Wan2.1

Wan-2.1 by the Wan Team.

This report presents Wan, a comprehensive and open suite of video foundation models designed to push the boundaries of video generation. Built upon the mainstream diffusion transformer paradigm, Wan achieves significant advancements in generative capabilities through a series of innovations, including our novel VAE, scalable pre-training strategies, large-scale data curation, and automated evaluation metrics. These contributions collectively enhance the model's performance and versatility. Specifically, Wan is characterized by four key features: Leading Performance: The 14B model of Wan, trained on a vast dataset comprising billions of images and videos, demonstrates the scaling laws of video generation with respect to both data and model size. It consistently outperforms the existing open-source models as well as state-of-the-art commercial solutions across multiple internal and external benchmarks, demonstrating a clear and significant performance superiority. Comprehensiveness: Wan offers two capable models, i.e., 1.3B and 14B parameters, for efficiency and effectiveness respectively. It also covers multiple downstream applications, including image-to-video, instruction-guided video editing, and personal video generation, encompassing up to eight tasks. Consumer-Grade Efficiency: The 1.3B model demonstrates exceptional resource efficiency, requiring only 8.19 GB VRAM, making it compatible with a wide range of consumer-grade GPUs. Openness: We open-source the entire series of Wan, including source code and all models, with the goal of fostering the growth of the video generation community. This openness seeks to significantly expand the creative possibilities of video production in the industry and provide academia with high-quality video foundation models. All the code and models are available at this https URL.

You can find all the original Wan2.1 checkpoints under the Wan-AI organization.

The following Wan models are supported in Diffusers:

• Wan 2.1 T2V 1.3B
• Wan 2.1 T2V 14B
• Wan 2.1 I2V 14B - 480P
• Wan 2.1 I2V 14B - 720P
• Wan 2.1 FLF2V 14B - 720P
• Wan 2.1 VACE 1.3B
• Wan 2.1 VACE 14B
• Wan 2.2 T2V 14B
• Wan 2.2 I2V 14B
• Wan 2.2 TI2V 5B

Tip

Click on the Wan2.1 models in the right sidebar for more examples of video generation.

Text-to-Video Generation

The example below demonstrates how to generate a video from text optimized for memory or inference speed.

Refer to the Reduce memory usage guide for more details about the various memory saving techniques.

The Wan2.1 text-to-video model below requires ~13GB of VRAM.

# pip install ftfy
import mindspore as ms
import numpy as np
from mindone.diffusers import AutoModel, WanPipeline
from mindone.diffusers.utils import export_to_video, load_image
from mindone.transformers import UMT5EncoderModel

text_encoder = UMT5EncoderModel.from_pretrained("Wan-AI/Wan2.1-T2V-14B-Diffusers", subfolder="text_encoder", mindspore_dtype=ms.float16)
vae = AutoModel.from_pretrained("Wan-AI/Wan2.1-T2V-14B-Diffusers", subfolder="vae", mindspore_dtype=ms.float32)
transformer = AutoModel.from_pretrained("Wan-AI/Wan2.1-T2V-14B-Diffusers", subfolder="transformer", mindspore_dtype=ms.float16)

pipeline = WanPipeline.from_pretrained(
    "Wan-AI/Wan2.1-T2V-14B-Diffusers",
    vae=vae,
    transformer=transformer,
    text_encoder=text_encoder,
    mindspore_dtype=ms.float16
)

prompt = """
The camera rushes from far to near in a low-angle shot,
revealing a white ferret on a log. It plays, leaps into the water, and emerges, as the camera zooms in
for a close-up. Water splashes berry bushes nearby, while moss, snow, and leaves blanket the ground.
Birch trees and a light blue sky frame the scene, with ferns in the foreground. Side lighting casts dynamic
shadows and warm highlights. Medium composition, front view, low angle, with depth of field.
"""
negative_prompt = """
Bright tones, overexposed, static, blurred details, subtitles, style, works, paintings, images, static, overall gray, worst quality,
low quality, JPEG compression residue, ugly, incomplete, extra fingers, poorly drawn hands, poorly drawn faces, deformed, disfigured,
misshapen limbs, fused fingers, still picture, messy background, three legs, many people in the background, walking backwards
"""

output = pipeline(
    prompt=prompt,
    negative_prompt=negative_prompt,
    num_frames=81,
    guidance_scale=5.0,
)[0][0]
export_to_video(output, "output.mp4", fps=16)

# pip install ftfy
import mindspore as ms
import numpy as np
from mindone.diffusers import AutoModel, WanPipeline
from mindone.diffusers.hooks.group_offloading import apply_group_offloading
from mindone.diffusers.utils import export_to_video, load_image
from mindone.transformers import UMT5EncoderModel

text_encoder = UMT5EncoderModel.from_pretrained("Wan-AI/Wan2.1-T2V-14B-Diffusers", subfolder="text_encoder", mindspore_dtype=ms.float16)
vae = AutoModel.from_pretrained("Wan-AI/Wan2.1-T2V-14B-Diffusers", subfolder="vae", mindspore_dtype=ms.float32)
transformer = AutoModel.from_pretrained("Wan-AI/Wan2.1-T2V-14B-Diffusers", subfolder="transformer", mindspore_dtype=ms.float16)

pipeline = WanPipeline.from_pretrained(
    "Wan-AI/Wan2.1-T2V-14B-Diffusers",
    vae=vae,
    transformer=transformer,
    text_encoder=text_encoder,
    mindspore_dtype=ms.float16
)

pipeline.transformer.construct = ms.jit(pipeline.transformer.construct)

prompt = """
The camera rushes from far to near in a low-angle shot,
revealing a white ferret on a log. It plays, leaps into the water, and emerges, as the camera zooms in
for a close-up. Water splashes berry bushes nearby, while moss, snow, and leaves blanket the ground.
Birch trees and a light blue sky frame the scene, with ferns in the foreground. Side lighting casts dynamic
shadows and warm highlights. Medium composition, front view, low angle, with depth of field.
"""
negative_prompt = """
Bright tones, overexposed, static, blurred details, subtitles, style, works, paintings, images, static, overall gray, worst quality,
low quality, JPEG compression residue, ugly, incomplete, extra fingers, poorly drawn hands, poorly drawn faces, deformed, disfigured,
misshapen limbs, fused fingers, still picture, messy background, three legs, many people in the background, walking backwards
"""

output = pipeline(
    prompt=prompt,
    negative_prompt=negative_prompt,
    num_frames=81,
    guidance_scale=5.0,
)[0][0]
export_to_video(output, "output.mp4", fps=16)

Any-to-Video Controllable Generation

Wan VACE supports various generation techniques which achieve controllable video generation. Some of the capabilities include: - Control to Video (Depth, Pose, Sketch, Flow, Grayscale, Scribble, Layout, Boundary Box, etc.). Recommended library for preprocessing videos to obtain control videos: huggingface/controlnet_aux - Image/Video to Video (first frame, last frame, starting clip, ending clip, random clips) - Inpainting and Outpainting - Subject to Video (faces, object, characters, etc.) - Composition to Video (reference anything, animate anything, swap anything, expand anything, move anything, etc.)

The code snippets available in this pull request demonstrate some examples of how videos can be generated with controllability signals.

The general rule of thumb to keep in mind when preparing inputs for the VACE pipeline is that the input images, or frames of a video that you want to use for conditioning, should have a corresponding mask that is black in color. The black mask signifies that the model will not generate new content for that area, and only use those parts for conditioning the generation process. For parts/frames that should be generated by the model, the mask should be white in color.

Notes

• Wan2.1 supports LoRAs with load_lora_weights.

Show example code

# pip install ftfy
import mindspore as ms
from mindone.diffusers import AutoModel, WanPipeline
from mindone.diffusers.schedulers.scheduling_unipc_multistep import UniPCMultistepScheduler
from mindone.diffusers.utils import export_to_video

vae = AutoModel.from_pretrained(
    "Wan-AI/Wan2.1-T2V-1.3B-Diffusers", subfolder="vae", mindspore_dtype=ms.float32
)
pipeline = WanPipeline.from_pretrained(
    "Wan-AI/Wan2.1-T2V-1.3B-Diffusers", vae=vae, mindspore_dtype=ms.float16
)
pipeline.scheduler = UniPCMultistepScheduler.from_config(
    pipeline.scheduler.config, flow_shift=5.0
)

pipeline.load_lora_weights("benjamin-paine/steamboat-willie-1.3b", adapter_name="steamboat-willie")
pipeline.set_adapters("steamboat-willie")

# use "steamboat willie style" to trigger the LoRA
prompt = """
steamboat willie style, golden era animation, The camera rushes from far to near in a low-angle shot,
revealing a white ferret on a log. It plays, leaps into the water, and emerges, as the camera zooms in
for a close-up. Water splashes berry bushes nearby, while moss, snow, and leaves blanket the ground.
Birch trees and a light blue sky frame the scene, with ferns in the foreground. Side lighting casts dynamic
shadows and warm highlights. Medium composition, front view, low angle, with depth of field.
"""

output = pipeline(
    prompt=prompt,
    num_frames=81,
    guidance_scale=5.0,
)[0][0]
export_to_video(output, "output.mp4", fps=16)

• [WanTransformer3DModel] and [AutoencoderKLWan] supports loading from single files with from_single_file.

Show example code

# pip install ftfy
import mindspore as ms
from mindone.diffusers import WanPipeline, WanTransformer3DModel, AutoencoderKLWan

vae = AutoencoderKLWan.from_single_file(
    "https://huggingface.co/Comfy-Org/Wan_2.1_ComfyUI_repackaged/blob/main/split_files/vae/wan_2.1_vae.safetensors"
)
transformer = WanTransformer3DModel.from_single_file(
    "https://huggingface.co/Comfy-Org/Wan_2.1_ComfyUI_repackaged/blob/main/split_files/diffusion_models/wan2.1_t2v_1.3B_bf16.safetensors",
    mindspore_dtype=ms.float16
)
pipeline = WanPipeline.from_pretrained(
    "Wan-AI/Wan2.1-T2V-1.3B-Diffusers",
    vae=vae,
    transformer=transformer,
    mindspore_dtype=ms.float16
)

• Set the [AutoencoderKLWan] dtype to mindspore.float32 for better decoding quality.

• The number of frames per second (fps) or k should be calculated by 4 * k + 1.

• Try lower shift values (2.0 to 5.0) for lower resolution videos and higher shift values (7.0 to 12.0) for higher resolution images.

• Wan 2.1 and 2.2 support using LightX2V LoRAs to speed up inference. Using them on Wan 2.2 is slightly more involed. Refer to this code snippet to learn more.

• Wan 2.2 has two denoisers. By default, LoRAs are only loaded into the first denoiser. One can set load_into_transformer_2=True to load LoRAs into the second denoiser. Refer to this and this examples to learn more.

mindone.diffusers.WanPipeline

Bases: DiffusionPipeline, WanLoraLoaderMixin

Pipeline for text-to-video generation using Wan.

This model inherits from [DiffusionPipeline]. Check the superclass documentation for the generic methods implemented for all pipelines (downloading, saving, running on a particular device, etc.).

PARAMETER	DESCRIPTION
tokenizer	Tokenizer from T5, specifically the google/umt5-xxl variant. TYPE: [`T5Tokenizer`]
text_encoder	T5, specifically the google/umt5-xxl variant. TYPE: [`T5EncoderModel`]
transformer	Conditional Transformer to denoise the input latents. TYPE: [`WanTransformer3DModel`] DEFAULT: None
scheduler	A scheduler to be used in combination with transformer to denoise the encoded image latents. TYPE: [`UniPCMultistepScheduler`]
vae	Variational Auto-Encoder (VAE) Model to encode and decode videos to and from latent representations. TYPE: [`AutoencoderKLWan`]
transformer_2	Conditional Transformer to denoise the input latents during the low-noise stage. If provided, enables two-stage denoising where transformer handles high-noise stages and transformer_2 handles low-noise stages. If not provided, only transformer is used. TYPE: [`WanTransformer3DModel`], *optional* DEFAULT: None
boundary_ratio	Ratio of total timesteps to use as the boundary for switching between transformers in two-stage denoising. The actual boundary timestep is calculated as boundary_ratio * num_train_timesteps. When provided, transformer handles timesteps >= boundary_timestep and transformer_2 handles timesteps < boundary_timestep. If None, only transformer is used for the entire denoising process. TYPE: `float`, *optional*, defaults to `None` DEFAULT: None

Source code in mindone/diffusers/pipelines/wan/pipeline_wan.py

class WanPipeline(DiffusionPipeline, WanLoraLoaderMixin):
    r"""
    Pipeline for text-to-video generation using Wan.

    This model inherits from [`DiffusionPipeline`]. Check the superclass documentation for the generic methods
    implemented for all pipelines (downloading, saving, running on a particular device, etc.).

    Args:
        tokenizer ([`T5Tokenizer`]):
            Tokenizer from [T5](https://huggingface.co/docs/transformers/en/model_doc/t5#transformers.T5Tokenizer),
            specifically the [google/umt5-xxl](https://huggingface.co/google/umt5-xxl) variant.
        text_encoder ([`T5EncoderModel`]):
            [T5](https://huggingface.co/docs/transformers/en/model_doc/t5#transformers.T5EncoderModel), specifically
            the [google/umt5-xxl](https://huggingface.co/google/umt5-xxl) variant.
        transformer ([`WanTransformer3DModel`]):
            Conditional Transformer to denoise the input latents.
        scheduler ([`UniPCMultistepScheduler`]):
            A scheduler to be used in combination with `transformer` to denoise the encoded image latents.
        vae ([`AutoencoderKLWan`]):
            Variational Auto-Encoder (VAE) Model to encode and decode videos to and from latent representations.
        transformer_2 ([`WanTransformer3DModel`], *optional*):
            Conditional Transformer to denoise the input latents during the low-noise stage. If provided, enables
            two-stage denoising where `transformer` handles high-noise stages and `transformer_2` handles low-noise
            stages. If not provided, only `transformer` is used.
        boundary_ratio (`float`, *optional*, defaults to `None`):
            Ratio of total timesteps to use as the boundary for switching between transformers in two-stage denoising.
            The actual boundary timestep is calculated as `boundary_ratio * num_train_timesteps`. When provided,
            `transformer` handles timesteps >= boundary_timestep and `transformer_2` handles timesteps <
            boundary_timestep. If `None`, only `transformer` is used for the entire denoising process.
    """

    model_cpu_offload_seq = "text_encoder->transformer->transformer_2->vae"
    _callback_tensor_inputs = ["latents", "prompt_embeds", "negative_prompt_embeds"]
    _optional_components = ["transformer", "transformer_2"]

    def __init__(
        self,
        tokenizer: AutoTokenizer,
        text_encoder: UMT5EncoderModel,
        vae: AutoencoderKLWan,
        scheduler: FlowMatchEulerDiscreteScheduler,
        transformer: Optional[WanTransformer3DModel] = None,
        transformer_2: Optional[WanTransformer3DModel] = None,
        boundary_ratio: Optional[float] = None,
        expand_timesteps: bool = False,  # Wan2.2 ti2v
    ):
        super().__init__()

        self.register_modules(
            vae=vae,
            text_encoder=text_encoder,
            tokenizer=tokenizer,
            transformer=transformer,
            scheduler=scheduler,
            transformer_2=transformer_2,
        )
        self.register_to_config(boundary_ratio=boundary_ratio)
        self.register_to_config(expand_timesteps=expand_timesteps)
        self.vae_scale_factor_temporal = self.vae.config.scale_factor_temporal if getattr(self, "vae", None) else 4
        self.vae_scale_factor_spatial = self.vae.config.scale_factor_spatial if getattr(self, "vae", None) else 8
        self.video_processor = VideoProcessor(vae_scale_factor=self.vae_scale_factor_spatial)

    def _get_t5_prompt_embeds(
        self,
        prompt: Union[str, List[str]] = None,
        num_videos_per_prompt: int = 1,
        max_sequence_length: int = 226,
        dtype: Optional[ms.Type] = None,
    ):
        dtype = dtype or self.text_encoder.dtype

        prompt = [prompt] if isinstance(prompt, str) else prompt
        prompt = [prompt_clean(u) for u in prompt]
        batch_size = len(prompt)

        text_inputs = self.tokenizer(
            prompt,
            padding="max_length",
            max_length=max_sequence_length,
            truncation=True,
            add_special_tokens=True,
            return_attention_mask=True,
            return_tensors="np",
        )
        text_input_ids, mask = ms.tensor(text_inputs.input_ids), ms.tensor(text_inputs.attention_mask)
        seq_lens = mask.gt(0).sum(dim=1).long()

        prompt_embeds = self.text_encoder(text_input_ids, mask)[0]
        prompt_embeds = prompt_embeds.to(dtype=dtype)
        prompt_embeds = [u[:v] for u, v in zip(prompt_embeds, seq_lens)]
        prompt_embeds = mint.stack(
            [mint.cat([u, u.new_zeros((max_sequence_length - u.shape[0], u.shape[1]))]) for u in prompt_embeds], dim=0
        )

        # duplicate text embeddings for each generation per prompt, using mps friendly method
        _, seq_len, _ = prompt_embeds.shape
        prompt_embeds = prompt_embeds.tile((1, num_videos_per_prompt, 1))
        prompt_embeds = prompt_embeds.view(batch_size * num_videos_per_prompt, seq_len, -1)

        return prompt_embeds

    def encode_prompt(
        self,
        prompt: Union[str, List[str]],
        negative_prompt: Optional[Union[str, List[str]]] = None,
        do_classifier_free_guidance: bool = True,
        num_videos_per_prompt: int = 1,
        prompt_embeds: Optional[ms.Tensor] = None,
        negative_prompt_embeds: Optional[ms.Tensor] = None,
        max_sequence_length: int = 226,
        dtype: Optional[ms.Type] = None,
    ):
        r"""
        Encodes the prompt into text encoder hidden states.

        Args:
            prompt (`str` or `List[str]`, *optional*):
                prompt to be encoded
            negative_prompt (`str` or `List[str]`, *optional*):
                The prompt or prompts not to guide the image generation. If not defined, one has to pass
                `negative_prompt_embeds` instead. Ignored when not using guidance (i.e., ignored if `guidance_scale` is
                less than `1`).
            do_classifier_free_guidance (`bool`, *optional*, defaults to `True`):
                Whether to use classifier free guidance or not.
            num_videos_per_prompt (`int`, *optional*, defaults to 1):
                Number of videos that should be generated per prompt.
            prompt_embeds (`ms.Tensor`, *optional*):
                Pre-generated text embeddings. Can be used to easily tweak text inputs, *e.g.* prompt weighting. If not
                provided, text embeddings will be generated from `prompt` input argument.
            negative_prompt_embeds (`ms.Tensor`, *optional*):
                Pre-generated negative text embeddings. Can be used to easily tweak text inputs, *e.g.* prompt
                weighting. If not provided, negative_prompt_embeds will be generated from `negative_prompt` input
                argument.
            dtype: (`ms.Type`, *optional*):
                mindspore dtype
        """
        prompt = [prompt] if isinstance(prompt, str) else prompt
        if prompt is not None:
            batch_size = len(prompt)
        else:
            batch_size = prompt_embeds.shape[0]

        if prompt_embeds is None:
            prompt_embeds = self._get_t5_prompt_embeds(
                prompt=prompt,
                num_videos_per_prompt=num_videos_per_prompt,
                max_sequence_length=max_sequence_length,
                dtype=dtype,
            )

        if do_classifier_free_guidance and negative_prompt_embeds is None:
            negative_prompt = negative_prompt or ""
            negative_prompt = batch_size * [negative_prompt] if isinstance(negative_prompt, str) else negative_prompt

            if prompt is not None and type(prompt) is not type(negative_prompt):
                raise TypeError(
                    f"`negative_prompt` should be the same type to `prompt`, but got {type(negative_prompt)} !="
                    f" {type(prompt)}."
                )
            elif batch_size != len(negative_prompt):
                raise ValueError(
                    f"`negative_prompt`: {negative_prompt} has batch size {len(negative_prompt)}, but `prompt`:"
                    f" {prompt} has batch size {batch_size}. Please make sure that passed `negative_prompt` matches"
                    " the batch size of `prompt`."
                )

            negative_prompt_embeds = self._get_t5_prompt_embeds(
                prompt=negative_prompt,
                num_videos_per_prompt=num_videos_per_prompt,
                max_sequence_length=max_sequence_length,
                dtype=dtype,
            )

        return prompt_embeds, negative_prompt_embeds

    def check_inputs(
        self,
        prompt,
        negative_prompt,
        height,
        width,
        prompt_embeds=None,
        negative_prompt_embeds=None,
        callback_on_step_end_tensor_inputs=None,
        guidance_scale_2=None,
    ):
        if height % 16 != 0 or width % 16 != 0:
            raise ValueError(f"`height` and `width` have to be divisible by 16 but are {height} and {width}.")

        if callback_on_step_end_tensor_inputs is not None and not all(
            k in self._callback_tensor_inputs for k in callback_on_step_end_tensor_inputs
        ):
            raise ValueError(
                f"`callback_on_step_end_tensor_inputs` has to be in {self._callback_tensor_inputs}, but found {[k for k in callback_on_step_end_tensor_inputs if k not in self._callback_tensor_inputs]}"  # noqa: E501
            )

        if prompt is not None and prompt_embeds is not None:
            raise ValueError(
                f"Cannot forward both `prompt`: {prompt} and `prompt_embeds`: {prompt_embeds}. Please make sure to"
                " only forward one of the two."
            )
        elif negative_prompt is not None and negative_prompt_embeds is not None:
            raise ValueError(
                f"Cannot forward both `negative_prompt`: {negative_prompt} and `negative_prompt_embeds`: {negative_prompt_embeds}. Please make sure to"
                " only forward one of the two."
            )
        elif prompt is None and prompt_embeds is None:
            raise ValueError(
                "Provide either `prompt` or `prompt_embeds`. Cannot leave both `prompt` and `prompt_embeds` undefined."
            )
        elif prompt is not None and (not isinstance(prompt, str) and not isinstance(prompt, list)):
            raise ValueError(f"`prompt` has to be of type `str` or `list` but is {type(prompt)}")
        elif negative_prompt is not None and (
            not isinstance(negative_prompt, str) and not isinstance(negative_prompt, list)
        ):
            raise ValueError(f"`negative_prompt` has to be of type `str` or `list` but is {type(negative_prompt)}")

        if self.config.boundary_ratio is None and guidance_scale_2 is not None:
            raise ValueError("`guidance_scale_2` is only supported when the pipeline's `boundary_ratio` is not None.")

    def prepare_latents(
        self,
        batch_size: int,
        num_channels_latents: int = 16,
        height: int = 480,
        width: int = 832,
        num_frames: int = 81,
        dtype: Optional[ms.Type] = None,
        generator: Optional[Union[np.random.Generator, List[np.random.Generator]]] = None,
        latents: Optional[ms.Tensor] = None,
    ) -> ms.Tensor:
        if latents is not None:
            return latents.to(dtype=dtype)

        num_latent_frames = (num_frames - 1) // self.vae_scale_factor_temporal + 1
        shape = (
            batch_size,
            num_channels_latents,
            num_latent_frames,
            int(height) // self.vae_scale_factor_spatial,
            int(width) // self.vae_scale_factor_spatial,
        )
        if isinstance(generator, list) and len(generator) != batch_size:
            raise ValueError(
                f"You have passed a list of generators of length {len(generator)}, but requested an effective batch"
                f" size of {batch_size}. Make sure the batch size matches the length of the generators."
            )

        latents = randn_tensor(shape, generator=generator, dtype=dtype)
        return latents

    @property
    def guidance_scale(self):
        return self._guidance_scale

    @property
    def do_classifier_free_guidance(self):
        return self._guidance_scale > 1.0

    @property
    def num_timesteps(self):
        return self._num_timesteps

    @property
    def current_timestep(self):
        return self._current_timestep

    @property
    def interrupt(self):
        return self._interrupt

    @property
    def attention_kwargs(self):
        return self._attention_kwargs

    def __call__(
        self,
        prompt: Union[str, List[str]] = None,
        negative_prompt: Union[str, List[str]] = None,
        height: int = 480,
        width: int = 832,
        num_frames: int = 81,
        num_inference_steps: int = 50,
        guidance_scale: float = 5.0,
        guidance_scale_2: Optional[float] = None,
        num_videos_per_prompt: Optional[int] = 1,
        generator: Optional[Union[np.random.Generator, List[np.random.Generator]]] = None,
        latents: Optional[ms.Tensor] = None,
        prompt_embeds: Optional[ms.Tensor] = None,
        negative_prompt_embeds: Optional[ms.Tensor] = None,
        output_type: Optional[str] = "np",
        return_dict: bool = False,
        attention_kwargs: Optional[Dict[str, Any]] = None,
        callback_on_step_end: Optional[
            Union[Callable[[int, int, Dict], None], PipelineCallback, MultiPipelineCallbacks]
        ] = None,
        callback_on_step_end_tensor_inputs: List[str] = ["latents"],
        max_sequence_length: int = 512,
    ):
        r"""
        The call function to the pipeline for generation.

        Args:
            prompt (`str` or `List[str]`, *optional*):
                The prompt or prompts to guide the image generation. If not defined, pass `prompt_embeds` instead.
            negative_prompt (`str` or `List[str]`, *optional*):
                The prompt or prompts to avoid during image generation. If not defined, pass `negative_prompt_embeds`
                instead. Ignored when not using guidance (`guidance_scale` < `1`).
            height (`int`, defaults to `480`):
                The height in pixels of the generated image.
            width (`int`, defaults to `832`):
                The width in pixels of the generated image.
            num_frames (`int`, defaults to `81`):
                The number of frames in the generated video.
            num_inference_steps (`int`, defaults to `50`):
                The number of denoising steps. More denoising steps usually lead to a higher quality image at the
                expense of slower inference.
            guidance_scale (`float`, defaults to `5.0`):
                Guidance scale as defined in [Classifier-Free Diffusion
                Guidance](https://huggingface.co/papers/2207.12598). `guidance_scale` is defined as `w` of equation 2.
                of [Imagen Paper](https://huggingface.co/papers/2205.11487). Guidance scale is enabled by setting
                `guidance_scale > 1`. Higher guidance scale encourages to generate images that are closely linked to
                the text `prompt`, usually at the expense of lower image quality.
            guidance_scale_2 (`float`, *optional*, defaults to `None`):
                Guidance scale for the low-noise stage transformer (`transformer_2`). If `None` and the pipeline's
                `boundary_ratio` is not None, uses the same value as `guidance_scale`. Only used when `transformer_2`
                and the pipeline's `boundary_ratio` are not None.
            num_videos_per_prompt (`int`, *optional*, defaults to 1):
                The number of images to generate per prompt.
            generator (`np.random.Generator` or `List[np.random.Generator]`, *optional*):
                A [`np.random.Generator`](https://numpy.org/doc/stable/reference/random/generator.html) to make
                generation deterministic.
            latents (`ms.Tensor`, *optional*):
                Pre-generated noisy latents sampled from a Gaussian distribution, to be used as inputs for image
                generation. Can be used to tweak the same generation with different prompts. If not provided, a latents
                tensor is generated by sampling using the supplied random `generator`.
            prompt_embeds (`ms.Tensor`, *optional*):
                Pre-generated text embeddings. Can be used to easily tweak text inputs (prompt weighting). If not
                provided, text embeddings are generated from the `prompt` input argument.
            output_type (`str`, *optional*, defaults to `"np"`):
                The output format of the generated image. Choose between `PIL.Image` or `np.array`.
            return_dict (`bool`, *optional*, defaults to `False`):
                Whether or not to return a [`WanPipelineOutput`] instead of a plain tuple.
            attention_kwargs (`dict`, *optional*):
                A kwargs dictionary that if specified is passed along to the `AttentionProcessor` as defined under
                `self.processor` in
                [diffusers.models.attention_processor](https://github.com/huggingface/diffusers/blob/main/src/diffusers/models/attention_processor.py).
            callback_on_step_end (`Callable`, `PipelineCallback`, `MultiPipelineCallbacks`, *optional*):
                A function or a subclass of `PipelineCallback` or `MultiPipelineCallbacks` that is called at the end of
                each denoising step during the inference. with the following arguments: `callback_on_step_end(self:
                DiffusionPipeline, step: int, timestep: int, callback_kwargs: Dict)`. `callback_kwargs` will include a
                list of all tensors as specified by `callback_on_step_end_tensor_inputs`.
            callback_on_step_end_tensor_inputs (`List`, *optional*):
                The list of tensor inputs for the `callback_on_step_end` function. The tensors specified in the list
                will be passed as `callback_kwargs` argument. You will only be able to include variables listed in the
                `._callback_tensor_inputs` attribute of your pipeline class.
            max_sequence_length (`int`, defaults to `512`):
                The maximum sequence length of the text encoder. If the prompt is longer than this, it will be
                truncated. If the prompt is shorter, it will be padded to this length.

        Examples:

        Returns:
            [`~WanPipelineOutput`] or `tuple`:
                If `return_dict` is `True`, [`WanPipelineOutput`] is returned, otherwise a `tuple` is returned where
                the first element is a list with the generated images and the second element is a list of `bool`s
                indicating whether the corresponding generated image contains "not-safe-for-work" (nsfw) content.
        """

        if isinstance(callback_on_step_end, (PipelineCallback, MultiPipelineCallbacks)):
            callback_on_step_end_tensor_inputs = callback_on_step_end.tensor_inputs

        # 1. Check inputs. Raise error if not correct
        self.check_inputs(
            prompt,
            negative_prompt,
            height,
            width,
            prompt_embeds,
            negative_prompt_embeds,
            callback_on_step_end_tensor_inputs,
            guidance_scale_2,
        )

        if num_frames % self.vae_scale_factor_temporal != 1:
            logger.warning(
                f"`num_frames - 1` has to be divisible by {self.vae_scale_factor_temporal}. Rounding to the nearest number."
            )
            num_frames = num_frames // self.vae_scale_factor_temporal * self.vae_scale_factor_temporal + 1
        num_frames = max(num_frames, 1)

        if self.config.boundary_ratio is not None and guidance_scale_2 is None:
            guidance_scale_2 = guidance_scale

        self._guidance_scale = guidance_scale
        self._guidance_scale_2 = guidance_scale_2
        self._attention_kwargs = attention_kwargs
        self._current_timestep = None
        self._interrupt = False

        # 2. Define call parameters
        if prompt is not None and isinstance(prompt, str):
            batch_size = 1
        elif prompt is not None and isinstance(prompt, list):
            batch_size = len(prompt)
        else:
            batch_size = prompt_embeds.shape[0]

        # 3. Encode input prompt
        prompt_embeds, negative_prompt_embeds = self.encode_prompt(
            prompt=prompt,
            negative_prompt=negative_prompt,
            do_classifier_free_guidance=self.do_classifier_free_guidance,
            num_videos_per_prompt=num_videos_per_prompt,
            prompt_embeds=prompt_embeds,
            negative_prompt_embeds=negative_prompt_embeds,
            max_sequence_length=max_sequence_length,
        )

        transformer_dtype = self.transformer.dtype if self.transformer is not None else self.transformer_2.dtype
        prompt_embeds = prompt_embeds.to(transformer_dtype)
        if negative_prompt_embeds is not None:
            negative_prompt_embeds = negative_prompt_embeds.to(transformer_dtype)

        # 4. Prepare timesteps
        self.scheduler.set_timesteps(num_inference_steps)
        timesteps = self.scheduler.timesteps

        # 5. Prepare latent variables
        num_channels_latents = (
            self.transformer.config.in_channels
            if self.transformer is not None
            else self.transformer_2.config.in_channels
        )
        latents = self.prepare_latents(
            batch_size * num_videos_per_prompt,
            num_channels_latents,
            height,
            width,
            num_frames,
            ms.float32,
            generator,
            latents,
        )

        mask = mint.ones(latents.shape, dtype=ms.float32)

        # 6. Denoising loop
        num_warmup_steps = len(timesteps) - num_inference_steps * self.scheduler.order
        self._num_timesteps = len(timesteps)

        if self.config.boundary_ratio is not None:
            boundary_timestep = self.config.boundary_ratio * self.scheduler.config.num_train_timesteps
        else:
            boundary_timestep = None

        # we're popping the `scale` instead of getting it because otherwise `scale` will be propagated
        # to the transformer and will raise RuntimeError.
        lora_scale = attention_kwargs.pop("scale", None) if attention_kwargs is not None else None
        if lora_scale is not None:
            # weight the lora layers by setting `lora_scale` for each PEFT layer
            scale_lora_layers(self.transformer, lora_scale)

        with self.progress_bar(total=num_inference_steps) as progress_bar:
            for i, t in enumerate(timesteps):
                if self.interrupt:
                    continue

                self._current_timestep = t

                if boundary_timestep is None or t >= boundary_timestep:
                    # wan2.1 or high-noise stage in wan2.2
                    current_model = self.transformer
                    current_guidance_scale = guidance_scale
                else:
                    # low-noise stage in wan2.2
                    current_model = self.transformer_2
                    current_guidance_scale = guidance_scale_2

                latent_model_input = latents.to(transformer_dtype)
                if self.config.expand_timesteps:
                    # seq_len: num_latent_frames * latent_height//2 * latent_width//2
                    temp_ts = (mask[0][0][:, ::2, ::2] * t).flatten()
                    # batch_size, seq_len
                    timestep = temp_ts.unsqueeze(0).broadcast_to((latents.shape[0], -1))
                else:
                    timestep = t.broadcast_to((latents.shape[0],))

                with current_model.cache_context("cond"):
                    noise_pred = current_model(
                        hidden_states=latent_model_input,
                        timestep=timestep,
                        encoder_hidden_states=prompt_embeds,
                        attention_kwargs=attention_kwargs,
                        return_dict=False,
                    )[0]

                if self.do_classifier_free_guidance:
                    with current_model.cache_context("uncond"):
                        noise_uncond = current_model(
                            hidden_states=latent_model_input,
                            timestep=timestep,
                            encoder_hidden_states=negative_prompt_embeds,
                            attention_kwargs=attention_kwargs,
                            return_dict=False,
                        )[0]
                    noise_pred = noise_uncond + current_guidance_scale * (noise_pred - noise_uncond)

                # compute the previous noisy sample x_t -> x_t-1
                latents = self.scheduler.step(noise_pred, t, latents, return_dict=False)[0]

                if callback_on_step_end is not None:
                    callback_kwargs = {}
                    for k in callback_on_step_end_tensor_inputs:
                        callback_kwargs[k] = locals()[k]
                    callback_outputs = callback_on_step_end(self, i, t, callback_kwargs)

                    latents = callback_outputs.pop("latents", latents)
                    prompt_embeds = callback_outputs.pop("prompt_embeds", prompt_embeds)
                    negative_prompt_embeds = callback_outputs.pop("negative_prompt_embeds", negative_prompt_embeds)

                # call the callback, if provided
                if i == len(timesteps) - 1 or ((i + 1) > num_warmup_steps and (i + 1) % self.scheduler.order == 0):
                    progress_bar.update()

        if lora_scale is not None:
            # remove `lora_scale` from each PEFT layer
            unscale_lora_layers(self.transformer, lora_scale)
            attention_kwargs["scale"] = lora_scale

        self._current_timestep = None

        if not output_type == "latent":
            latents = latents.to(self.vae.dtype)
            latents_mean = (
                ms.tensor(self.vae.config.latents_mean).view(1, self.vae.config.z_dim, 1, 1, 1).to(latents.dtype)
            )
            latents_std = 1.0 / ms.tensor(self.vae.config.latents_std).view(1, self.vae.config.z_dim, 1, 1, 1).to(
                latents.dtype
            )
            latents = latents / latents_std + latents_mean
            # TODO: we use pynative mode here since cache in vae.decode which not supported in graph mode
            with pynative_context():
                video = self.vae.decode(latents, return_dict=False)[0]
            video = self.video_processor.postprocess_video(video, output_type=output_type)
        else:
            video = latents

        if not return_dict:
            return (video,)

        return WanPipelineOutput(frames=video)

mindone.diffusers.WanPipeline.__call__(prompt=None, negative_prompt=None, height=480, width=832, num_frames=81, num_inference_steps=50, guidance_scale=5.0, guidance_scale_2=None, num_videos_per_prompt=1, generator=None, latents=None, prompt_embeds=None, negative_prompt_embeds=None, output_type='np', return_dict=False, attention_kwargs=None, callback_on_step_end=None, callback_on_step_end_tensor_inputs=['latents'], max_sequence_length=512)

The call function to the pipeline for generation.

PARAMETER	DESCRIPTION
prompt	The prompt or prompts to guide the image generation. If not defined, pass prompt_embeds instead. TYPE: `str` or `List[str]`, *optional* DEFAULT: None
negative_prompt	The prompt or prompts to avoid during image generation. If not defined, pass negative_prompt_embeds instead. Ignored when not using guidance (guidance_scale < 1). TYPE: `str` or `List[str]`, *optional* DEFAULT: None
height	The height in pixels of the generated image. TYPE: `int`, defaults to `480` DEFAULT: 480
width	The width in pixels of the generated image. TYPE: `int`, defaults to `832` DEFAULT: 832
num_frames	The number of frames in the generated video. TYPE: `int`, defaults to `81` DEFAULT: 81
num_inference_steps	The number of denoising steps. More denoising steps usually lead to a higher quality image at the expense of slower inference. TYPE: `int`, defaults to `50` DEFAULT: 50
guidance_scale	Guidance scale as defined in Classifier-Free Diffusion Guidance. guidance_scale is defined as w of equation 2. of Imagen Paper. Guidance scale is enabled by setting guidance_scale > 1. Higher guidance scale encourages to generate images that are closely linked to the text prompt, usually at the expense of lower image quality. TYPE: `float`, defaults to `5.0` DEFAULT: 5.0
guidance_scale_2	Guidance scale for the low-noise stage transformer (transformer_2). If None and the pipeline's boundary_ratio is not None, uses the same value as guidance_scale. Only used when transformer_2 and the pipeline's boundary_ratio are not None. TYPE: `float`, *optional*, defaults to `None` DEFAULT: None
num_videos_per_prompt	The number of images to generate per prompt. TYPE: `int`, *optional*, defaults to 1 DEFAULT: 1
generator	A np.random.Generator to make generation deterministic. TYPE: `np.random.Generator` or `List[np.random.Generator]`, *optional* DEFAULT: None
latents	Pre-generated noisy latents sampled from a Gaussian distribution, to be used as inputs for image generation. Can be used to tweak the same generation with different prompts. If not provided, a latents tensor is generated by sampling using the supplied random generator. TYPE: `ms.Tensor`, *optional* DEFAULT: None
prompt_embeds	Pre-generated text embeddings. Can be used to easily tweak text inputs (prompt weighting). If not provided, text embeddings are generated from the prompt input argument. TYPE: `ms.Tensor`, *optional* DEFAULT: None
output_type	The output format of the generated image. Choose between PIL.Image or np.array. TYPE: `str`, *optional*, defaults to `"np"` DEFAULT: 'np'
return_dict	Whether or not to return a [WanPipelineOutput] instead of a plain tuple. TYPE: `bool`, *optional*, defaults to `False` DEFAULT: False
attention_kwargs	A kwargs dictionary that if specified is passed along to the AttentionProcessor as defined under self.processor in diffusers.models.attention_processor. TYPE: `dict`, *optional* DEFAULT: None
callback_on_step_end	A function or a subclass of PipelineCallback or MultiPipelineCallbacks that is called at the end of each denoising step during the inference. with the following arguments: callback_on_step_end(self: DiffusionPipeline, step: int, timestep: int, callback_kwargs: Dict). callback_kwargs will include a list of all tensors as specified by callback_on_step_end_tensor_inputs. TYPE: `Callable`, `PipelineCallback`, `MultiPipelineCallbacks`, *optional* DEFAULT: None
callback_on_step_end_tensor_inputs	The list of tensor inputs for the callback_on_step_end function. The tensors specified in the list will be passed as callback_kwargs argument. You will only be able to include variables listed in the ._callback_tensor_inputs attribute of your pipeline class. TYPE: `List`, *optional* DEFAULT: ['latents']
max_sequence_length	The maximum sequence length of the text encoder. If the prompt is longer than this, it will be truncated. If the prompt is shorter, it will be padded to this length. TYPE: `int`, defaults to `512` DEFAULT: 512

RETURNS	DESCRIPTION
	[~WanPipelineOutput] or tuple: If return_dict is True, [WanPipelineOutput] is returned, otherwise a tuple is returned where the first element is a list with the generated images and the second element is a list of bools indicating whether the corresponding generated image contains "not-safe-for-work" (nsfw) content.

Source code in mindone/diffusers/pipelines/wan/pipeline_wan.py

def __call__(
    self,
    prompt: Union[str, List[str]] = None,
    negative_prompt: Union[str, List[str]] = None,
    height: int = 480,
    width: int = 832,
    num_frames: int = 81,
    num_inference_steps: int = 50,
    guidance_scale: float = 5.0,
    guidance_scale_2: Optional[float] = None,
    num_videos_per_prompt: Optional[int] = 1,
    generator: Optional[Union[np.random.Generator, List[np.random.Generator]]] = None,
    latents: Optional[ms.Tensor] = None,
    prompt_embeds: Optional[ms.Tensor] = None,
    negative_prompt_embeds: Optional[ms.Tensor] = None,
    output_type: Optional[str] = "np",
    return_dict: bool = False,
    attention_kwargs: Optional[Dict[str, Any]] = None,
    callback_on_step_end: Optional[
        Union[Callable[[int, int, Dict], None], PipelineCallback, MultiPipelineCallbacks]
    ] = None,
    callback_on_step_end_tensor_inputs: List[str] = ["latents"],
    max_sequence_length: int = 512,
):
    r"""
    The call function to the pipeline for generation.

    Args:
        prompt (`str` or `List[str]`, *optional*):
            The prompt or prompts to guide the image generation. If not defined, pass `prompt_embeds` instead.
        negative_prompt (`str` or `List[str]`, *optional*):
            The prompt or prompts to avoid during image generation. If not defined, pass `negative_prompt_embeds`
            instead. Ignored when not using guidance (`guidance_scale` < `1`).
        height (`int`, defaults to `480`):
            The height in pixels of the generated image.
        width (`int`, defaults to `832`):
            The width in pixels of the generated image.
        num_frames (`int`, defaults to `81`):
            The number of frames in the generated video.
        num_inference_steps (`int`, defaults to `50`):
            The number of denoising steps. More denoising steps usually lead to a higher quality image at the
            expense of slower inference.
        guidance_scale (`float`, defaults to `5.0`):
            Guidance scale as defined in [Classifier-Free Diffusion
            Guidance](https://huggingface.co/papers/2207.12598). `guidance_scale` is defined as `w` of equation 2.
            of [Imagen Paper](https://huggingface.co/papers/2205.11487). Guidance scale is enabled by setting
            `guidance_scale > 1`. Higher guidance scale encourages to generate images that are closely linked to
            the text `prompt`, usually at the expense of lower image quality.
        guidance_scale_2 (`float`, *optional*, defaults to `None`):
            Guidance scale for the low-noise stage transformer (`transformer_2`). If `None` and the pipeline's
            `boundary_ratio` is not None, uses the same value as `guidance_scale`. Only used when `transformer_2`
            and the pipeline's `boundary_ratio` are not None.
        num_videos_per_prompt (`int`, *optional*, defaults to 1):
            The number of images to generate per prompt.
        generator (`np.random.Generator` or `List[np.random.Generator]`, *optional*):
            A [`np.random.Generator`](https://numpy.org/doc/stable/reference/random/generator.html) to make
            generation deterministic.
        latents (`ms.Tensor`, *optional*):
            Pre-generated noisy latents sampled from a Gaussian distribution, to be used as inputs for image
            generation. Can be used to tweak the same generation with different prompts. If not provided, a latents
            tensor is generated by sampling using the supplied random `generator`.
        prompt_embeds (`ms.Tensor`, *optional*):
            Pre-generated text embeddings. Can be used to easily tweak text inputs (prompt weighting). If not
            provided, text embeddings are generated from the `prompt` input argument.
        output_type (`str`, *optional*, defaults to `"np"`):
            The output format of the generated image. Choose between `PIL.Image` or `np.array`.
        return_dict (`bool`, *optional*, defaults to `False`):
            Whether or not to return a [`WanPipelineOutput`] instead of a plain tuple.
        attention_kwargs (`dict`, *optional*):
            A kwargs dictionary that if specified is passed along to the `AttentionProcessor` as defined under
            `self.processor` in
            [diffusers.models.attention_processor](https://github.com/huggingface/diffusers/blob/main/src/diffusers/models/attention_processor.py).
        callback_on_step_end (`Callable`, `PipelineCallback`, `MultiPipelineCallbacks`, *optional*):
            A function or a subclass of `PipelineCallback` or `MultiPipelineCallbacks` that is called at the end of
            each denoising step during the inference. with the following arguments: `callback_on_step_end(self:
            DiffusionPipeline, step: int, timestep: int, callback_kwargs: Dict)`. `callback_kwargs` will include a
            list of all tensors as specified by `callback_on_step_end_tensor_inputs`.
        callback_on_step_end_tensor_inputs (`List`, *optional*):
            The list of tensor inputs for the `callback_on_step_end` function. The tensors specified in the list
            will be passed as `callback_kwargs` argument. You will only be able to include variables listed in the
            `._callback_tensor_inputs` attribute of your pipeline class.
        max_sequence_length (`int`, defaults to `512`):
            The maximum sequence length of the text encoder. If the prompt is longer than this, it will be
            truncated. If the prompt is shorter, it will be padded to this length.

    Examples:

    Returns:
        [`~WanPipelineOutput`] or `tuple`:
            If `return_dict` is `True`, [`WanPipelineOutput`] is returned, otherwise a `tuple` is returned where
            the first element is a list with the generated images and the second element is a list of `bool`s
            indicating whether the corresponding generated image contains "not-safe-for-work" (nsfw) content.
    """

    if isinstance(callback_on_step_end, (PipelineCallback, MultiPipelineCallbacks)):
        callback_on_step_end_tensor_inputs = callback_on_step_end.tensor_inputs

    # 1. Check inputs. Raise error if not correct
    self.check_inputs(
        prompt,
        negative_prompt,
        height,
        width,
        prompt_embeds,
        negative_prompt_embeds,
        callback_on_step_end_tensor_inputs,
        guidance_scale_2,
    )

    if num_frames % self.vae_scale_factor_temporal != 1:
        logger.warning(
            f"`num_frames - 1` has to be divisible by {self.vae_scale_factor_temporal}. Rounding to the nearest number."
        )
        num_frames = num_frames // self.vae_scale_factor_temporal * self.vae_scale_factor_temporal + 1
    num_frames = max(num_frames, 1)

    if self.config.boundary_ratio is not None and guidance_scale_2 is None:
        guidance_scale_2 = guidance_scale

    self._guidance_scale = guidance_scale
    self._guidance_scale_2 = guidance_scale_2
    self._attention_kwargs = attention_kwargs
    self._current_timestep = None
    self._interrupt = False

    # 2. Define call parameters
    if prompt is not None and isinstance(prompt, str):
        batch_size = 1
    elif prompt is not None and isinstance(prompt, list):
        batch_size = len(prompt)
    else:
        batch_size = prompt_embeds.shape[0]

    # 3. Encode input prompt
    prompt_embeds, negative_prompt_embeds = self.encode_prompt(
        prompt=prompt,
        negative_prompt=negative_prompt,
        do_classifier_free_guidance=self.do_classifier_free_guidance,
        num_videos_per_prompt=num_videos_per_prompt,
        prompt_embeds=prompt_embeds,
        negative_prompt_embeds=negative_prompt_embeds,
        max_sequence_length=max_sequence_length,
    )

    transformer_dtype = self.transformer.dtype if self.transformer is not None else self.transformer_2.dtype
    prompt_embeds = prompt_embeds.to(transformer_dtype)
    if negative_prompt_embeds is not None:
        negative_prompt_embeds = negative_prompt_embeds.to(transformer_dtype)

    # 4. Prepare timesteps
    self.scheduler.set_timesteps(num_inference_steps)
    timesteps = self.scheduler.timesteps

    # 5. Prepare latent variables
    num_channels_latents = (
        self.transformer.config.in_channels
        if self.transformer is not None
        else self.transformer_2.config.in_channels
    )
    latents = self.prepare_latents(
        batch_size * num_videos_per_prompt,
        num_channels_latents,
        height,
        width,
        num_frames,
        ms.float32,
        generator,
        latents,
    )

    mask = mint.ones(latents.shape, dtype=ms.float32)

    # 6. Denoising loop
    num_warmup_steps = len(timesteps) - num_inference_steps * self.scheduler.order
    self._num_timesteps = len(timesteps)

    if self.config.boundary_ratio is not None:
        boundary_timestep = self.config.boundary_ratio * self.scheduler.config.num_train_timesteps
    else:
        boundary_timestep = None

    # we're popping the `scale` instead of getting it because otherwise `scale` will be propagated
    # to the transformer and will raise RuntimeError.
    lora_scale = attention_kwargs.pop("scale", None) if attention_kwargs is not None else None
    if lora_scale is not None:
        # weight the lora layers by setting `lora_scale` for each PEFT layer
        scale_lora_layers(self.transformer, lora_scale)

    with self.progress_bar(total=num_inference_steps) as progress_bar:
        for i, t in enumerate(timesteps):
            if self.interrupt:
                continue

            self._current_timestep = t

            if boundary_timestep is None or t >= boundary_timestep:
                # wan2.1 or high-noise stage in wan2.2
                current_model = self.transformer
                current_guidance_scale = guidance_scale
            else:
                # low-noise stage in wan2.2
                current_model = self.transformer_2
                current_guidance_scale = guidance_scale_2

            latent_model_input = latents.to(transformer_dtype)
            if self.config.expand_timesteps:
                # seq_len: num_latent_frames * latent_height//2 * latent_width//2
                temp_ts = (mask[0][0][:, ::2, ::2] * t).flatten()
                # batch_size, seq_len
                timestep = temp_ts.unsqueeze(0).broadcast_to((latents.shape[0], -1))
            else:
                timestep = t.broadcast_to((latents.shape[0],))

            with current_model.cache_context("cond"):
                noise_pred = current_model(
                    hidden_states=latent_model_input,
                    timestep=timestep,
                    encoder_hidden_states=prompt_embeds,
                    attention_kwargs=attention_kwargs,
                    return_dict=False,
                )[0]

            if self.do_classifier_free_guidance:
                with current_model.cache_context("uncond"):
                    noise_uncond = current_model(
                        hidden_states=latent_model_input,
                        timestep=timestep,
                        encoder_hidden_states=negative_prompt_embeds,
                        attention_kwargs=attention_kwargs,
                        return_dict=False,
                    )[0]
                noise_pred = noise_uncond + current_guidance_scale * (noise_pred - noise_uncond)

            # compute the previous noisy sample x_t -> x_t-1
            latents = self.scheduler.step(noise_pred, t, latents, return_dict=False)[0]

            if callback_on_step_end is not None:
                callback_kwargs = {}
                for k in callback_on_step_end_tensor_inputs:
                    callback_kwargs[k] = locals()[k]
                callback_outputs = callback_on_step_end(self, i, t, callback_kwargs)

                latents = callback_outputs.pop("latents", latents)
                prompt_embeds = callback_outputs.pop("prompt_embeds", prompt_embeds)
                negative_prompt_embeds = callback_outputs.pop("negative_prompt_embeds", negative_prompt_embeds)

            # call the callback, if provided
            if i == len(timesteps) - 1 or ((i + 1) > num_warmup_steps and (i + 1) % self.scheduler.order == 0):
                progress_bar.update()

    if lora_scale is not None:
        # remove `lora_scale` from each PEFT layer
        unscale_lora_layers(self.transformer, lora_scale)
        attention_kwargs["scale"] = lora_scale

    self._current_timestep = None

    if not output_type == "latent":
        latents = latents.to(self.vae.dtype)
        latents_mean = (
            ms.tensor(self.vae.config.latents_mean).view(1, self.vae.config.z_dim, 1, 1, 1).to(latents.dtype)
        )
        latents_std = 1.0 / ms.tensor(self.vae.config.latents_std).view(1, self.vae.config.z_dim, 1, 1, 1).to(
            latents.dtype
        )
        latents = latents / latents_std + latents_mean
        # TODO: we use pynative mode here since cache in vae.decode which not supported in graph mode
        with pynative_context():
            video = self.vae.decode(latents, return_dict=False)[0]
        video = self.video_processor.postprocess_video(video, output_type=output_type)
    else:
        video = latents

    if not return_dict:
        return (video,)

    return WanPipelineOutput(frames=video)

mindone.diffusers.WanPipeline.encode_prompt(prompt, negative_prompt=None, do_classifier_free_guidance=True, num_videos_per_prompt=1, prompt_embeds=None, negative_prompt_embeds=None, max_sequence_length=226, dtype=None)

Encodes the prompt into text encoder hidden states.

PARAMETER	DESCRIPTION
prompt	prompt to be encoded TYPE: `str` or `List[str]`, *optional*
negative_prompt	The prompt or prompts not to guide the image generation. If not defined, one has to pass negative_prompt_embeds instead. Ignored when not using guidance (i.e., ignored if guidance_scale is less than 1). TYPE: `str` or `List[str]`, *optional* DEFAULT: None
do_classifier_free_guidance	Whether to use classifier free guidance or not. TYPE: `bool`, *optional*, defaults to `True` DEFAULT: True
num_videos_per_prompt	Number of videos that should be generated per prompt. TYPE: `int`, *optional*, defaults to 1 DEFAULT: 1
prompt_embeds	Pre-generated text embeddings. Can be used to easily tweak text inputs, e.g. prompt weighting. If not provided, text embeddings will be generated from prompt input argument. TYPE: `ms.Tensor`, *optional* DEFAULT: None
negative_prompt_embeds	Pre-generated negative text embeddings. Can be used to easily tweak text inputs, e.g. prompt weighting. If not provided, negative_prompt_embeds will be generated from negative_prompt input argument. TYPE: `ms.Tensor`, *optional* DEFAULT: None
dtype	(ms.Type, optional): mindspore dtype TYPE: Optional[Type] DEFAULT: None

Source code in mindone/diffusers/pipelines/wan/pipeline_wan.py

def encode_prompt(
    self,
    prompt: Union[str, List[str]],
    negative_prompt: Optional[Union[str, List[str]]] = None,
    do_classifier_free_guidance: bool = True,
    num_videos_per_prompt: int = 1,
    prompt_embeds: Optional[ms.Tensor] = None,
    negative_prompt_embeds: Optional[ms.Tensor] = None,
    max_sequence_length: int = 226,
    dtype: Optional[ms.Type] = None,
):
    r"""
    Encodes the prompt into text encoder hidden states.

    Args:
        prompt (`str` or `List[str]`, *optional*):
            prompt to be encoded
        negative_prompt (`str` or `List[str]`, *optional*):
            The prompt or prompts not to guide the image generation. If not defined, one has to pass
            `negative_prompt_embeds` instead. Ignored when not using guidance (i.e., ignored if `guidance_scale` is
            less than `1`).
        do_classifier_free_guidance (`bool`, *optional*, defaults to `True`):
            Whether to use classifier free guidance or not.
        num_videos_per_prompt (`int`, *optional*, defaults to 1):
            Number of videos that should be generated per prompt.
        prompt_embeds (`ms.Tensor`, *optional*):
            Pre-generated text embeddings. Can be used to easily tweak text inputs, *e.g.* prompt weighting. If not
            provided, text embeddings will be generated from `prompt` input argument.
        negative_prompt_embeds (`ms.Tensor`, *optional*):
            Pre-generated negative text embeddings. Can be used to easily tweak text inputs, *e.g.* prompt
            weighting. If not provided, negative_prompt_embeds will be generated from `negative_prompt` input
            argument.
        dtype: (`ms.Type`, *optional*):
            mindspore dtype
    """
    prompt = [prompt] if isinstance(prompt, str) else prompt
    if prompt is not None:
        batch_size = len(prompt)
    else:
        batch_size = prompt_embeds.shape[0]

    if prompt_embeds is None:
        prompt_embeds = self._get_t5_prompt_embeds(
            prompt=prompt,
            num_videos_per_prompt=num_videos_per_prompt,
            max_sequence_length=max_sequence_length,
            dtype=dtype,
        )

    if do_classifier_free_guidance and negative_prompt_embeds is None:
        negative_prompt = negative_prompt or ""
        negative_prompt = batch_size * [negative_prompt] if isinstance(negative_prompt, str) else negative_prompt

        if prompt is not None and type(prompt) is not type(negative_prompt):
            raise TypeError(
                f"`negative_prompt` should be the same type to `prompt`, but got {type(negative_prompt)} !="
                f" {type(prompt)}."
            )
        elif batch_size != len(negative_prompt):
            raise ValueError(
                f"`negative_prompt`: {negative_prompt} has batch size {len(negative_prompt)}, but `prompt`:"
                f" {prompt} has batch size {batch_size}. Please make sure that passed `negative_prompt` matches"
                " the batch size of `prompt`."
            )

        negative_prompt_embeds = self._get_t5_prompt_embeds(
            prompt=negative_prompt,
            num_videos_per_prompt=num_videos_per_prompt,
            max_sequence_length=max_sequence_length,
            dtype=dtype,
        )

    return prompt_embeds, negative_prompt_embeds

mindone.diffusers.WanImageToVideoPipeline

Bases: DiffusionPipeline, WanLoraLoaderMixin

Pipeline for image-to-video generation using Wan.

This model inherits from [DiffusionPipeline]. Check the superclass documentation for the generic methods implemented for all pipelines (downloading, saving, running on a particular device, etc.).

PARAMETER	DESCRIPTION
tokenizer	Tokenizer from T5, specifically the google/umt5-xxl variant. TYPE: [`T5Tokenizer`]
text_encoder	T5, specifically the google/umt5-xxl variant. TYPE: [`T5EncoderModel`]
image_encoder	CLIP, specifically the clip-vit-huge-patch14 variant. TYPE: [`CLIPVisionModel`] DEFAULT: None
transformer	Conditional Transformer to denoise the input latents. TYPE: [`WanTransformer3DModel`] DEFAULT: None
scheduler	A scheduler to be used in combination with transformer to denoise the encoded image latents. TYPE: [`UniPCMultistepScheduler`]
vae	Variational Auto-Encoder (VAE) Model to encode and decode videos to and from latent representations. TYPE: [`AutoencoderKLWan`]
transformer_2	Conditional Transformer to denoise the input latents during the low-noise stage. In two-stage denoising, transformer handles high-noise stages and transformer_2 handles low-noise stages. If not provided, only transformer is used. TYPE: [`WanTransformer3DModel`], *optional* DEFAULT: None
boundary_ratio	Ratio of total timesteps to use as the boundary for switching between transformers in two-stage denoising. The actual boundary timestep is calculated as boundary_ratio * num_train_timesteps. When provided, transformer handles timesteps >= boundary_timestep and transformer_2 handles timesteps < boundary_timestep. If None, only transformer is used for the entire denoising process. TYPE: `float`, *optional*, defaults to `None` DEFAULT: None

Source code in mindone/diffusers/pipelines/wan/pipeline_wan_i2v.py

class WanImageToVideoPipeline(DiffusionPipeline, WanLoraLoaderMixin):
    r"""
    Pipeline for image-to-video generation using Wan.

    This model inherits from [`DiffusionPipeline`]. Check the superclass documentation for the generic methods
    implemented for all pipelines (downloading, saving, running on a particular device, etc.).

    Args:
        tokenizer ([`T5Tokenizer`]):
            Tokenizer from [T5](https://huggingface.co/docs/transformers/en/model_doc/t5#transformers.T5Tokenizer),
            specifically the [google/umt5-xxl](https://huggingface.co/google/umt5-xxl) variant.
        text_encoder ([`T5EncoderModel`]):
            [T5](https://huggingface.co/docs/transformers/en/model_doc/t5#transformers.T5EncoderModel), specifically
            the [google/umt5-xxl](https://huggingface.co/google/umt5-xxl) variant.
        image_encoder ([`CLIPVisionModel`]):
            [CLIP](https://huggingface.co/docs/transformers/model_doc/clip#transformers.CLIPVisionModel), specifically
            the
            [clip-vit-huge-patch14](https://github.com/mlfoundations/open_clip/blob/main/docs/PRETRAINED.md#vit-h14-xlm-roberta-large)
            variant.
        transformer ([`WanTransformer3DModel`]):
            Conditional Transformer to denoise the input latents.
        scheduler ([`UniPCMultistepScheduler`]):
            A scheduler to be used in combination with `transformer` to denoise the encoded image latents.
        vae ([`AutoencoderKLWan`]):
            Variational Auto-Encoder (VAE) Model to encode and decode videos to and from latent representations.
        transformer_2 ([`WanTransformer3DModel`], *optional*):
            Conditional Transformer to denoise the input latents during the low-noise stage. In two-stage denoising,
            `transformer` handles high-noise stages and `transformer_2` handles low-noise stages. If not provided, only
            `transformer` is used.
        boundary_ratio (`float`, *optional*, defaults to `None`):
            Ratio of total timesteps to use as the boundary for switching between transformers in two-stage denoising.
            The actual boundary timestep is calculated as `boundary_ratio * num_train_timesteps`. When provided,
            `transformer` handles timesteps >= boundary_timestep and `transformer_2` handles timesteps <
            boundary_timestep. If `None`, only `transformer` is used for the entire denoising process.
    """

    model_cpu_offload_seq = "text_encoder->image_encoder->transformer->transformer_2->vae"
    _callback_tensor_inputs = ["latents", "prompt_embeds", "negative_prompt_embeds"]
    _optional_components = ["transformer", "transformer_2", "image_encoder", "image_processor"]

    def __init__(
        self,
        tokenizer: AutoTokenizer,
        text_encoder: UMT5EncoderModel,
        vae: AutoencoderKLWan,
        scheduler: FlowMatchEulerDiscreteScheduler,
        image_processor: CLIPImageProcessor = None,
        image_encoder: CLIPVisionModel = None,
        transformer: WanTransformer3DModel = None,
        transformer_2: WanTransformer3DModel = None,
        boundary_ratio: Optional[float] = None,
        expand_timesteps: bool = False,
    ):
        super().__init__()

        self.register_modules(
            vae=vae,
            text_encoder=text_encoder,
            tokenizer=tokenizer,
            image_encoder=image_encoder,
            transformer=transformer,
            scheduler=scheduler,
            image_processor=image_processor,
            transformer_2=transformer_2,
        )
        self.register_to_config(boundary_ratio=boundary_ratio, expand_timesteps=expand_timesteps)

        self.vae_scale_factor_temporal = self.vae.config.scale_factor_temporal if getattr(self, "vae", None) else 4
        self.vae_scale_factor_spatial = self.vae.config.scale_factor_spatial if getattr(self, "vae", None) else 8
        self.video_processor = VideoProcessor(vae_scale_factor=self.vae_scale_factor_spatial)
        self.image_processor = image_processor

    def _get_t5_prompt_embeds(
        self,
        prompt: Union[str, List[str]] = None,
        num_videos_per_prompt: int = 1,
        max_sequence_length: int = 512,
        dtype: Optional[ms.Type] = None,
    ):
        dtype = dtype or self.text_encoder.dtype

        prompt = [prompt] if isinstance(prompt, str) else prompt
        prompt = [prompt_clean(u) for u in prompt]
        batch_size = len(prompt)

        text_inputs = self.tokenizer(
            prompt,
            padding="max_length",
            max_length=max_sequence_length,
            truncation=True,
            add_special_tokens=True,
            return_attention_mask=True,
            return_tensors="np",
        )
        text_input_ids, mask = ms.tensor(text_inputs.input_ids), ms.tensor(text_inputs.attention_mask)
        seq_lens = mask.gt(0).sum(dim=1).long()

        prompt_embeds = self.text_encoder(text_input_ids, mask)[0]
        prompt_embeds = prompt_embeds.to(dtype=dtype)
        prompt_embeds = [u[:v] for u, v in zip(prompt_embeds, seq_lens)]
        prompt_embeds = mint.stack(
            [mint.cat([u, u.new_zeros((max_sequence_length - u.shape[0], u.shape[1]))]) for u in prompt_embeds], dim=0
        )

        # duplicate text embeddings for each generation per prompt, using mps friendly method
        _, seq_len, _ = prompt_embeds.shape
        prompt_embeds = prompt_embeds.tile((1, num_videos_per_prompt, 1))
        prompt_embeds = prompt_embeds.view(batch_size * num_videos_per_prompt, seq_len, -1)

        return prompt_embeds

    def encode_image(
        self,
        image: PipelineImageInput,
    ):
        image = self.image_processor(images=image, return_tensors="np")
        for key in image.keys():
            image[key] = ms.tensor(image[key])
        image_embeds = self.image_encoder(**image, output_hidden_states=True)
        return image_embeds[2][-2]

    # Copied from diffusers.pipelines.wan.pipeline_wan.WanPipeline.encode_prompt
    def encode_prompt(
        self,
        prompt: Union[str, List[str]],
        negative_prompt: Optional[Union[str, List[str]]] = None,
        do_classifier_free_guidance: bool = True,
        num_videos_per_prompt: int = 1,
        prompt_embeds: Optional[ms.Tensor] = None,
        negative_prompt_embeds: Optional[ms.Tensor] = None,
        max_sequence_length: int = 226,
        dtype: Optional[ms.Type] = None,
    ):
        r"""
        Encodes the prompt into text encoder hidden states.

        Args:
            prompt (`str` or `List[str]`, *optional*):
                prompt to be encoded
            negative_prompt (`str` or `List[str]`, *optional*):
                The prompt or prompts not to guide the image generation. If not defined, one has to pass
                `negative_prompt_embeds` instead. Ignored when not using guidance (i.e., ignored if `guidance_scale` is
                less than `1`).
            do_classifier_free_guidance (`bool`, *optional*, defaults to `True`):
                Whether to use classifier free guidance or not.
            num_videos_per_prompt (`int`, *optional*, defaults to 1):
                Number of videos that should be generated per prompt.
            prompt_embeds (`ms.Tensor`, *optional*):
                Pre-generated text embeddings. Can be used to easily tweak text inputs, *e.g.* prompt weighting. If not
                provided, text embeddings will be generated from `prompt` input argument.
            negative_prompt_embeds (`ms.Tensor`, *optional*):
                Pre-generated negative text embeddings. Can be used to easily tweak text inputs, *e.g.* prompt
                weighting. If not provided, negative_prompt_embeds will be generated from `negative_prompt` input
                argument.
            dtype: (`ms.Type`, *optional*):
                mindspore dtype
        """
        prompt = [prompt] if isinstance(prompt, str) else prompt
        if prompt is not None:
            batch_size = len(prompt)
        else:
            batch_size = prompt_embeds.shape[0]

        if prompt_embeds is None:
            prompt_embeds = self._get_t5_prompt_embeds(
                prompt=prompt,
                num_videos_per_prompt=num_videos_per_prompt,
                max_sequence_length=max_sequence_length,
                dtype=dtype,
            )

        if do_classifier_free_guidance and negative_prompt_embeds is None:
            negative_prompt = negative_prompt or ""
            negative_prompt = batch_size * [negative_prompt] if isinstance(negative_prompt, str) else negative_prompt

            if prompt is not None and type(prompt) is not type(negative_prompt):
                raise TypeError(
                    f"`negative_prompt` should be the same type to `prompt`, but got {type(negative_prompt)} !="
                    f" {type(prompt)}."
                )
            elif batch_size != len(negative_prompt):
                raise ValueError(
                    f"`negative_prompt`: {negative_prompt} has batch size {len(negative_prompt)}, but `prompt`:"
                    f" {prompt} has batch size {batch_size}. Please make sure that passed `negative_prompt` matches"
                    " the batch size of `prompt`."
                )

            negative_prompt_embeds = self._get_t5_prompt_embeds(
                prompt=negative_prompt,
                num_videos_per_prompt=num_videos_per_prompt,
                max_sequence_length=max_sequence_length,
                dtype=dtype,
            )

        return prompt_embeds, negative_prompt_embeds

    def check_inputs(
        self,
        prompt,
        negative_prompt,
        image,
        height,
        width,
        prompt_embeds=None,
        negative_prompt_embeds=None,
        image_embeds=None,
        callback_on_step_end_tensor_inputs=None,
        guidance_scale_2=None,
    ):
        if image is not None and image_embeds is not None:
            raise ValueError(
                f"Cannot forward both `image`: {image} and `image_embeds`: {image_embeds}. Please make sure to"
                " only forward one of the two."
            )
        if image is None and image_embeds is None:
            raise ValueError(
                "Provide either `image` or `prompt_embeds`. Cannot leave both `image` and `image_embeds` undefined."
            )
        if image is not None and not isinstance(image, ms.Tensor) and not isinstance(image, PIL.Image.Image):
            raise ValueError(f"`image` has to be of type `ms.Tensor` or `PIL.Image.Image` but is {type(image)}")
        if height % 16 != 0 or width % 16 != 0:
            raise ValueError(f"`height` and `width` have to be divisible by 16 but are {height} and {width}.")

        if callback_on_step_end_tensor_inputs is not None and not all(
            k in self._callback_tensor_inputs for k in callback_on_step_end_tensor_inputs
        ):
            raise ValueError(
                f"`callback_on_step_end_tensor_inputs` has to be in {self._callback_tensor_inputs}, but found {[k for k in callback_on_step_end_tensor_inputs if k not in self._callback_tensor_inputs]}"  # noqa: E501
            )

        if prompt is not None and prompt_embeds is not None:
            raise ValueError(
                f"Cannot forward both `prompt`: {prompt} and `prompt_embeds`: {prompt_embeds}. Please make sure to"
                " only forward one of the two."
            )
        elif negative_prompt is not None and negative_prompt_embeds is not None:
            raise ValueError(
                f"Cannot forward both `negative_prompt`: {negative_prompt} and `negative_prompt_embeds`: {negative_prompt_embeds}. Please make sure to"  # noqa: E501
                " only forward one of the two."
            )
        elif prompt is None and prompt_embeds is None:
            raise ValueError(
                "Provide either `prompt` or `prompt_embeds`. Cannot leave both `prompt` and `prompt_embeds` undefined."
            )
        elif prompt is not None and (not isinstance(prompt, str) and not isinstance(prompt, list)):
            raise ValueError(f"`prompt` has to be of type `str` or `list` but is {type(prompt)}")
        elif negative_prompt is not None and (
            not isinstance(negative_prompt, str) and not isinstance(negative_prompt, list)
        ):
            raise ValueError(f"`negative_prompt` has to be of type `str` or `list` but is {type(negative_prompt)}")

        if self.config.boundary_ratio is None and guidance_scale_2 is not None:
            raise ValueError("`guidance_scale_2` is only supported when the pipeline's `boundary_ratio` is not None.")

        if self.config.boundary_ratio is not None and image_embeds is not None:
            raise ValueError("Cannot forward `image_embeds` when the pipeline's `boundary_ratio` is not configured.")

    def prepare_latents(
        self,
        image: PipelineImageInput,
        batch_size: int,
        num_channels_latents: int = 16,
        height: int = 480,
        width: int = 832,
        num_frames: int = 81,
        dtype: Optional[ms.Type] = None,
        generator: Optional[Union[np.random.Generator, List[np.random.Generator]]] = None,
        latents: Optional[ms.Tensor] = None,
        last_image: Optional[ms.Tensor] = None,
    ) -> Tuple[ms.Tensor, ms.Tensor]:
        num_latent_frames = (num_frames - 1) // self.vae_scale_factor_temporal + 1
        latent_height = height // self.vae_scale_factor_spatial
        latent_width = width // self.vae_scale_factor_spatial

        shape = (batch_size, num_channels_latents, num_latent_frames, latent_height, latent_width)
        if isinstance(generator, list) and len(generator) != batch_size:
            raise ValueError(
                f"You have passed a list of generators of length {len(generator)}, but requested an effective batch"
                f" size of {batch_size}. Make sure the batch size matches the length of the generators."
            )

        if latents is None:
            latents = randn_tensor(shape, generator=generator, dtype=dtype)
        else:
            latents = latents.to(dtype=dtype)

        image = image.unsqueeze(2)  # [batch_size, channels, 1, height, width]

        if self.config.expand_timesteps:
            video_condition = image

        elif last_image is None:
            video_condition = mint.cat(
                [image, image.new_zeros((image.shape[0], image.shape[1], num_frames - 1, height, width))], dim=2
            )
        else:
            last_image = last_image.unsqueeze(2)
            video_condition = mint.cat(
                [image, image.new_zeros((image.shape[0], image.shape[1], num_frames - 2, height, width)), last_image],
                dim=2,
            )
        video_condition = video_condition.to(dtype=self.vae.dtype)

        latents_mean = ms.tensor(self.vae.config.latents_mean).view(1, self.vae.config.z_dim, 1, 1, 1).to(latents.dtype)
        latents_std = 1.0 / ms.tensor(self.vae.config.latents_std).view(1, self.vae.config.z_dim, 1, 1, 1).to(
            latents.dtype
        )

        # TODO: we use pynative mode here since there is cache in vae.encode which not supported in graph mode
        with pynative_context():
            if isinstance(generator, list):
                latent_condition = [
                    retrieve_latents(self.vae, self.vae.encode(video_condition)[0], sample_mode="argmax")
                    for _ in generator
                ]
                latent_condition = mint.cat(latent_condition)
            else:
                latent_condition = retrieve_latents(self.vae, self.vae.encode(video_condition)[0], sample_mode="argmax")
                latent_condition = latent_condition.tile((batch_size, 1, 1, 1, 1))

        latent_condition = latent_condition.to(dtype)
        latent_condition = (latent_condition - latents_mean) * latents_std

        if self.config.expand_timesteps:
            first_frame_mask = mint.ones((1, 1, num_latent_frames, latent_height, latent_width), dtype=dtype)
            first_frame_mask[:, :, 0] = 0
            return latents, latent_condition, first_frame_mask

        mask_lat_size = mint.ones((batch_size, 1, num_frames, latent_height, latent_width))

        if last_image is None:
            mask_lat_size[:, :, list(range(1, num_frames))] = 0
        else:
            mask_lat_size[:, :, list(range(1, num_frames - 1))] = 0
        first_frame_mask = mask_lat_size[:, :, 0:1]
        first_frame_mask = mint.repeat_interleave(first_frame_mask, dim=2, repeats=self.vae_scale_factor_temporal)
        mask_lat_size = mint.concat([first_frame_mask, mask_lat_size[:, :, 1:, :]], dim=2)
        mask_lat_size = mask_lat_size.view(batch_size, -1, self.vae_scale_factor_temporal, latent_height, latent_width)
        mask_lat_size = mask_lat_size.swapaxes(1, 2)

        return latents, mint.concat([mask_lat_size, latent_condition], dim=1)

    @property
    def guidance_scale(self):
        return self._guidance_scale

    @property
    def do_classifier_free_guidance(self):
        return self._guidance_scale > 1

    @property
    def num_timesteps(self):
        return self._num_timesteps

    @property
    def current_timestep(self):
        return self._current_timestep

    @property
    def interrupt(self):
        return self._interrupt

    @property
    def attention_kwargs(self):
        return self._attention_kwargs

    def __call__(
        self,
        image: PipelineImageInput,
        prompt: Union[str, List[str]] = None,
        negative_prompt: Union[str, List[str]] = None,
        height: int = 480,
        width: int = 832,
        num_frames: int = 81,
        num_inference_steps: int = 50,
        guidance_scale: float = 5.0,
        guidance_scale_2: Optional[float] = None,
        num_videos_per_prompt: Optional[int] = 1,
        generator: Optional[Union[np.random.Generator, List[np.random.Generator]]] = None,
        latents: Optional[ms.Tensor] = None,
        prompt_embeds: Optional[ms.Tensor] = None,
        negative_prompt_embeds: Optional[ms.Tensor] = None,
        image_embeds: Optional[ms.Tensor] = None,
        last_image: Optional[ms.Tensor] = None,
        output_type: Optional[str] = "np",
        return_dict: bool = False,
        attention_kwargs: Optional[Dict[str, Any]] = None,
        callback_on_step_end: Optional[
            Union[Callable[[int, int, Dict], None], PipelineCallback, MultiPipelineCallbacks]
        ] = None,
        callback_on_step_end_tensor_inputs: List[str] = ["latents"],
        max_sequence_length: int = 512,
    ):
        r"""
        The call function to the pipeline for generation.

        Args:
            image (`PipelineImageInput`):
                The input image to condition the generation on. Must be an image, a list of images or a `ms.Tensor`.
            prompt (`str` or `List[str]`, *optional*):
                The prompt or prompts to guide the image generation. If not defined, one has to pass `prompt_embeds`.
                instead.
            negative_prompt (`str` or `List[str]`, *optional*):
                The prompt or prompts not to guide the image generation. If not defined, one has to pass
                `negative_prompt_embeds` instead. Ignored when not using guidance (i.e., ignored if `guidance_scale` is
                less than `1`).
            height (`int`, defaults to `480`):
                The height of the generated video.
            width (`int`, defaults to `832`):
                The width of the generated video.
            num_frames (`int`, defaults to `81`):
                The number of frames in the generated video.
            num_inference_steps (`int`, defaults to `50`):
                The number of denoising steps. More denoising steps usually lead to a higher quality image at the
                expense of slower inference.
            guidance_scale (`float`, defaults to `5.0`):
                Guidance scale as defined in [Classifier-Free Diffusion
                Guidance](https://huggingface.co/papers/2207.12598). `guidance_scale` is defined as `w` of equation 2.
                of [Imagen Paper](https://huggingface.co/papers/2205.11487). Guidance scale is enabled by setting
                `guidance_scale > 1`. Higher guidance scale encourages to generate images that are closely linked to
                the text `prompt`, usually at the expense of lower image quality.
            guidance_scale_2 (`float`, *optional*, defaults to `None`):
                Guidance scale for the low-noise stage transformer (`transformer_2`). If `None` and the pipeline's
                `boundary_ratio` is not None, uses the same value as `guidance_scale`. Only used when `transformer_2`
                and the pipeline's `boundary_ratio` are not None.
            num_videos_per_prompt (`int`, *optional*, defaults to 1):
                The number of images to generate per prompt.
            generator (`np.random.Generator` or `List[np.random.Generator]`, *optional*):
                A [`np.random.Generator`](https://numpy.org/doc/stable/reference/random/generator.html) to make
                generation deterministic.
            latents (`ms.Tensor`, *optional*):
                Pre-generated noisy latents sampled from a Gaussian distribution, to be used as inputs for image
                generation. Can be used to tweak the same generation with different prompts. If not provided, a latents
                tensor is generated by sampling using the supplied random `generator`.
            prompt_embeds (`ms.Tensor`, *optional*):
                Pre-generated text embeddings. Can be used to easily tweak text inputs (prompt weighting). If not
                provided, text embeddings are generated from the `prompt` input argument.
            negative_prompt_embeds (`ms.Tensor`, *optional*):
                Pre-generated text embeddings. Can be used to easily tweak text inputs (prompt weighting). If not
                provided, text embeddings are generated from the `negative_prompt` input argument.
            image_embeds (`ms.Tensor`, *optional*):
                Pre-generated image embeddings. Can be used to easily tweak image inputs (weighting). If not provided,
                image embeddings are generated from the `image` input argument.
            output_type (`str`, *optional*, defaults to `"np"`):
                The output format of the generated image. Choose between `PIL.Image` or `np.array`.
            return_dict (`bool`, *optional*, defaults to `False`):
                Whether or not to return a [`WanPipelineOutput`] instead of a plain tuple.
            attention_kwargs (`dict`, *optional*):
                A kwargs dictionary that if specified is passed along to the `AttentionProcessor` as defined under
                `self.processor` in
                [diffusers.models.attention_processor](https://github.com/huggingface/diffusers/blob/main/src/diffusers/models/attention_processor.py).
            callback_on_step_end (`Callable`, `PipelineCallback`, `MultiPipelineCallbacks`, *optional*):
                A function or a subclass of `PipelineCallback` or `MultiPipelineCallbacks` that is called at the end of
                each denoising step during the inference. with the following arguments: `callback_on_step_end(self:
                DiffusionPipeline, step: int, timestep: int, callback_kwargs: Dict)`. `callback_kwargs` will include a
                list of all tensors as specified by `callback_on_step_end_tensor_inputs`.
            callback_on_step_end_tensor_inputs (`List`, *optional*):
                The list of tensor inputs for the `callback_on_step_end` function. The tensors specified in the list
                will be passed as `callback_kwargs` argument. You will only be able to include variables listed in the
                `._callback_tensor_inputs` attribute of your pipeline class.
            max_sequence_length (`int`, defaults to `512`):
                The maximum sequence length of the text encoder. If the prompt is longer than this, it will be
                truncated. If the prompt is shorter, it will be padded to this length.

        Examples:

        Returns:
            [`~WanPipelineOutput`] or `tuple`:
                If `return_dict` is `True`, [`WanPipelineOutput`] is returned, otherwise a `tuple` is returned where
                the first element is a list with the generated images and the second element is a list of `bool`s
                indicating whether the corresponding generated image contains "not-safe-for-work" (nsfw) content.
        """

        if isinstance(callback_on_step_end, (PipelineCallback, MultiPipelineCallbacks)):
            callback_on_step_end_tensor_inputs = callback_on_step_end.tensor_inputs

        # 1. Check inputs. Raise error if not correct
        self.check_inputs(
            prompt,
            negative_prompt,
            image,
            height,
            width,
            prompt_embeds,
            negative_prompt_embeds,
            image_embeds,
            callback_on_step_end_tensor_inputs,
            guidance_scale_2,
        )

        if num_frames % self.vae_scale_factor_temporal != 1:
            logger.warning(
                f"`num_frames - 1` has to be divisible by {self.vae_scale_factor_temporal}. Rounding to the nearest number."
            )
            num_frames = num_frames // self.vae_scale_factor_temporal * self.vae_scale_factor_temporal + 1
        num_frames = max(num_frames, 1)

        if self.config.boundary_ratio is not None and guidance_scale_2 is None:
            guidance_scale_2 = guidance_scale

        self._guidance_scale = guidance_scale
        self._guidance_scale_2 = guidance_scale_2
        self._attention_kwargs = attention_kwargs
        self._current_timestep = None
        self._interrupt = False

        # 2. Define call parameters
        if prompt is not None and isinstance(prompt, str):
            batch_size = 1
        elif prompt is not None and isinstance(prompt, list):
            batch_size = len(prompt)
        else:
            batch_size = prompt_embeds.shape[0]

        # 3. Encode input prompt
        prompt_embeds, negative_prompt_embeds = self.encode_prompt(
            prompt=prompt,
            negative_prompt=negative_prompt,
            do_classifier_free_guidance=self.do_classifier_free_guidance,
            num_videos_per_prompt=num_videos_per_prompt,
            prompt_embeds=prompt_embeds,
            negative_prompt_embeds=negative_prompt_embeds,
            max_sequence_length=max_sequence_length,
        )

        # Encode image embedding
        transformer_dtype = self.transformer.dtype if self.transformer is not None else self.transformer_2.dtype
        prompt_embeds = prompt_embeds.to(transformer_dtype)
        if negative_prompt_embeds is not None:
            negative_prompt_embeds = negative_prompt_embeds.to(transformer_dtype)

        # only wan 2.1 i2v transformer accepts image_embeds
        if self.transformer is not None and self.transformer.config.image_dim is not None:
            if image_embeds is None:
                if last_image is None:
                    image_embeds = self.encode_image(image)
                else:
                    image_embeds = self.encode_image([image, last_image])
            image_embeds = image_embeds.tile((batch_size, 1, 1))
            image_embeds = image_embeds.to(transformer_dtype)

        # 4. Prepare timesteps
        self.scheduler.set_timesteps(num_inference_steps)
        timesteps = self.scheduler.timesteps

        # 5. Prepare latent variables
        num_channels_latents = self.vae.config.z_dim
        image = self.video_processor.preprocess(image, height=height, width=width).to(dtype=ms.float32)
        if last_image is not None:
            last_image = self.video_processor.preprocess(last_image, height=height, width=width).to(dtype=ms.float32)

        latents_outputs = self.prepare_latents(
            image,
            batch_size * num_videos_per_prompt,
            num_channels_latents,
            height,
            width,
            num_frames,
            ms.float32,
            generator,
            latents,
            last_image,
        )
        if self.config.expand_timesteps:
            # wan 2.2 5b i2v use firt_frame_mask to mask timesteps
            latents, condition, first_frame_mask = latents_outputs
        else:
            latents, condition = latents_outputs

        # 6. Denoising loop
        num_warmup_steps = len(timesteps) - num_inference_steps * self.scheduler.order
        self._num_timesteps = len(timesteps)

        if self.config.boundary_ratio is not None:
            boundary_timestep = self.config.boundary_ratio * self.scheduler.config.num_train_timesteps
        else:
            boundary_timestep = None

        # we're popping the `scale` instead of getting it because otherwise `scale` will be propagated
        # to the transformer and will raise RuntimeError.
        lora_scale = attention_kwargs.pop("scale", None) if attention_kwargs is not None else None
        if lora_scale is not None:
            # weight the lora layers by setting `lora_scale` for each PEFT layer
            scale_lora_layers(self.transformer, lora_scale)

        with self.progress_bar(total=num_inference_steps) as progress_bar:
            for i, t in enumerate(timesteps):
                if self.interrupt:
                    continue

                self._current_timestep = t

                if boundary_timestep is None or t >= boundary_timestep:
                    # wan2.1 or high-noise stage in wan2.2
                    current_model = self.transformer
                    current_guidance_scale = guidance_scale
                else:
                    # low-noise stage in wan2.2
                    current_model = self.transformer_2
                    current_guidance_scale = guidance_scale_2

                if self.config.expand_timesteps:
                    latent_model_input = (1 - first_frame_mask) * condition + first_frame_mask * latents
                    latent_model_input = latent_model_input.to(transformer_dtype)

                    # seq_len: num_latent_frames * (latent_height // patch_size) * (latent_width // patch_size)
                    temp_ts = (first_frame_mask[0][0][:, ::2, ::2] * t).flatten()
                    # batch_size, seq_len
                    timestep = temp_ts.unsqueeze(0).broadcast_to((latents.shape[0], -1))
                else:
                    latent_model_input = mint.cat([latents, condition], dim=1).to(transformer_dtype)
                    timestep = t.broadcast_to((latents.shape[0],))

                with current_model.cache_context("cond"):
                    noise_pred = current_model(
                        hidden_states=latent_model_input,
                        timestep=timestep,
                        encoder_hidden_states=prompt_embeds,
                        encoder_hidden_states_image=image_embeds,
                        attention_kwargs=attention_kwargs,
                        return_dict=False,
                    )[0]

                if self.do_classifier_free_guidance:
                    with current_model.cache_context("uncond"):
                        noise_uncond = current_model(
                            hidden_states=latent_model_input,
                            timestep=timestep,
                            encoder_hidden_states=negative_prompt_embeds,
                            encoder_hidden_states_image=image_embeds,
                            attention_kwargs=attention_kwargs,
                            return_dict=False,
                        )[0]
                        noise_pred = noise_uncond + current_guidance_scale * (noise_pred - noise_uncond)

                # compute the previous noisy sample x_t -> x_t-1
                latents = self.scheduler.step(noise_pred, t, latents, return_dict=False)[0]

                if callback_on_step_end is not None:
                    callback_kwargs = {}
                    for k in callback_on_step_end_tensor_inputs:
                        callback_kwargs[k] = locals()[k]
                    callback_outputs = callback_on_step_end(self, i, t, callback_kwargs)

                    latents = callback_outputs.pop("latents", latents)
                    prompt_embeds = callback_outputs.pop("prompt_embeds", prompt_embeds)
                    negative_prompt_embeds = callback_outputs.pop("negative_prompt_embeds", negative_prompt_embeds)

                # call the callback, if provided
                if i == len(timesteps) - 1 or ((i + 1) > num_warmup_steps and (i + 1) % self.scheduler.order == 0):
                    progress_bar.update()

        if lora_scale is not None:
            # remove `lora_scale` from each PEFT layer
            unscale_lora_layers(self.transformer, lora_scale)
            attention_kwargs["scale"] = lora_scale

        self._current_timestep = None

        if self.config.expand_timesteps:
            latents = (1 - first_frame_mask) * condition + first_frame_mask * latents

        if not output_type == "latent":
            latents = latents.to(self.vae.dtype)
            latents_mean = (
                ms.tensor(self.vae.config.latents_mean).view(1, self.vae.config.z_dim, 1, 1, 1).to(latents.dtype)
            )
            latents_std = 1.0 / ms.tensor(self.vae.config.latents_std).view(1, self.vae.config.z_dim, 1, 1, 1).to(
                latents.dtype
            )
            latents = latents / latents_std + latents_mean
            # TODO: we use pynative mode here since cache in vae.decode which not supported in graph mode
            with pynative_context():
                video = self.vae.decode(latents, return_dict=False)[0]
            video = self.video_processor.postprocess_video(video, output_type=output_type)
        else:
            video = latents

        if not return_dict:
            return (video,)

        return WanPipelineOutput(frames=video)

mindone.diffusers.WanImageToVideoPipeline.__call__(image, prompt=None, negative_prompt=None, height=480, width=832, num_frames=81, num_inference_steps=50, guidance_scale=5.0, guidance_scale_2=None, num_videos_per_prompt=1, generator=None, latents=None, prompt_embeds=None, negative_prompt_embeds=None, image_embeds=None, last_image=None, output_type='np', return_dict=False, attention_kwargs=None, callback_on_step_end=None, callback_on_step_end_tensor_inputs=['latents'], max_sequence_length=512)

The call function to the pipeline for generation.

PARAMETER	DESCRIPTION
image	The input image to condition the generation on. Must be an image, a list of images or a ms.Tensor. TYPE: `PipelineImageInput`
prompt	The prompt or prompts to guide the image generation. If not defined, one has to pass prompt_embeds. instead. TYPE: `str` or `List[str]`, *optional* DEFAULT: None
negative_prompt	The prompt or prompts not to guide the image generation. If not defined, one has to pass negative_prompt_embeds instead. Ignored when not using guidance (i.e., ignored if guidance_scale is less than 1). TYPE: `str` or `List[str]`, *optional* DEFAULT: None
height	The height of the generated video. TYPE: `int`, defaults to `480` DEFAULT: 480
width	The width of the generated video. TYPE: `int`, defaults to `832` DEFAULT: 832
num_frames	The number of frames in the generated video. TYPE: `int`, defaults to `81` DEFAULT: 81
num_inference_steps	The number of denoising steps. More denoising steps usually lead to a higher quality image at the expense of slower inference. TYPE: `int`, defaults to `50` DEFAULT: 50
guidance_scale	Guidance scale as defined in Classifier-Free Diffusion Guidance. guidance_scale is defined as w of equation 2. of Imagen Paper. Guidance scale is enabled by setting guidance_scale > 1. Higher guidance scale encourages to generate images that are closely linked to the text prompt, usually at the expense of lower image quality. TYPE: `float`, defaults to `5.0` DEFAULT: 5.0
guidance_scale_2	Guidance scale for the low-noise stage transformer (transformer_2). If None and the pipeline's boundary_ratio is not None, uses the same value as guidance_scale. Only used when transformer_2 and the pipeline's boundary_ratio are not None. TYPE: `float`, *optional*, defaults to `None` DEFAULT: None
num_videos_per_prompt	The number of images to generate per prompt. TYPE: `int`, *optional*, defaults to 1 DEFAULT: 1
generator	A np.random.Generator to make generation deterministic. TYPE: `np.random.Generator` or `List[np.random.Generator]`, *optional* DEFAULT: None
latents	Pre-generated noisy latents sampled from a Gaussian distribution, to be used as inputs for image generation. Can be used to tweak the same generation with different prompts. If not provided, a latents tensor is generated by sampling using the supplied random generator. TYPE: `ms.Tensor`, *optional* DEFAULT: None
prompt_embeds	Pre-generated text embeddings. Can be used to easily tweak text inputs (prompt weighting). If not provided, text embeddings are generated from the prompt input argument. TYPE: `ms.Tensor`, *optional* DEFAULT: None
negative_prompt_embeds	Pre-generated text embeddings. Can be used to easily tweak text inputs (prompt weighting). If not provided, text embeddings are generated from the negative_prompt input argument. TYPE: `ms.Tensor`, *optional* DEFAULT: None
image_embeds	Pre-generated image embeddings. Can be used to easily tweak image inputs (weighting). If not provided, image embeddings are generated from the image input argument. TYPE: `ms.Tensor`, *optional* DEFAULT: None
output_type	The output format of the generated image. Choose between PIL.Image or np.array. TYPE: `str`, *optional*, defaults to `"np"` DEFAULT: 'np'
return_dict	Whether or not to return a [WanPipelineOutput] instead of a plain tuple. TYPE: `bool`, *optional*, defaults to `False` DEFAULT: False
attention_kwargs	A kwargs dictionary that if specified is passed along to the AttentionProcessor as defined under self.processor in diffusers.models.attention_processor. TYPE: `dict`, *optional* DEFAULT: None
callback_on_step_end	A function or a subclass of PipelineCallback or MultiPipelineCallbacks that is called at the end of each denoising step during the inference. with the following arguments: callback_on_step_end(self: DiffusionPipeline, step: int, timestep: int, callback_kwargs: Dict). callback_kwargs will include a list of all tensors as specified by callback_on_step_end_tensor_inputs. TYPE: `Callable`, `PipelineCallback`, `MultiPipelineCallbacks`, *optional* DEFAULT: None
callback_on_step_end_tensor_inputs	The list of tensor inputs for the callback_on_step_end function. The tensors specified in the list will be passed as callback_kwargs argument. You will only be able to include variables listed in the ._callback_tensor_inputs attribute of your pipeline class. TYPE: `List`, *optional* DEFAULT: ['latents']
max_sequence_length	The maximum sequence length of the text encoder. If the prompt is longer than this, it will be truncated. If the prompt is shorter, it will be padded to this length. TYPE: `int`, defaults to `512` DEFAULT: 512

RETURNS	DESCRIPTION
	[~WanPipelineOutput] or tuple: If return_dict is True, [WanPipelineOutput] is returned, otherwise a tuple is returned where the first element is a list with the generated images and the second element is a list of bools indicating whether the corresponding generated image contains "not-safe-for-work" (nsfw) content.

Source code in mindone/diffusers/pipelines/wan/pipeline_wan_i2v.py

def __call__(
    self,
    image: PipelineImageInput,
    prompt: Union[str, List[str]] = None,
    negative_prompt: Union[str, List[str]] = None,
    height: int = 480,
    width: int = 832,
    num_frames: int = 81,
    num_inference_steps: int = 50,
    guidance_scale: float = 5.0,
    guidance_scale_2: Optional[float] = None,
    num_videos_per_prompt: Optional[int] = 1,
    generator: Optional[Union[np.random.Generator, List[np.random.Generator]]] = None,
    latents: Optional[ms.Tensor] = None,
    prompt_embeds: Optional[ms.Tensor] = None,
    negative_prompt_embeds: Optional[ms.Tensor] = None,
    image_embeds: Optional[ms.Tensor] = None,
    last_image: Optional[ms.Tensor] = None,
    output_type: Optional[str] = "np",
    return_dict: bool = False,
    attention_kwargs: Optional[Dict[str, Any]] = None,
    callback_on_step_end: Optional[
        Union[Callable[[int, int, Dict], None], PipelineCallback, MultiPipelineCallbacks]
    ] = None,
    callback_on_step_end_tensor_inputs: List[str] = ["latents"],
    max_sequence_length: int = 512,
):
    r"""
    The call function to the pipeline for generation.

    Args:
        image (`PipelineImageInput`):
            The input image to condition the generation on. Must be an image, a list of images or a `ms.Tensor`.
        prompt (`str` or `List[str]`, *optional*):
            The prompt or prompts to guide the image generation. If not defined, one has to pass `prompt_embeds`.
            instead.
        negative_prompt (`str` or `List[str]`, *optional*):
            The prompt or prompts not to guide the image generation. If not defined, one has to pass
            `negative_prompt_embeds` instead. Ignored when not using guidance (i.e., ignored if `guidance_scale` is
            less than `1`).
        height (`int`, defaults to `480`):
            The height of the generated video.
        width (`int`, defaults to `832`):
            The width of the generated video.
        num_frames (`int`, defaults to `81`):
            The number of frames in the generated video.
        num_inference_steps (`int`, defaults to `50`):
            The number of denoising steps. More denoising steps usually lead to a higher quality image at the
            expense of slower inference.
        guidance_scale (`float`, defaults to `5.0`):
            Guidance scale as defined in [Classifier-Free Diffusion
            Guidance](https://huggingface.co/papers/2207.12598). `guidance_scale` is defined as `w` of equation 2.
            of [Imagen Paper](https://huggingface.co/papers/2205.11487). Guidance scale is enabled by setting
            `guidance_scale > 1`. Higher guidance scale encourages to generate images that are closely linked to
            the text `prompt`, usually at the expense of lower image quality.
        guidance_scale_2 (`float`, *optional*, defaults to `None`):
            Guidance scale for the low-noise stage transformer (`transformer_2`). If `None` and the pipeline's
            `boundary_ratio` is not None, uses the same value as `guidance_scale`. Only used when `transformer_2`
            and the pipeline's `boundary_ratio` are not None.
        num_videos_per_prompt (`int`, *optional*, defaults to 1):
            The number of images to generate per prompt.
        generator (`np.random.Generator` or `List[np.random.Generator]`, *optional*):
            A [`np.random.Generator`](https://numpy.org/doc/stable/reference/random/generator.html) to make
            generation deterministic.
        latents (`ms.Tensor`, *optional*):
            Pre-generated noisy latents sampled from a Gaussian distribution, to be used as inputs for image
            generation. Can be used to tweak the same generation with different prompts. If not provided, a latents
            tensor is generated by sampling using the supplied random `generator`.
        prompt_embeds (`ms.Tensor`, *optional*):
            Pre-generated text embeddings. Can be used to easily tweak text inputs (prompt weighting). If not
            provided, text embeddings are generated from the `prompt` input argument.
        negative_prompt_embeds (`ms.Tensor`, *optional*):
            Pre-generated text embeddings. Can be used to easily tweak text inputs (prompt weighting). If not
            provided, text embeddings are generated from the `negative_prompt` input argument.
        image_embeds (`ms.Tensor`, *optional*):
            Pre-generated image embeddings. Can be used to easily tweak image inputs (weighting). If not provided,
            image embeddings are generated from the `image` input argument.
        output_type (`str`, *optional*, defaults to `"np"`):
            The output format of the generated image. Choose between `PIL.Image` or `np.array`.
        return_dict (`bool`, *optional*, defaults to `False`):
            Whether or not to return a [`WanPipelineOutput`] instead of a plain tuple.
        attention_kwargs (`dict`, *optional*):
            A kwargs dictionary that if specified is passed along to the `AttentionProcessor` as defined under
            `self.processor` in
            [diffusers.models.attention_processor](https://github.com/huggingface/diffusers/blob/main/src/diffusers/models/attention_processor.py).
        callback_on_step_end (`Callable`, `PipelineCallback`, `MultiPipelineCallbacks`, *optional*):
            A function or a subclass of `PipelineCallback` or `MultiPipelineCallbacks` that is called at the end of
            each denoising step during the inference. with the following arguments: `callback_on_step_end(self:
            DiffusionPipeline, step: int, timestep: int, callback_kwargs: Dict)`. `callback_kwargs` will include a
            list of all tensors as specified by `callback_on_step_end_tensor_inputs`.
        callback_on_step_end_tensor_inputs (`List`, *optional*):
            The list of tensor inputs for the `callback_on_step_end` function. The tensors specified in the list
            will be passed as `callback_kwargs` argument. You will only be able to include variables listed in the
            `._callback_tensor_inputs` attribute of your pipeline class.
        max_sequence_length (`int`, defaults to `512`):
            The maximum sequence length of the text encoder. If the prompt is longer than this, it will be
            truncated. If the prompt is shorter, it will be padded to this length.

    Examples:

    Returns:
        [`~WanPipelineOutput`] or `tuple`:
            If `return_dict` is `True`, [`WanPipelineOutput`] is returned, otherwise a `tuple` is returned where
            the first element is a list with the generated images and the second element is a list of `bool`s
            indicating whether the corresponding generated image contains "not-safe-for-work" (nsfw) content.
    """

    if isinstance(callback_on_step_end, (PipelineCallback, MultiPipelineCallbacks)):
        callback_on_step_end_tensor_inputs = callback_on_step_end.tensor_inputs

    # 1. Check inputs. Raise error if not correct
    self.check_inputs(
        prompt,
        negative_prompt,
        image,
        height,
        width,
        prompt_embeds,
        negative_prompt_embeds,
        image_embeds,
        callback_on_step_end_tensor_inputs,
        guidance_scale_2,
    )

    if num_frames % self.vae_scale_factor_temporal != 1:
        logger.warning(
            f"`num_frames - 1` has to be divisible by {self.vae_scale_factor_temporal}. Rounding to the nearest number."
        )
        num_frames = num_frames // self.vae_scale_factor_temporal * self.vae_scale_factor_temporal + 1
    num_frames = max(num_frames, 1)

    if self.config.boundary_ratio is not None and guidance_scale_2 is None:
        guidance_scale_2 = guidance_scale

    self._guidance_scale = guidance_scale
    self._guidance_scale_2 = guidance_scale_2
    self._attention_kwargs = attention_kwargs
    self._current_timestep = None
    self._interrupt = False

    # 2. Define call parameters
    if prompt is not None and isinstance(prompt, str):
        batch_size = 1
    elif prompt is not None and isinstance(prompt, list):
        batch_size = len(prompt)
    else:
        batch_size = prompt_embeds.shape[0]

    # 3. Encode input prompt
    prompt_embeds, negative_prompt_embeds = self.encode_prompt(
        prompt=prompt,
        negative_prompt=negative_prompt,
        do_classifier_free_guidance=self.do_classifier_free_guidance,
        num_videos_per_prompt=num_videos_per_prompt,
        prompt_embeds=prompt_embeds,
        negative_prompt_embeds=negative_prompt_embeds,
        max_sequence_length=max_sequence_length,
    )

    # Encode image embedding
    transformer_dtype = self.transformer.dtype if self.transformer is not None else self.transformer_2.dtype
    prompt_embeds = prompt_embeds.to(transformer_dtype)
    if negative_prompt_embeds is not None:
        negative_prompt_embeds = negative_prompt_embeds.to(transformer_dtype)

    # only wan 2.1 i2v transformer accepts image_embeds
    if self.transformer is not None and self.transformer.config.image_dim is not None:
        if image_embeds is None:
            if last_image is None:
                image_embeds = self.encode_image(image)
            else:
                image_embeds = self.encode_image([image, last_image])
        image_embeds = image_embeds.tile((batch_size, 1, 1))
        image_embeds = image_embeds.to(transformer_dtype)

    # 4. Prepare timesteps
    self.scheduler.set_timesteps(num_inference_steps)
    timesteps = self.scheduler.timesteps

    # 5. Prepare latent variables
    num_channels_latents = self.vae.config.z_dim
    image = self.video_processor.preprocess(image, height=height, width=width).to(dtype=ms.float32)
    if last_image is not None:
        last_image = self.video_processor.preprocess(last_image, height=height, width=width).to(dtype=ms.float32)

    latents_outputs = self.prepare_latents(
        image,
        batch_size * num_videos_per_prompt,
        num_channels_latents,
        height,
        width,
        num_frames,
        ms.float32,
        generator,
        latents,
        last_image,
    )
    if self.config.expand_timesteps:
        # wan 2.2 5b i2v use firt_frame_mask to mask timesteps
        latents, condition, first_frame_mask = latents_outputs
    else:
        latents, condition = latents_outputs

    # 6. Denoising loop
    num_warmup_steps = len(timesteps) - num_inference_steps * self.scheduler.order
    self._num_timesteps = len(timesteps)

    if self.config.boundary_ratio is not None:
        boundary_timestep = self.config.boundary_ratio * self.scheduler.config.num_train_timesteps
    else:
        boundary_timestep = None

    # we're popping the `scale` instead of getting it because otherwise `scale` will be propagated
    # to the transformer and will raise RuntimeError.
    lora_scale = attention_kwargs.pop("scale", None) if attention_kwargs is not None else None
    if lora_scale is not None:
        # weight the lora layers by setting `lora_scale` for each PEFT layer
        scale_lora_layers(self.transformer, lora_scale)

    with self.progress_bar(total=num_inference_steps) as progress_bar:
        for i, t in enumerate(timesteps):
            if self.interrupt:
                continue

            self._current_timestep = t

            if boundary_timestep is None or t >= boundary_timestep:
                # wan2.1 or high-noise stage in wan2.2
                current_model = self.transformer
                current_guidance_scale = guidance_scale
            else:
                # low-noise stage in wan2.2
                current_model = self.transformer_2
                current_guidance_scale = guidance_scale_2

            if self.config.expand_timesteps:
                latent_model_input = (1 - first_frame_mask) * condition + first_frame_mask * latents
                latent_model_input = latent_model_input.to(transformer_dtype)

                # seq_len: num_latent_frames * (latent_height // patch_size) * (latent_width // patch_size)
                temp_ts = (first_frame_mask[0][0][:, ::2, ::2] * t).flatten()
                # batch_size, seq_len
                timestep = temp_ts.unsqueeze(0).broadcast_to((latents.shape[0], -1))
            else:
                latent_model_input = mint.cat([latents, condition], dim=1).to(transformer_dtype)
                timestep = t.broadcast_to((latents.shape[0],))

            with current_model.cache_context("cond"):
                noise_pred = current_model(
                    hidden_states=latent_model_input,
                    timestep=timestep,
                    encoder_hidden_states=prompt_embeds,
                    encoder_hidden_states_image=image_embeds,
                    attention_kwargs=attention_kwargs,
                    return_dict=False,
                )[0]

            if self.do_classifier_free_guidance:
                with current_model.cache_context("uncond"):
                    noise_uncond = current_model(
                        hidden_states=latent_model_input,
                        timestep=timestep,
                        encoder_hidden_states=negative_prompt_embeds,
                        encoder_hidden_states_image=image_embeds,
                        attention_kwargs=attention_kwargs,
                        return_dict=False,
                    )[0]
                    noise_pred = noise_uncond + current_guidance_scale * (noise_pred - noise_uncond)

            # compute the previous noisy sample x_t -> x_t-1
            latents = self.scheduler.step(noise_pred, t, latents, return_dict=False)[0]

            if callback_on_step_end is not None:
                callback_kwargs = {}
                for k in callback_on_step_end_tensor_inputs:
                    callback_kwargs[k] = locals()[k]
                callback_outputs = callback_on_step_end(self, i, t, callback_kwargs)

                latents = callback_outputs.pop("latents", latents)
                prompt_embeds = callback_outputs.pop("prompt_embeds", prompt_embeds)
                negative_prompt_embeds = callback_outputs.pop("negative_prompt_embeds", negative_prompt_embeds)

            # call the callback, if provided
            if i == len(timesteps) - 1 or ((i + 1) > num_warmup_steps and (i + 1) % self.scheduler.order == 0):
                progress_bar.update()

    if lora_scale is not None:
        # remove `lora_scale` from each PEFT layer
        unscale_lora_layers(self.transformer, lora_scale)
        attention_kwargs["scale"] = lora_scale

    self._current_timestep = None

    if self.config.expand_timesteps:
        latents = (1 - first_frame_mask) * condition + first_frame_mask * latents

    if not output_type == "latent":
        latents = latents.to(self.vae.dtype)
        latents_mean = (
            ms.tensor(self.vae.config.latents_mean).view(1, self.vae.config.z_dim, 1, 1, 1).to(latents.dtype)
        )
        latents_std = 1.0 / ms.tensor(self.vae.config.latents_std).view(1, self.vae.config.z_dim, 1, 1, 1).to(
            latents.dtype
        )
        latents = latents / latents_std + latents_mean
        # TODO: we use pynative mode here since cache in vae.decode which not supported in graph mode
        with pynative_context():
            video = self.vae.decode(latents, return_dict=False)[0]
        video = self.video_processor.postprocess_video(video, output_type=output_type)
    else:
        video = latents

    if not return_dict:
        return (video,)

    return WanPipelineOutput(frames=video)

mindone.diffusers.WanImageToVideoPipeline.encode_prompt(prompt, negative_prompt=None, do_classifier_free_guidance=True, num_videos_per_prompt=1, prompt_embeds=None, negative_prompt_embeds=None, max_sequence_length=226, dtype=None)

Encodes the prompt into text encoder hidden states.

PARAMETER	DESCRIPTION
prompt	prompt to be encoded TYPE: `str` or `List[str]`, *optional*
negative_prompt	The prompt or prompts not to guide the image generation. If not defined, one has to pass negative_prompt_embeds instead. Ignored when not using guidance (i.e., ignored if guidance_scale is less than 1). TYPE: `str` or `List[str]`, *optional* DEFAULT: None
do_classifier_free_guidance	Whether to use classifier free guidance or not. TYPE: `bool`, *optional*, defaults to `True` DEFAULT: True
num_videos_per_prompt	Number of videos that should be generated per prompt. TYPE: `int`, *optional*, defaults to 1 DEFAULT: 1
prompt_embeds	Pre-generated text embeddings. Can be used to easily tweak text inputs, e.g. prompt weighting. If not provided, text embeddings will be generated from prompt input argument. TYPE: `ms.Tensor`, *optional* DEFAULT: None
negative_prompt_embeds	Pre-generated negative text embeddings. Can be used to easily tweak text inputs, e.g. prompt weighting. If not provided, negative_prompt_embeds will be generated from negative_prompt input argument. TYPE: `ms.Tensor`, *optional* DEFAULT: None
dtype	(ms.Type, optional): mindspore dtype TYPE: Optional[Type] DEFAULT: None

Source code in mindone/diffusers/pipelines/wan/pipeline_wan_i2v.py

def encode_prompt(
    self,
    prompt: Union[str, List[str]],
    negative_prompt: Optional[Union[str, List[str]]] = None,
    do_classifier_free_guidance: bool = True,
    num_videos_per_prompt: int = 1,
    prompt_embeds: Optional[ms.Tensor] = None,
    negative_prompt_embeds: Optional[ms.Tensor] = None,
    max_sequence_length: int = 226,
    dtype: Optional[ms.Type] = None,
):
    r"""
    Encodes the prompt into text encoder hidden states.

    Args:
        prompt (`str` or `List[str]`, *optional*):
            prompt to be encoded
        negative_prompt (`str` or `List[str]`, *optional*):
            The prompt or prompts not to guide the image generation. If not defined, one has to pass
            `negative_prompt_embeds` instead. Ignored when not using guidance (i.e., ignored if `guidance_scale` is
            less than `1`).
        do_classifier_free_guidance (`bool`, *optional*, defaults to `True`):
            Whether to use classifier free guidance or not.
        num_videos_per_prompt (`int`, *optional*, defaults to 1):
            Number of videos that should be generated per prompt.
        prompt_embeds (`ms.Tensor`, *optional*):
            Pre-generated text embeddings. Can be used to easily tweak text inputs, *e.g.* prompt weighting. If not
            provided, text embeddings will be generated from `prompt` input argument.
        negative_prompt_embeds (`ms.Tensor`, *optional*):
            Pre-generated negative text embeddings. Can be used to easily tweak text inputs, *e.g.* prompt
            weighting. If not provided, negative_prompt_embeds will be generated from `negative_prompt` input
            argument.
        dtype: (`ms.Type`, *optional*):
            mindspore dtype
    """
    prompt = [prompt] if isinstance(prompt, str) else prompt
    if prompt is not None:
        batch_size = len(prompt)
    else:
        batch_size = prompt_embeds.shape[0]

    if prompt_embeds is None:
        prompt_embeds = self._get_t5_prompt_embeds(
            prompt=prompt,
            num_videos_per_prompt=num_videos_per_prompt,
            max_sequence_length=max_sequence_length,
            dtype=dtype,
        )

    if do_classifier_free_guidance and negative_prompt_embeds is None:
        negative_prompt = negative_prompt or ""
        negative_prompt = batch_size * [negative_prompt] if isinstance(negative_prompt, str) else negative_prompt

        if prompt is not None and type(prompt) is not type(negative_prompt):
            raise TypeError(
                f"`negative_prompt` should be the same type to `prompt`, but got {type(negative_prompt)} !="
                f" {type(prompt)}."
            )
        elif batch_size != len(negative_prompt):
            raise ValueError(
                f"`negative_prompt`: {negative_prompt} has batch size {len(negative_prompt)}, but `prompt`:"
                f" {prompt} has batch size {batch_size}. Please make sure that passed `negative_prompt` matches"
                " the batch size of `prompt`."
            )

        negative_prompt_embeds = self._get_t5_prompt_embeds(
            prompt=negative_prompt,
            num_videos_per_prompt=num_videos_per_prompt,
            max_sequence_length=max_sequence_length,
            dtype=dtype,
        )

    return prompt_embeds, negative_prompt_embeds

mindone.diffusers.WanVACEPipeline

Bases: DiffusionPipeline, WanLoraLoaderMixin

Pipeline for controllable generation using Wan.

This model inherits from [DiffusionPipeline]. Check the superclass documentation for the generic methods implemented for all pipelines (downloading, saving, etc.).

PARAMETER	DESCRIPTION
tokenizer	Tokenizer from T5, specifically the google/umt5-xxl variant. TYPE: [`T5Tokenizer`]
text_encoder	T5, specifically the google/umt5-xxl variant. TYPE: [`T5EncoderModel`]
transformer	Conditional Transformer to denoise the input latents. TYPE: [`WanTransformer3DModel`]
scheduler	A scheduler to be used in combination with transformer to denoise the encoded image latents. TYPE: [`UniPCMultistepScheduler`]
vae	Variational Auto-Encoder (VAE) Model to encode and decode videos to and from latent representations. TYPE: [`AutoencoderKLWan`]

Source code in mindone/diffusers/pipelines/wan/pipeline_wan_vace.py

class WanVACEPipeline(DiffusionPipeline, WanLoraLoaderMixin):
    r"""
    Pipeline for controllable generation using Wan.

    This model inherits from [`DiffusionPipeline`]. Check the superclass documentation for the generic methods
    implemented for all pipelines (downloading, saving, etc.).

    Args:
        tokenizer ([`T5Tokenizer`]):
            Tokenizer from [T5](https://huggingface.co/docs/transformers/en/model_doc/t5#transformers.T5Tokenizer),
            specifically the [google/umt5-xxl](https://huggingface.co/google/umt5-xxl) variant.
        text_encoder ([`T5EncoderModel`]):
            [T5](https://huggingface.co/docs/transformers/en/model_doc/t5#transformers.T5EncoderModel), specifically
            the [google/umt5-xxl](https://huggingface.co/google/umt5-xxl) variant.
        transformer ([`WanTransformer3DModel`]):
            Conditional Transformer to denoise the input latents.
        scheduler ([`UniPCMultistepScheduler`]):
            A scheduler to be used in combination with `transformer` to denoise the encoded image latents.
        vae ([`AutoencoderKLWan`]):
            Variational Auto-Encoder (VAE) Model to encode and decode videos to and from latent representations.
    """

    model_cpu_offload_seq = "text_encoder->transformer->vae"
    _callback_tensor_inputs = ["latents", "prompt_embeds", "negative_prompt_embeds"]

    def __init__(
        self,
        tokenizer: AutoTokenizer,
        text_encoder: UMT5EncoderModel,
        transformer: WanVACETransformer3DModel,
        vae: AutoencoderKLWan,
        scheduler: FlowMatchEulerDiscreteScheduler,
    ):
        super().__init__()

        self.register_modules(
            vae=vae,
            text_encoder=text_encoder,
            tokenizer=tokenizer,
            transformer=transformer,
            scheduler=scheduler,
        )

        self.vae_scale_factor_temporal = 2 ** sum(self.vae.temperal_downsample) if getattr(self, "vae", None) else 4
        self.vae_scale_factor_spatial = 2 ** len(self.vae.temperal_downsample) if getattr(self, "vae", None) else 8
        self.video_processor = VideoProcessor(vae_scale_factor=self.vae_scale_factor_spatial)

    # Copied from diffusers.pipelines.wan.pipeline_wan.WanPipeline._get_t5_prompt_embeds
    def _get_t5_prompt_embeds(
        self,
        prompt: Union[str, List[str]] = None,
        num_videos_per_prompt: int = 1,
        max_sequence_length: int = 226,
        dtype: Optional[ms.Type] = None,
    ):
        dtype = dtype or self.text_encoder.dtype

        prompt = [prompt] if isinstance(prompt, str) else prompt
        prompt = [prompt_clean(u) for u in prompt]
        batch_size = len(prompt)

        text_inputs = self.tokenizer(
            prompt,
            padding="max_length",
            max_length=max_sequence_length,
            truncation=True,
            add_special_tokens=True,
            return_attention_mask=True,
            return_tensors="np",
        )
        text_input_ids, mask = ms.tensor(text_inputs.input_ids), ms.tensor(text_inputs.attention_mask)
        seq_lens = mask.gt(0).sum(dim=1).long()

        with pynative_context():
            prompt_embeds = self.text_encoder(text_input_ids, mask)[0]
        prompt_embeds = prompt_embeds.to(dtype=dtype)
        prompt_embeds = [u[:v] for u, v in zip(prompt_embeds, seq_lens)]
        prompt_embeds = mint.stack(
            [mint.cat([u, u.new_zeros((max_sequence_length - u.shape[0], u.shape[1]))]) for u in prompt_embeds], dim=0
        )

        # duplicate text embeddings for each generation per prompt, using mps friendly method
        _, seq_len, _ = prompt_embeds.shape
        prompt_embeds = prompt_embeds.tile((1, num_videos_per_prompt, 1))
        prompt_embeds = prompt_embeds.view(batch_size * num_videos_per_prompt, seq_len, -1)

        return prompt_embeds

    # Copied from diffusers.pipelines.wan.pipeline_wan.WanPipeline.encode_prompt
    def encode_prompt(
        self,
        prompt: Union[str, List[str]],
        negative_prompt: Optional[Union[str, List[str]]] = None,
        do_classifier_free_guidance: bool = True,
        num_videos_per_prompt: int = 1,
        prompt_embeds: Optional[ms.Tensor] = None,
        negative_prompt_embeds: Optional[ms.Tensor] = None,
        max_sequence_length: int = 226,
        dtype: Optional[ms.Type] = None,
    ):
        r"""
        Encodes the prompt into text encoder hidden states.

        Args:
            prompt (`str` or `List[str]`, *optional*):
                prompt to be encoded
            negative_prompt (`str` or `List[str]`, *optional*):
                The prompt or prompts not to guide the image generation. If not defined, one has to pass
                `negative_prompt_embeds` instead. Ignored when not using guidance (i.e., ignored if `guidance_scale` is
                less than `1`).
            do_classifier_free_guidance (`bool`, *optional*, defaults to `True`):
                Whether to use classifier free guidance or not.
            num_videos_per_prompt (`int`, *optional*, defaults to 1):
                Number of videos that should be generated per prompt.
            prompt_embeds (`ms.Tensor`, *optional*):
                Pre-generated text embeddings. Can be used to easily tweak text inputs, *e.g.* prompt weighting. If not
                provided, text embeddings will be generated from `prompt` input argument.
            negative_prompt_embeds (`ms.Tensor`, *optional*):
                Pre-generated negative text embeddings. Can be used to easily tweak text inputs, *e.g.* prompt
                weighting. If not provided, negative_prompt_embeds will be generated from `negative_prompt` input
                argument.
            dtype: (`ms.dtype`, *optional*):
                ms dtype
        """
        prompt = [prompt] if isinstance(prompt, str) else prompt
        if prompt is not None:
            batch_size = len(prompt)
        else:
            batch_size = prompt_embeds.shape[0]

        if prompt_embeds is None:
            prompt_embeds = self._get_t5_prompt_embeds(
                prompt=prompt,
                num_videos_per_prompt=num_videos_per_prompt,
                max_sequence_length=max_sequence_length,
                dtype=dtype,
            )

        if do_classifier_free_guidance and negative_prompt_embeds is None:
            negative_prompt = negative_prompt or ""
            negative_prompt = batch_size * [negative_prompt] if isinstance(negative_prompt, str) else negative_prompt

            if prompt is not None and type(prompt) is not type(negative_prompt):
                raise TypeError(
                    f"`negative_prompt` should be the same type to `prompt`, but got {type(negative_prompt)} !="
                    f" {type(prompt)}."
                )
            elif batch_size != len(negative_prompt):
                raise ValueError(
                    f"`negative_prompt`: {negative_prompt} has batch size {len(negative_prompt)}, but `prompt`:"
                    f" {prompt} has batch size {batch_size}. Please make sure that passed `negative_prompt` matches"
                    " the batch size of `prompt`."
                )

            negative_prompt_embeds = self._get_t5_prompt_embeds(
                prompt=negative_prompt,
                num_videos_per_prompt=num_videos_per_prompt,
                max_sequence_length=max_sequence_length,
                dtype=dtype,
            )

        return prompt_embeds, negative_prompt_embeds

    def check_inputs(
        self,
        prompt,
        negative_prompt,
        height,
        width,
        prompt_embeds=None,
        negative_prompt_embeds=None,
        callback_on_step_end_tensor_inputs=None,
        video=None,
        mask=None,
        reference_images=None,
    ):
        base = self.vae_scale_factor_spatial * self.transformer.config.patch_size[1]
        if height % base != 0 or width % base != 0:
            raise ValueError(f"`height` and `width` have to be divisible by {base} but are {height} and {width}.")

        if callback_on_step_end_tensor_inputs is not None and not all(
            k in self._callback_tensor_inputs for k in callback_on_step_end_tensor_inputs
        ):
            raise ValueError(
                f"`callback_on_step_end_tensor_inputs` has to be in "
                f"{self._callback_tensor_inputs}, but found "
                f"{[k for k in callback_on_step_end_tensor_inputs if k not in self._callback_tensor_inputs]}"
            )

        if prompt is not None and prompt_embeds is not None:
            raise ValueError(
                f"Cannot forward both `prompt`: {prompt} and `prompt_embeds`: {prompt_embeds}. Please make sure to"
                " only forward one of the two."
            )
        elif negative_prompt is not None and negative_prompt_embeds is not None:
            raise ValueError(
                f"Cannot forward both `negative_prompt`: {negative_prompt} and `negative_prompt_embeds`: {negative_prompt_embeds}. Please make sure to"
                " only forward one of the two."
            )
        elif prompt is None and prompt_embeds is None:
            raise ValueError(
                "Provide either `prompt` or `prompt_embeds`. Cannot leave both `prompt` and `prompt_embeds` undefined."
            )
        elif prompt is not None and (not isinstance(prompt, str) and not isinstance(prompt, list)):
            raise ValueError(f"`prompt` has to be of type `str` or `list` but is {type(prompt)}")
        elif negative_prompt is not None and (
            not isinstance(negative_prompt, str) and not isinstance(negative_prompt, list)
        ):
            raise ValueError(f"`negative_prompt` has to be of type `str` or `list` but is {type(negative_prompt)}")

        if video is not None:
            if mask is not None:
                if len(video) != len(mask):
                    raise ValueError(
                        f"Length of `video` {len(video)} and `mask` {len(mask)} do not match. Please make sure that"
                        " they have the same length."
                    )
            if reference_images is not None:
                is_pil_image = isinstance(reference_images, PIL.Image.Image)
                is_list_of_pil_images = isinstance(reference_images, list) and all(
                    isinstance(ref_img, PIL.Image.Image) for ref_img in reference_images
                )
                is_list_of_list_of_pil_images = isinstance(reference_images, list) and all(
                    isinstance(ref_img, list) and all(isinstance(ref_img_, PIL.Image.Image) for ref_img_ in ref_img)
                    for ref_img in reference_images
                )
                if not (is_pil_image or is_list_of_pil_images or is_list_of_list_of_pil_images):
                    raise ValueError(
                        "`reference_images` has to be of type `PIL.Image.Image` or `list` of `PIL.Image.Image`, or "
                        "`list` of `list` of `PIL.Image.Image`, but is {type(reference_images)}"
                    )
                if is_list_of_list_of_pil_images and len(reference_images) != 1:
                    raise ValueError(
                        "The pipeline only supports generating one video at a time at the moment. When passing a list "
                        "of list of reference images, where the outer list corresponds to the batch size and the inner "
                        "list corresponds to list of conditioning images per video, please make sure to only pass "
                        "one inner list of reference images (i.e., `[[<image1>, <image2>, ...]]`"
                    )
        elif mask is not None:
            raise ValueError("`mask` can only be passed if `video` is passed as well.")

    def preprocess_conditions(
        self,
        video: Optional[List[PipelineImageInput]] = None,
        mask: Optional[List[PipelineImageInput]] = None,
        reference_images: Optional[Union[PIL.Image.Image, List[PIL.Image.Image], List[List[PIL.Image.Image]]]] = None,
        batch_size: int = 1,
        height: int = 480,
        width: int = 832,
        num_frames: int = 81,
        dtype: Optional[ms.Type] = None,
    ):
        if video is not None:
            base = self.vae_scale_factor_spatial * self.transformer.config.patch_size[1]
            video_height, video_width = self.video_processor.get_default_height_width(video[0])

            if video_height * video_width > height * width:
                scale = min(width / video_width, height / video_height)
                video_height, video_width = int(video_height * scale), int(video_width * scale)

            if video_height % base != 0 or video_width % base != 0:
                logger.warning(
                    f"Video height and width should be divisible by {base}, but got {video_height} and {video_width}. "
                )
                video_height = (video_height // base) * base
                video_width = (video_width // base) * base

            assert video_height * video_width <= height * width

            video = self.video_processor.preprocess_video(video, video_height, video_width)
            image_size = (video_height, video_width)  # Use the height/width of video (with possible rescaling)
        else:
            video = mint.zeros((batch_size, 3, num_frames, height, width), dtype=dtype)
            image_size = (height, width)  # Use the height/width provider by user

        if mask is not None:
            mask = self.video_processor.preprocess_video(mask, image_size[0], image_size[1])
            mask = mint.clamp((mask + 1) / 2, min=0, max=1)
        else:
            mask = mint.ones_like(video)

        video = video.to(dtype=dtype)
        mask = mask.to(dtype=dtype)

        # Make a list of list of images where the outer list corresponds to video batch size and the inner list
        # corresponds to list of conditioning images per video
        if reference_images is None or isinstance(reference_images, PIL.Image.Image):
            reference_images = [[reference_images] for _ in range(video.shape[0])]
        elif isinstance(reference_images, (list, tuple)) and isinstance(next(iter(reference_images)), PIL.Image.Image):
            reference_images = [reference_images]
        elif (
            isinstance(reference_images, (list, tuple))
            and isinstance(next(iter(reference_images)), list)
            and isinstance(next(iter(reference_images[0])), PIL.Image.Image)
        ):
            reference_images = reference_images
        else:
            raise ValueError(
                "`reference_images` has to be of type `PIL.Image.Image` or `list` of `PIL.Image.Image`, or "
                "`list` of `list` of `PIL.Image.Image`, but is {type(reference_images)}"
            )

        if video.shape[0] != len(reference_images):
            raise ValueError(
                f"Batch size of `video` {video.shape[0]} and length of `reference_images` {len(reference_images)} "
                f"does not match."
            )

        ref_images_lengths = [len(reference_images_batch) for reference_images_batch in reference_images]
        if any(length != ref_images_lengths[0] for length in ref_images_lengths):
            raise ValueError(
                f"All batches of `reference_images` should have the same length, but got {ref_images_lengths}. Support for this "
                "may be added in the future."
            )

        reference_images_preprocessed = []
        for i, reference_images_batch in enumerate(reference_images):
            preprocessed_images = []
            for j, image in enumerate(reference_images_batch):
                if image is None:
                    continue
                image = self.video_processor.preprocess(image, None, None)
                img_height, img_width = image.shape[-2:]
                scale = min(image_size[0] / img_height, image_size[1] / img_width)
                new_height, new_width = int(img_height * scale), int(img_width * scale)
                resized_image = mint.nn.functional.interpolate(
                    image, size=(new_height, new_width), mode="bilinear", align_corners=False
                ).squeeze(
                    0
                )  # [C, H, W]
                top = (image_size[0] - new_height) // 2
                left = (image_size[1] - new_width) // 2
                canvas = mint.ones((3, *image_size), dtype=dtype)
                canvas[:, top : top + new_height, left : left + new_width] = resized_image
                preprocessed_images.append(canvas)
            reference_images_preprocessed.append(preprocessed_images)

        return video, mask, reference_images_preprocessed

    def prepare_video_latents(
        self,
        video: ms.Tensor,
        mask: ms.Tensor,
        reference_images: Optional[List[List[ms.Tensor]]] = None,
        generator: Optional[Union[ms.Generator, List[ms.Generator]]] = None,
    ) -> ms.Tensor:
        if isinstance(generator, list):
            # TODO: support this
            raise ValueError("Passing a list of generators is not yet supported. This may be supported in the future.")

        if reference_images is None:
            # For each batch of video, we set no re
            # ference image (as one or more can be passed by user)
            reference_images = [[None] for _ in range(video.shape[0])]
        else:
            if video.shape[0] != len(reference_images):
                raise ValueError(
                    f"Batch size of `video` {video.shape[0]} and length of `reference_images` {len(reference_images)} "
                    f"does not match."
                )

        if video.shape[0] != 1:
            # TODO: support this
            raise ValueError(
                "Generating with more than one video is not yet supported. This may be supported in the future."
            )

        vae_dtype = self.vae.dtype
        video = video.to(dtype=vae_dtype)

        latents_mean = ms.tensor(self.vae.config.latents_mean, dtype=ms.float32).view(1, self.vae.config.z_dim, 1, 1, 1)
        latents_std = 1.0 / ms.tensor(self.vae.config.latents_std, dtype=ms.float32).view(
            1, self.vae.config.z_dim, 1, 1, 1
        )

        if mask is None:
            latents = retrieve_latents(self.vae.encode(video), generator, sample_mode="argmax").unbind(0)
            latents = ((latents.float() - latents_mean) * latents_std).to(vae_dtype)
        else:
            mask = mask.to(dtype=vae_dtype)
            mask = mint.where(mask > 0.5, 1.0, 0.0)
            inactive = video * (1 - mask)
            reactive = video * mask
            with pynative_context():
                inactive = retrieve_latents(self.vae, self.vae.encode(inactive)[0], sample_mode="argmax")
                reactive = retrieve_latents(self.vae, self.vae.encode(reactive)[0], sample_mode="argmax")
            inactive = ((inactive.float() - latents_mean) * latents_std).to(vae_dtype)
            reactive = ((reactive.float() - latents_mean) * latents_std).to(vae_dtype)
            latents = mint.cat([inactive, reactive], dim=1)

        latent_list = []
        for latent, reference_images_batch in zip(latents, reference_images):
            for reference_image in reference_images_batch:
                assert reference_image.ndim == 3
                reference_image = reference_image.to(dtype=vae_dtype)
                reference_image = reference_image[None, :, None, :, :]  # [1, C, 1, H, W]
                reference_latent = retrieve_latents(self.vae, self.vae.encode(reference_image)[0], sample_mode="argmax")
                reference_latent = ((reference_latent.float() - latents_mean) * latents_std).to(vae_dtype)
                reference_latent = reference_latent.squeeze(0)  # [C, 1, H, W]
                reference_latent = mint.cat([reference_latent, mint.zeros_like(reference_latent)], dim=0)
                latent = mint.cat([reference_latent.squeeze(0), latent], dim=1)
            latent_list.append(latent)
        return mint.stack(latent_list)

    def prepare_masks(
        self,
        mask: ms.Tensor,
        reference_images: Optional[List[ms.Tensor]] = None,
        generator: Optional[Union[ms.Generator, List[ms.Generator]]] = None,
    ) -> ms.Tensor:
        if isinstance(generator, list):
            # TODO: support this
            raise ValueError("Passing a list of generators is not yet supported. This may be supported in the future.")

        if reference_images is None:
            # For each batch of video, we set no reference image (as one or more can be passed by user)
            reference_images = [[None] for _ in range(mask.shape[0])]
        else:
            if mask.shape[0] != len(reference_images):
                raise ValueError(
                    f"Batch size of `mask` {mask.shape[0]} and length of `reference_images` {len(reference_images)} "
                    f"does not match."
                )

        if mask.shape[0] != 1:
            # TODO: support this
            raise ValueError(
                "Generating with more than one video is not yet supported. This may be supported in the future."
            )

        transformer_patch_size = self.transformer.config.patch_size[1]

        mask_list = []
        for mask_, reference_images_batch in zip(mask, reference_images):
            num_channels, num_frames, height, width = mask_.shape
            new_num_frames = (num_frames + self.vae_scale_factor_temporal - 1) // self.vae_scale_factor_temporal
            new_height = height // (self.vae_scale_factor_spatial * transformer_patch_size) * transformer_patch_size
            new_width = width // (self.vae_scale_factor_spatial * transformer_patch_size) * transformer_patch_size
            mask_ = mask_[0, :, :, :]
            mask_ = mask_.view(
                num_frames, new_height, self.vae_scale_factor_spatial, new_width, self.vae_scale_factor_spatial
            )
            mask_ = mask_.permute(2, 4, 0, 1, 3).flatten(0, 1)  # [8x8, num_frames, new_height, new_width]
            mask_ = mint.nn.functional.interpolate(
                mask_.unsqueeze(0), size=(new_num_frames, new_height, new_width), mode="nearest"
            ).squeeze(0)
            num_ref_images = len(reference_images_batch)
            if num_ref_images > 0:
                mask_padding = mint.zeros_like(mask_[:, :num_ref_images, :, :])
                mask_ = mint.cat([mask_padding, mask_], dim=1)
            mask_list.append(mask_)
        return mint.stack(mask_list)

    def prepare_latents(
        self,
        batch_size: int,
        num_channels_latents: int = 16,
        height: int = 480,
        width: int = 832,
        num_frames: int = 81,
        dtype: Optional[ms.Type] = None,
        generator: Optional[Union[ms.Generator, List[ms.Generator]]] = None,
        latents: Optional[ms.Tensor] = None,
    ) -> ms.Tensor:
        if latents is not None:
            return latents.to(dtype=dtype)

        num_latent_frames = (num_frames - 1) // self.vae_scale_factor_temporal + 1
        shape = (
            batch_size,
            num_channels_latents,
            num_latent_frames,
            int(height) // self.vae_scale_factor_spatial,
            int(width) // self.vae_scale_factor_spatial,
        )
        if isinstance(generator, list) and len(generator) != batch_size:
            raise ValueError(
                f"You have passed a list of generators of length {len(generator)}, but requested an effective batch"
                f" size of {batch_size}. Make sure the batch size matches the length of the generators."
            )

        latents = randn_tensor(shape, generator=generator, dtype=dtype)
        return latents

    @property
    def guidance_scale(self):
        return self._guidance_scale

    @property
    def do_classifier_free_guidance(self):
        return self._guidance_scale > 1.0

    @property
    def num_timesteps(self):
        return self._num_timesteps

    @property
    def current_timestep(self):
        return self._current_timestep

    @property
    def interrupt(self):
        return self._interrupt

    @property
    def attention_kwargs(self):
        return self._attention_kwargs

    def __call__(
        self,
        prompt: Union[str, List[str]] = None,
        negative_prompt: Union[str, List[str]] = None,
        video: Optional[List[PipelineImageInput]] = None,
        mask: Optional[List[PipelineImageInput]] = None,
        reference_images: Optional[List[PipelineImageInput]] = None,
        conditioning_scale: Union[float, List[float], ms.Tensor] = 1.0,
        height: int = 480,
        width: int = 832,
        num_frames: int = 81,
        num_inference_steps: int = 50,
        guidance_scale: float = 5.0,
        num_videos_per_prompt: Optional[int] = 1,
        generator: np.random.Generator = None,
        latents: Optional[ms.Tensor] = None,
        prompt_embeds: Optional[ms.Tensor] = None,
        negative_prompt_embeds: Optional[ms.Tensor] = None,
        output_type: Optional[str] = "np",
        return_dict: bool = False,
        attention_kwargs: Optional[Dict[str, Any]] = None,
        callback_on_step_end: Optional[
            Union[Callable[[int, int, Dict], None], PipelineCallback, MultiPipelineCallbacks]
        ] = None,
        callback_on_step_end_tensor_inputs: List[str] = ["latents"],
        max_sequence_length: int = 512,
    ):
        r"""
        The call function to the pipeline for generation.

        Args:
            prompt (`str` or `List[str]`, *optional*):
                The prompt or prompts to guide the image generation. If not defined, one has to pass `prompt_embeds`
                instead.
            negative_prompt (`str` or `List[str]`, *optional*):
                The prompt or prompts not to guide the image generation. If not defined, one has to pass
                `negative_prompt_embeds` instead. Ignored when not using guidance (i.e., ignored if `guidance_scale` is
                less than `1`).
            video (`List[PIL.Image.Image]`, *optional*):
                The input video or videos to be used as a starting point for the generation. The video should be a list
                of PIL images, a numpy array, or a mindspore tensor. Currently, the pipeline only supports generating one
                video at a time.
            mask (`List[PIL.Image.Image]`, *optional*):
                The input mask defines which video regions to condition on and which to generate. Black areas in the
                mask indicate conditioning regions, while white areas indicate regions for generation. The mask should
                be a list of PIL images, a numpy array, or a mindspore tensor. Currently supports generating a single video
                at a time.
            reference_images (`List[PIL.Image.Image]`, *optional*):
                A list of one or more reference images as extra conditioning for the generation. For example, if you
                are trying to inpaint a video to change the character, you can pass reference images of the new
                character here. Refer to the Diffusers [examples](https://github.com/huggingface/diffusers/pull/11582)
                and original [user
                guide](https://github.com/ali-vilab/VACE/blob/0897c6d055d7d9ea9e191dce763006664d9780f8/UserGuide.md)
                for a full list of supported tasks and use cases.
            conditioning_scale (`float`, `List[float]`, `ms.Tensor`, defaults to `1.0`):
                The conditioning scale to be applied when adding the control conditioning latent stream to the
                denoising latent stream in each control layer of the model. If a float is provided, it will be applied
                uniformly to all layers. If a list or tensor is provided, it should have the same length as the number
                of control layers in the model (`len(transformer.config.vace_layers)`).
            height (`int`, defaults to `480`):
                The height in pixels of the generated image.
            width (`int`, defaults to `832`):
                The width in pixels of the generated image.
            num_frames (`int`, defaults to `81`):
                The number of frames in the generated video.
            num_inference_steps (`int`, defaults to `50`):
                The number of denoising steps. More denoising steps usually lead to a higher quality image at the
                expense of slower inference.
            guidance_scale (`float`, defaults to `5.0`):
                Guidance scale as defined in [Classifier-Free Diffusion Guidance](https://arxiv.org/abs/2207.12598).
                `guidance_scale` is defined as `w` of equation 2. of [Imagen
                Paper](https://arxiv.org/pdf/2205.11487.pdf). Guidance scale is enabled by setting `guidance_scale >
                1`. Higher guidance scale encourages to generate images that are closely linked to the text `prompt`,
                usually at the expense of lower image quality.
            num_videos_per_prompt (`int`, *optional*, defaults to 1):
                The number of images to generate per prompt.
            generator (`np.random.Generator`):
                A [`np.random.Generator`] to make
                generation deterministic.
            latents (`ms.Tensor`, *optional*):
                Pre-generated noisy latents sampled from a Gaussian distribution, to be used as inputs for image
                generation. Can be used to tweak the same generation with different prompts. If not provided, a latents
                tensor is generated by sampling using the supplied random `generator`.
            prompt_embeds (`ms.Tensor`, *optional*):
                Pre-generated text embeddings. Can be used to easily tweak text inputs (prompt weighting). If not
                provided, text embeddings are generated from the `prompt` input argument.
            output_type (`str`, *optional*, defaults to `"np"`):
                The output format of the generated image. Choose between `PIL.Image` or `np.array`.
            return_dict (`bool`, *optional*, defaults to `False`):
                Whether or not to return a [`WanPipelineOutput`] instead of a plain tuple.
            attention_kwargs (`dict`, *optional*):
                A kwargs dictionary that if specified is passed along to the `AttentionProcessor` as defined under
                `self.processor` in
                [diffusers.models.attention_processor](https://github.com/huggingface/diffusers/blob/main/src/diffusers/models/attention_processor.py).
            callback_on_step_end (`Callable`, `PipelineCallback`, `MultiPipelineCallbacks`, *optional*):
                A function or a subclass of `PipelineCallback` or `MultiPipelineCallbacks` that is called at the end of
                each denoising step during the inference. with the following arguments: `callback_on_step_end(self:
                DiffusionPipeline, step: int, timestep: int, callback_kwargs: Dict)`. `callback_kwargs` will include a
                list of all tensors as specified by `callback_on_step_end_tensor_inputs`.
            callback_on_step_end_tensor_inputs (`List`, *optional*):
                The list of tensor inputs for the `callback_on_step_end` function. The tensors specified in the list
                will be passed as `callback_kwargs` argument. You will only be able to include variables listed in the
                `._callback_tensor_inputs` attribute of your pipeline class.
            max_sequence_length (`int`, defaults to `512`):
                The maximum sequence length of the text encoder. If the prompt is longer than this, it will be
                truncated. If the prompt is shorter, it will be padded to this length.

        Examples:

        Returns:
            [`~WanPipelineOutput`] or `tuple`:
                If `return_dict` is `False`, [`WanPipelineOutput`] is returned, otherwise a `tuple` is returned where
                the first element is a list with the generated images and the second element is a list of `bool`s
                indicating whether the corresponding generated image contains "not-safe-for-work" (nsfw) content.
        """

        if isinstance(callback_on_step_end, (PipelineCallback, MultiPipelineCallbacks)):
            callback_on_step_end_tensor_inputs = callback_on_step_end.tensor_inputs

        # Simplification of implementation for now
        if not isinstance(prompt, str):
            raise ValueError("Passing a list of prompts is not yet supported. This may be supported in the future.")
        if num_videos_per_prompt != 1:
            raise ValueError(
                "Generating multiple videos per prompt is not yet supported. This may be supported in the future."
            )

        # 1. Check inputs. Raise error if not correct
        self.check_inputs(
            prompt,
            negative_prompt,
            height,
            width,
            prompt_embeds,
            negative_prompt_embeds,
            callback_on_step_end_tensor_inputs,
            video,
            mask,
            reference_images,
        )

        if num_frames % self.vae_scale_factor_temporal != 1:
            logger.warning(
                f"`num_frames - 1` has to be divisible by {self.vae_scale_factor_temporal}. "
                f"Rounding to the nearest number."
            )
            num_frames = num_frames // self.vae_scale_factor_temporal * self.vae_scale_factor_temporal + 1
        num_frames = max(num_frames, 1)

        self._guidance_scale = guidance_scale
        self._attention_kwargs = attention_kwargs
        self._current_timestep = None
        self._interrupt = False

        # 2. Define call parameters
        if prompt is not None and isinstance(prompt, str):
            batch_size = 1
        elif prompt is not None and isinstance(prompt, list):
            batch_size = len(prompt)
        else:
            batch_size = prompt_embeds.shape[0]

        vae_dtype = self.vae.dtype
        transformer_dtype = self.transformer.dtype

        if isinstance(conditioning_scale, (int, float)):
            conditioning_scale = [conditioning_scale] * len(self.transformer.config.vace_layers)
        if isinstance(conditioning_scale, list):
            if len(conditioning_scale) != len(self.transformer.config.vace_layers):
                raise ValueError(
                    f"Length of `conditioning_scale` {len(conditioning_scale)} does not match number of layers "
                    f"{len(self.transformer.config.vace_layers)}."
                )
            conditioning_scale = ms.tensor(conditioning_scale)
        if isinstance(conditioning_scale, ms.Tensor):
            if conditioning_scale.shape[0] != len(self.transformer.config.vace_layers):
                raise ValueError(
                    f"Length of `conditioning_scale` {conditioning_scale.shape[0]} does not match number of layers "
                    f"{len(self.transformer.config.vace_layers)}."
                )
            conditioning_scale = conditioning_scale.to(dtype=transformer_dtype)

        # 3. Encode input prompt
        prompt_embeds, negative_prompt_embeds = self.encode_prompt(
            prompt=prompt,
            negative_prompt=negative_prompt,
            do_classifier_free_guidance=self.do_classifier_free_guidance,
            num_videos_per_prompt=num_videos_per_prompt,
            prompt_embeds=prompt_embeds,
            negative_prompt_embeds=negative_prompt_embeds,
            max_sequence_length=max_sequence_length,
        )

        prompt_embeds = prompt_embeds.to(transformer_dtype)
        if negative_prompt_embeds is not None:
            negative_prompt_embeds = negative_prompt_embeds.to(transformer_dtype)

        # 4. Prepare timesteps
        self.scheduler.set_timesteps(num_inference_steps)
        timesteps = self.scheduler.timesteps

        # 5. Prepare latent variables
        video, mask, reference_images = self.preprocess_conditions(
            video,
            mask,
            reference_images,
            batch_size,
            height,
            width,
            num_frames,
            ms.float32,
        )
        num_reference_images = len(reference_images[0])

        conditioning_latents = self.prepare_video_latents(video, mask, reference_images, generator)
        mask = self.prepare_masks(mask, reference_images, generator).to(conditioning_latents.dtype)
        conditioning_latents = mint.cat([conditioning_latents, mask], dim=1)
        conditioning_latents = conditioning_latents.to(transformer_dtype)

        num_channels_latents = self.transformer.config.in_channels
        latents = self.prepare_latents(
            batch_size * num_videos_per_prompt,
            num_channels_latents,
            height,
            width,
            num_frames + num_reference_images * self.vae_scale_factor_temporal,
            ms.float32,
            generator,
            latents,
        )

        if conditioning_latents.shape[2] != latents.shape[2]:
            logger.warning(
                "The number of frames in the conditioning latents does not match the number of frames to be generated. "
                "Generation quality may be affected."
            )

        # 6. Denoising loop
        num_warmup_steps = len(timesteps) - num_inference_steps * self.scheduler.order
        self._num_timesteps = len(timesteps)

        with self.progress_bar(total=num_inference_steps) as progress_bar:
            for i, t in enumerate(timesteps):
                if self.interrupt:
                    continue

                self._current_timestep = t
                latent_model_input = latents.to(transformer_dtype)
                timestep = t.expand([latents.shape[0]])

                noise_pred = self.transformer(
                    hidden_states=latent_model_input,
                    timestep=timestep,
                    encoder_hidden_states=prompt_embeds,
                    control_hidden_states=conditioning_latents,
                    control_hidden_states_scale=conditioning_scale,
                    attention_kwargs=attention_kwargs,
                    return_dict=False,
                )[0]

                if self.do_classifier_free_guidance:
                    noise_uncond = self.transformer(
                        hidden_states=latent_model_input,
                        timestep=timestep,
                        encoder_hidden_states=negative_prompt_embeds,
                        control_hidden_states=conditioning_latents,
                        control_hidden_states_scale=conditioning_scale,
                        attention_kwargs=attention_kwargs,
                        return_dict=False,
                    )[0]
                    noise_pred = noise_uncond + guidance_scale * (noise_pred - noise_uncond)

                # compute the previous noisy sample x_t -> x_t-1
                latents = self.scheduler.step(noise_pred, t, latents, return_dict=False)[0]

                if callback_on_step_end is not None:
                    callback_kwargs = {}
                    for k in callback_on_step_end_tensor_inputs:
                        callback_kwargs[k] = locals()[k]
                    callback_outputs = callback_on_step_end(self, i, t, callback_kwargs)

                    latents = callback_outputs.pop("latents", latents)
                    prompt_embeds = callback_outputs.pop("prompt_embeds", prompt_embeds)
                    negative_prompt_embeds = callback_outputs.pop("negative_prompt_embeds", negative_prompt_embeds)

                # call the callback, if provided
                if i == len(timesteps) - 1 or ((i + 1) > num_warmup_steps and (i + 1) % self.scheduler.order == 0):
                    progress_bar.update()

        self._current_timestep = None

        if not output_type == "latent":
            latents = latents[:, :, num_reference_images:]
            latents = latents.to(vae_dtype)
            latents_mean = (
                ms.tensor(self.vae.config.latents_mean).view(1, self.vae.config.z_dim, 1, 1, 1).to(latents.dtype)
            )
            latents_std = 1.0 / ms.tensor(self.vae.config.latents_std, dtype=ms.float32).view(
                1, self.vae.config.z_dim, 1, 1, 1
            )
            latents = latents / latents_std + latents_mean
            with pynative_context():
                video = self.vae.decode(latents, return_dict=False)[0]
            video = self.video_processor.postprocess_video(video, output_type=output_type)
        else:
            video = latents

        if not return_dict:
            return (video,)

        return WanPipelineOutput(frames=video)

mindone.diffusers.WanVACEPipeline.__call__(prompt=None, negative_prompt=None, video=None, mask=None, reference_images=None, conditioning_scale=1.0, height=480, width=832, num_frames=81, num_inference_steps=50, guidance_scale=5.0, num_videos_per_prompt=1, generator=None, latents=None, prompt_embeds=None, negative_prompt_embeds=None, output_type='np', return_dict=False, attention_kwargs=None, callback_on_step_end=None, callback_on_step_end_tensor_inputs=['latents'], max_sequence_length=512)

The call function to the pipeline for generation.

PARAMETER	DESCRIPTION
prompt	The prompt or prompts to guide the image generation. If not defined, one has to pass prompt_embeds instead. TYPE: `str` or `List[str]`, *optional* DEFAULT: None
negative_prompt	The prompt or prompts not to guide the image generation. If not defined, one has to pass negative_prompt_embeds instead. Ignored when not using guidance (i.e., ignored if guidance_scale is less than 1). TYPE: `str` or `List[str]`, *optional* DEFAULT: None
video	The input video or videos to be used as a starting point for the generation. The video should be a list of PIL images, a numpy array, or a mindspore tensor. Currently, the pipeline only supports generating one video at a time. TYPE: `List[PIL.Image.Image]`, *optional* DEFAULT: None
mask	The input mask defines which video regions to condition on and which to generate. Black areas in the mask indicate conditioning regions, while white areas indicate regions for generation. The mask should be a list of PIL images, a numpy array, or a mindspore tensor. Currently supports generating a single video at a time. TYPE: `List[PIL.Image.Image]`, *optional* DEFAULT: None
reference_images	A list of one or more reference images as extra conditioning for the generation. For example, if you are trying to inpaint a video to change the character, you can pass reference images of the new character here. Refer to the Diffusers examples and original user guide for a full list of supported tasks and use cases. TYPE: `List[PIL.Image.Image]`, *optional* DEFAULT: None
conditioning_scale	The conditioning scale to be applied when adding the control conditioning latent stream to the denoising latent stream in each control layer of the model. If a float is provided, it will be applied uniformly to all layers. If a list or tensor is provided, it should have the same length as the number of control layers in the model (len(transformer.config.vace_layers)). TYPE: `float`, `List[float]`, `ms.Tensor`, defaults to `1.0` DEFAULT: 1.0
height	The height in pixels of the generated image. TYPE: `int`, defaults to `480` DEFAULT: 480
width	The width in pixels of the generated image. TYPE: `int`, defaults to `832` DEFAULT: 832
num_frames	The number of frames in the generated video. TYPE: `int`, defaults to `81` DEFAULT: 81
num_inference_steps	The number of denoising steps. More denoising steps usually lead to a higher quality image at the expense of slower inference. TYPE: `int`, defaults to `50` DEFAULT: 50
guidance_scale	Guidance scale as defined in Classifier-Free Diffusion Guidance. guidance_scale is defined as w of equation 2. of Imagen Paper. Guidance scale is enabled by setting guidance_scale > 1. Higher guidance scale encourages to generate images that are closely linked to the text prompt, usually at the expense of lower image quality. TYPE: `float`, defaults to `5.0` DEFAULT: 5.0
num_videos_per_prompt	The number of images to generate per prompt. TYPE: `int`, *optional*, defaults to 1 DEFAULT: 1
generator	A [np.random.Generator] to make generation deterministic. TYPE: `np.random.Generator` DEFAULT: None
latents	Pre-generated noisy latents sampled from a Gaussian distribution, to be used as inputs for image generation. Can be used to tweak the same generation with different prompts. If not provided, a latents tensor is generated by sampling using the supplied random generator. TYPE: `ms.Tensor`, *optional* DEFAULT: None
prompt_embeds	Pre-generated text embeddings. Can be used to easily tweak text inputs (prompt weighting). If not provided, text embeddings are generated from the prompt input argument. TYPE: `ms.Tensor`, *optional* DEFAULT: None
output_type	The output format of the generated image. Choose between PIL.Image or np.array. TYPE: `str`, *optional*, defaults to `"np"` DEFAULT: 'np'
return_dict	Whether or not to return a [WanPipelineOutput] instead of a plain tuple. TYPE: `bool`, *optional*, defaults to `False` DEFAULT: False
attention_kwargs	A kwargs dictionary that if specified is passed along to the AttentionProcessor as defined under self.processor in diffusers.models.attention_processor. TYPE: `dict`, *optional* DEFAULT: None
callback_on_step_end	A function or a subclass of PipelineCallback or MultiPipelineCallbacks that is called at the end of each denoising step during the inference. with the following arguments: callback_on_step_end(self: DiffusionPipeline, step: int, timestep: int, callback_kwargs: Dict). callback_kwargs will include a list of all tensors as specified by callback_on_step_end_tensor_inputs. TYPE: `Callable`, `PipelineCallback`, `MultiPipelineCallbacks`, *optional* DEFAULT: None
callback_on_step_end_tensor_inputs	The list of tensor inputs for the callback_on_step_end function. The tensors specified in the list will be passed as callback_kwargs argument. You will only be able to include variables listed in the ._callback_tensor_inputs attribute of your pipeline class. TYPE: `List`, *optional* DEFAULT: ['latents']
max_sequence_length	The maximum sequence length of the text encoder. If the prompt is longer than this, it will be truncated. If the prompt is shorter, it will be padded to this length. TYPE: `int`, defaults to `512` DEFAULT: 512

RETURNS	DESCRIPTION
	[~WanPipelineOutput] or tuple: If return_dict is False, [WanPipelineOutput] is returned, otherwise a tuple is returned where the first element is a list with the generated images and the second element is a list of bools indicating whether the corresponding generated image contains "not-safe-for-work" (nsfw) content.

Source code in mindone/diffusers/pipelines/wan/pipeline_wan_vace.py

def __call__(
    self,
    prompt: Union[str, List[str]] = None,
    negative_prompt: Union[str, List[str]] = None,
    video: Optional[List[PipelineImageInput]] = None,
    mask: Optional[List[PipelineImageInput]] = None,
    reference_images: Optional[List[PipelineImageInput]] = None,
    conditioning_scale: Union[float, List[float], ms.Tensor] = 1.0,
    height: int = 480,
    width: int = 832,
    num_frames: int = 81,
    num_inference_steps: int = 50,
    guidance_scale: float = 5.0,
    num_videos_per_prompt: Optional[int] = 1,
    generator: np.random.Generator = None,
    latents: Optional[ms.Tensor] = None,
    prompt_embeds: Optional[ms.Tensor] = None,
    negative_prompt_embeds: Optional[ms.Tensor] = None,
    output_type: Optional[str] = "np",
    return_dict: bool = False,
    attention_kwargs: Optional[Dict[str, Any]] = None,
    callback_on_step_end: Optional[
        Union[Callable[[int, int, Dict], None], PipelineCallback, MultiPipelineCallbacks]
    ] = None,
    callback_on_step_end_tensor_inputs: List[str] = ["latents"],
    max_sequence_length: int = 512,
):
    r"""
    The call function to the pipeline for generation.

    Args:
        prompt (`str` or `List[str]`, *optional*):
            The prompt or prompts to guide the image generation. If not defined, one has to pass `prompt_embeds`
            instead.
        negative_prompt (`str` or `List[str]`, *optional*):
            The prompt or prompts not to guide the image generation. If not defined, one has to pass
            `negative_prompt_embeds` instead. Ignored when not using guidance (i.e., ignored if `guidance_scale` is
            less than `1`).
        video (`List[PIL.Image.Image]`, *optional*):
            The input video or videos to be used as a starting point for the generation. The video should be a list
            of PIL images, a numpy array, or a mindspore tensor. Currently, the pipeline only supports generating one
            video at a time.
        mask (`List[PIL.Image.Image]`, *optional*):
            The input mask defines which video regions to condition on and which to generate. Black areas in the
            mask indicate conditioning regions, while white areas indicate regions for generation. The mask should
            be a list of PIL images, a numpy array, or a mindspore tensor. Currently supports generating a single video
            at a time.
        reference_images (`List[PIL.Image.Image]`, *optional*):
            A list of one or more reference images as extra conditioning for the generation. For example, if you
            are trying to inpaint a video to change the character, you can pass reference images of the new
            character here. Refer to the Diffusers [examples](https://github.com/huggingface/diffusers/pull/11582)
            and original [user
            guide](https://github.com/ali-vilab/VACE/blob/0897c6d055d7d9ea9e191dce763006664d9780f8/UserGuide.md)
            for a full list of supported tasks and use cases.
        conditioning_scale (`float`, `List[float]`, `ms.Tensor`, defaults to `1.0`):
            The conditioning scale to be applied when adding the control conditioning latent stream to the
            denoising latent stream in each control layer of the model. If a float is provided, it will be applied
            uniformly to all layers. If a list or tensor is provided, it should have the same length as the number
            of control layers in the model (`len(transformer.config.vace_layers)`).
        height (`int`, defaults to `480`):
            The height in pixels of the generated image.
        width (`int`, defaults to `832`):
            The width in pixels of the generated image.
        num_frames (`int`, defaults to `81`):
            The number of frames in the generated video.
        num_inference_steps (`int`, defaults to `50`):
            The number of denoising steps. More denoising steps usually lead to a higher quality image at the
            expense of slower inference.
        guidance_scale (`float`, defaults to `5.0`):
            Guidance scale as defined in [Classifier-Free Diffusion Guidance](https://arxiv.org/abs/2207.12598).
            `guidance_scale` is defined as `w` of equation 2. of [Imagen
            Paper](https://arxiv.org/pdf/2205.11487.pdf). Guidance scale is enabled by setting `guidance_scale >
            1`. Higher guidance scale encourages to generate images that are closely linked to the text `prompt`,
            usually at the expense of lower image quality.
        num_videos_per_prompt (`int`, *optional*, defaults to 1):
            The number of images to generate per prompt.
        generator (`np.random.Generator`):
            A [`np.random.Generator`] to make
            generation deterministic.
        latents (`ms.Tensor`, *optional*):
            Pre-generated noisy latents sampled from a Gaussian distribution, to be used as inputs for image
            generation. Can be used to tweak the same generation with different prompts. If not provided, a latents
            tensor is generated by sampling using the supplied random `generator`.
        prompt_embeds (`ms.Tensor`, *optional*):
            Pre-generated text embeddings. Can be used to easily tweak text inputs (prompt weighting). If not
            provided, text embeddings are generated from the `prompt` input argument.
        output_type (`str`, *optional*, defaults to `"np"`):
            The output format of the generated image. Choose between `PIL.Image` or `np.array`.
        return_dict (`bool`, *optional*, defaults to `False`):
            Whether or not to return a [`WanPipelineOutput`] instead of a plain tuple.
        attention_kwargs (`dict`, *optional*):
            A kwargs dictionary that if specified is passed along to the `AttentionProcessor` as defined under
            `self.processor` in
            [diffusers.models.attention_processor](https://github.com/huggingface/diffusers/blob/main/src/diffusers/models/attention_processor.py).
        callback_on_step_end (`Callable`, `PipelineCallback`, `MultiPipelineCallbacks`, *optional*):
            A function or a subclass of `PipelineCallback` or `MultiPipelineCallbacks` that is called at the end of
            each denoising step during the inference. with the following arguments: `callback_on_step_end(self:
            DiffusionPipeline, step: int, timestep: int, callback_kwargs: Dict)`. `callback_kwargs` will include a
            list of all tensors as specified by `callback_on_step_end_tensor_inputs`.
        callback_on_step_end_tensor_inputs (`List`, *optional*):
            The list of tensor inputs for the `callback_on_step_end` function. The tensors specified in the list
            will be passed as `callback_kwargs` argument. You will only be able to include variables listed in the
            `._callback_tensor_inputs` attribute of your pipeline class.
        max_sequence_length (`int`, defaults to `512`):
            The maximum sequence length of the text encoder. If the prompt is longer than this, it will be
            truncated. If the prompt is shorter, it will be padded to this length.

    Examples:

    Returns:
        [`~WanPipelineOutput`] or `tuple`:
            If `return_dict` is `False`, [`WanPipelineOutput`] is returned, otherwise a `tuple` is returned where
            the first element is a list with the generated images and the second element is a list of `bool`s
            indicating whether the corresponding generated image contains "not-safe-for-work" (nsfw) content.
    """

    if isinstance(callback_on_step_end, (PipelineCallback, MultiPipelineCallbacks)):
        callback_on_step_end_tensor_inputs = callback_on_step_end.tensor_inputs

    # Simplification of implementation for now
    if not isinstance(prompt, str):
        raise ValueError("Passing a list of prompts is not yet supported. This may be supported in the future.")
    if num_videos_per_prompt != 1:
        raise ValueError(
            "Generating multiple videos per prompt is not yet supported. This may be supported in the future."
        )

    # 1. Check inputs. Raise error if not correct
    self.check_inputs(
        prompt,
        negative_prompt,
        height,
        width,
        prompt_embeds,
        negative_prompt_embeds,
        callback_on_step_end_tensor_inputs,
        video,
        mask,
        reference_images,
    )

    if num_frames % self.vae_scale_factor_temporal != 1:
        logger.warning(
            f"`num_frames - 1` has to be divisible by {self.vae_scale_factor_temporal}. "
            f"Rounding to the nearest number."
        )
        num_frames = num_frames // self.vae_scale_factor_temporal * self.vae_scale_factor_temporal + 1
    num_frames = max(num_frames, 1)

    self._guidance_scale = guidance_scale
    self._attention_kwargs = attention_kwargs
    self._current_timestep = None
    self._interrupt = False

    # 2. Define call parameters
    if prompt is not None and isinstance(prompt, str):
        batch_size = 1
    elif prompt is not None and isinstance(prompt, list):
        batch_size = len(prompt)
    else:
        batch_size = prompt_embeds.shape[0]

    vae_dtype = self.vae.dtype
    transformer_dtype = self.transformer.dtype

    if isinstance(conditioning_scale, (int, float)):
        conditioning_scale = [conditioning_scale] * len(self.transformer.config.vace_layers)
    if isinstance(conditioning_scale, list):
        if len(conditioning_scale) != len(self.transformer.config.vace_layers):
            raise ValueError(
                f"Length of `conditioning_scale` {len(conditioning_scale)} does not match number of layers "
                f"{len(self.transformer.config.vace_layers)}."
            )
        conditioning_scale = ms.tensor(conditioning_scale)
    if isinstance(conditioning_scale, ms.Tensor):
        if conditioning_scale.shape[0] != len(self.transformer.config.vace_layers):
            raise ValueError(
                f"Length of `conditioning_scale` {conditioning_scale.shape[0]} does not match number of layers "
                f"{len(self.transformer.config.vace_layers)}."
            )
        conditioning_scale = conditioning_scale.to(dtype=transformer_dtype)

    # 3. Encode input prompt
    prompt_embeds, negative_prompt_embeds = self.encode_prompt(
        prompt=prompt,
        negative_prompt=negative_prompt,
        do_classifier_free_guidance=self.do_classifier_free_guidance,
        num_videos_per_prompt=num_videos_per_prompt,
        prompt_embeds=prompt_embeds,
        negative_prompt_embeds=negative_prompt_embeds,
        max_sequence_length=max_sequence_length,
    )

    prompt_embeds = prompt_embeds.to(transformer_dtype)
    if negative_prompt_embeds is not None:
        negative_prompt_embeds = negative_prompt_embeds.to(transformer_dtype)

    # 4. Prepare timesteps
    self.scheduler.set_timesteps(num_inference_steps)
    timesteps = self.scheduler.timesteps

    # 5. Prepare latent variables
    video, mask, reference_images = self.preprocess_conditions(
        video,
        mask,
        reference_images,
        batch_size,
        height,
        width,
        num_frames,
        ms.float32,
    )
    num_reference_images = len(reference_images[0])

    conditioning_latents = self.prepare_video_latents(video, mask, reference_images, generator)
    mask = self.prepare_masks(mask, reference_images, generator).to(conditioning_latents.dtype)
    conditioning_latents = mint.cat([conditioning_latents, mask], dim=1)
    conditioning_latents = conditioning_latents.to(transformer_dtype)

    num_channels_latents = self.transformer.config.in_channels
    latents = self.prepare_latents(
        batch_size * num_videos_per_prompt,
        num_channels_latents,
        height,
        width,
        num_frames + num_reference_images * self.vae_scale_factor_temporal,
        ms.float32,
        generator,
        latents,
    )

    if conditioning_latents.shape[2] != latents.shape[2]:
        logger.warning(
            "The number of frames in the conditioning latents does not match the number of frames to be generated. "
            "Generation quality may be affected."
        )

    # 6. Denoising loop
    num_warmup_steps = len(timesteps) - num_inference_steps * self.scheduler.order
    self._num_timesteps = len(timesteps)

    with self.progress_bar(total=num_inference_steps) as progress_bar:
        for i, t in enumerate(timesteps):
            if self.interrupt:
                continue

            self._current_timestep = t
            latent_model_input = latents.to(transformer_dtype)
            timestep = t.expand([latents.shape[0]])

            noise_pred = self.transformer(
                hidden_states=latent_model_input,
                timestep=timestep,
                encoder_hidden_states=prompt_embeds,
                control_hidden_states=conditioning_latents,
                control_hidden_states_scale=conditioning_scale,
                attention_kwargs=attention_kwargs,
                return_dict=False,
            )[0]

            if self.do_classifier_free_guidance:
                noise_uncond = self.transformer(
                    hidden_states=latent_model_input,
                    timestep=timestep,
                    encoder_hidden_states=negative_prompt_embeds,
                    control_hidden_states=conditioning_latents,
                    control_hidden_states_scale=conditioning_scale,
                    attention_kwargs=attention_kwargs,
                    return_dict=False,
                )[0]
                noise_pred = noise_uncond + guidance_scale * (noise_pred - noise_uncond)

            # compute the previous noisy sample x_t -> x_t-1
            latents = self.scheduler.step(noise_pred, t, latents, return_dict=False)[0]

            if callback_on_step_end is not None:
                callback_kwargs = {}
                for k in callback_on_step_end_tensor_inputs:
                    callback_kwargs[k] = locals()[k]
                callback_outputs = callback_on_step_end(self, i, t, callback_kwargs)

                latents = callback_outputs.pop("latents", latents)
                prompt_embeds = callback_outputs.pop("prompt_embeds", prompt_embeds)
                negative_prompt_embeds = callback_outputs.pop("negative_prompt_embeds", negative_prompt_embeds)

            # call the callback, if provided
            if i == len(timesteps) - 1 or ((i + 1) > num_warmup_steps and (i + 1) % self.scheduler.order == 0):
                progress_bar.update()

    self._current_timestep = None

    if not output_type == "latent":
        latents = latents[:, :, num_reference_images:]
        latents = latents.to(vae_dtype)
        latents_mean = (
            ms.tensor(self.vae.config.latents_mean).view(1, self.vae.config.z_dim, 1, 1, 1).to(latents.dtype)
        )
        latents_std = 1.0 / ms.tensor(self.vae.config.latents_std, dtype=ms.float32).view(
            1, self.vae.config.z_dim, 1, 1, 1
        )
        latents = latents / latents_std + latents_mean
        with pynative_context():
            video = self.vae.decode(latents, return_dict=False)[0]
        video = self.video_processor.postprocess_video(video, output_type=output_type)
    else:
        video = latents

    if not return_dict:
        return (video,)

    return WanPipelineOutput(frames=video)

mindone.diffusers.WanVACEPipeline.encode_prompt(prompt, negative_prompt=None, do_classifier_free_guidance=True, num_videos_per_prompt=1, prompt_embeds=None, negative_prompt_embeds=None, max_sequence_length=226, dtype=None)

Encodes the prompt into text encoder hidden states.

PARAMETER	DESCRIPTION
prompt	prompt to be encoded TYPE: `str` or `List[str]`, *optional*
negative_prompt	The prompt or prompts not to guide the image generation. If not defined, one has to pass negative_prompt_embeds instead. Ignored when not using guidance (i.e., ignored if guidance_scale is less than 1). TYPE: `str` or `List[str]`, *optional* DEFAULT: None
do_classifier_free_guidance	Whether to use classifier free guidance or not. TYPE: `bool`, *optional*, defaults to `True` DEFAULT: True
num_videos_per_prompt	Number of videos that should be generated per prompt. TYPE: `int`, *optional*, defaults to 1 DEFAULT: 1
prompt_embeds	Pre-generated text embeddings. Can be used to easily tweak text inputs, e.g. prompt weighting. If not provided, text embeddings will be generated from prompt input argument. TYPE: `ms.Tensor`, *optional* DEFAULT: None
negative_prompt_embeds	Pre-generated negative text embeddings. Can be used to easily tweak text inputs, e.g. prompt weighting. If not provided, negative_prompt_embeds will be generated from negative_prompt input argument. TYPE: `ms.Tensor`, *optional* DEFAULT: None
dtype	(ms.dtype, optional): ms dtype TYPE: Optional[Type] DEFAULT: None

Source code in mindone/diffusers/pipelines/wan/pipeline_wan_vace.py

def encode_prompt(
    self,
    prompt: Union[str, List[str]],
    negative_prompt: Optional[Union[str, List[str]]] = None,
    do_classifier_free_guidance: bool = True,
    num_videos_per_prompt: int = 1,
    prompt_embeds: Optional[ms.Tensor] = None,
    negative_prompt_embeds: Optional[ms.Tensor] = None,
    max_sequence_length: int = 226,
    dtype: Optional[ms.Type] = None,
):
    r"""
    Encodes the prompt into text encoder hidden states.

    Args:
        prompt (`str` or `List[str]`, *optional*):
            prompt to be encoded
        negative_prompt (`str` or `List[str]`, *optional*):
            The prompt or prompts not to guide the image generation. If not defined, one has to pass
            `negative_prompt_embeds` instead. Ignored when not using guidance (i.e., ignored if `guidance_scale` is
            less than `1`).
        do_classifier_free_guidance (`bool`, *optional*, defaults to `True`):
            Whether to use classifier free guidance or not.
        num_videos_per_prompt (`int`, *optional*, defaults to 1):
            Number of videos that should be generated per prompt.
        prompt_embeds (`ms.Tensor`, *optional*):
            Pre-generated text embeddings. Can be used to easily tweak text inputs, *e.g.* prompt weighting. If not
            provided, text embeddings will be generated from `prompt` input argument.
        negative_prompt_embeds (`ms.Tensor`, *optional*):
            Pre-generated negative text embeddings. Can be used to easily tweak text inputs, *e.g.* prompt
            weighting. If not provided, negative_prompt_embeds will be generated from `negative_prompt` input
            argument.
        dtype: (`ms.dtype`, *optional*):
            ms dtype
    """
    prompt = [prompt] if isinstance(prompt, str) else prompt
    if prompt is not None:
        batch_size = len(prompt)
    else:
        batch_size = prompt_embeds.shape[0]

    if prompt_embeds is None:
        prompt_embeds = self._get_t5_prompt_embeds(
            prompt=prompt,
            num_videos_per_prompt=num_videos_per_prompt,
            max_sequence_length=max_sequence_length,
            dtype=dtype,
        )

    if do_classifier_free_guidance and negative_prompt_embeds is None:
        negative_prompt = negative_prompt or ""
        negative_prompt = batch_size * [negative_prompt] if isinstance(negative_prompt, str) else negative_prompt

        if prompt is not None and type(prompt) is not type(negative_prompt):
            raise TypeError(
                f"`negative_prompt` should be the same type to `prompt`, but got {type(negative_prompt)} !="
                f" {type(prompt)}."
            )
        elif batch_size != len(negative_prompt):
            raise ValueError(
                f"`negative_prompt`: {negative_prompt} has batch size {len(negative_prompt)}, but `prompt`:"
                f" {prompt} has batch size {batch_size}. Please make sure that passed `negative_prompt` matches"
                " the batch size of `prompt`."
            )

        negative_prompt_embeds = self._get_t5_prompt_embeds(
            prompt=negative_prompt,
            num_videos_per_prompt=num_videos_per_prompt,
            max_sequence_length=max_sequence_length,
            dtype=dtype,
        )

    return prompt_embeds, negative_prompt_embeds

mindone.diffusers.WanVideoToVideoPipeline

Bases: DiffusionPipeline, WanLoraLoaderMixin

Pipeline for video-to-video generation using Wan.

This model inherits from [DiffusionPipeline]. Check the superclass documentation for the generic methods implemented for all pipelines (downloading, saving, running on a particular device, etc.).

PARAMETER	DESCRIPTION
tokenizer	Tokenizer from T5, specifically the google/umt5-xxl variant. TYPE: [`T5Tokenizer`]
text_encoder	T5, specifically the google/umt5-xxl variant. TYPE: [`T5EncoderModel`]
transformer	Conditional Transformer to denoise the input latents. TYPE: [`WanTransformer3DModel`]
scheduler	A scheduler to be used in combination with transformer to denoise the encoded image latents. TYPE: [`UniPCMultistepScheduler`]
vae	Variational Auto-Encoder (VAE) Model to encode and decode videos to and from latent representations. TYPE: [`AutoencoderKLWan`]

Source code in mindone/diffusers/pipelines/wan/pipeline_wan_video2video.py

class WanVideoToVideoPipeline(DiffusionPipeline, WanLoraLoaderMixin):
    r"""
    Pipeline for video-to-video generation using Wan.

    This model inherits from [`DiffusionPipeline`]. Check the superclass documentation for the generic methods
    implemented for all pipelines (downloading, saving, running on a particular device, etc.).

    Args:
        tokenizer ([`T5Tokenizer`]):
            Tokenizer from [T5](https://huggingface.co/docs/transformers/en/model_doc/t5#transformers.T5Tokenizer),
            specifically the [google/umt5-xxl](https://huggingface.co/google/umt5-xxl) variant.
        text_encoder ([`T5EncoderModel`]):
            [T5](https://huggingface.co/docs/transformers/en/model_doc/t5#transformers.T5EncoderModel), specifically
            the [google/umt5-xxl](https://huggingface.co/google/umt5-xxl) variant.
        transformer ([`WanTransformer3DModel`]):
            Conditional Transformer to denoise the input latents.
        scheduler ([`UniPCMultistepScheduler`]):
            A scheduler to be used in combination with `transformer` to denoise the encoded image latents.
        vae ([`AutoencoderKLWan`]):
            Variational Auto-Encoder (VAE) Model to encode and decode videos to and from latent representations.
    """

    model_cpu_offload_seq = "text_encoder->transformer->vae"
    _callback_tensor_inputs = ["latents", "prompt_embeds", "negative_prompt_embeds"]

    def __init__(
        self,
        tokenizer: AutoTokenizer,
        text_encoder: UMT5EncoderModel,
        transformer: WanTransformer3DModel,
        vae: AutoencoderKLWan,
        scheduler: FlowMatchEulerDiscreteScheduler,
    ):
        super().__init__()

        self.register_modules(
            vae=vae,
            text_encoder=text_encoder,
            tokenizer=tokenizer,
            transformer=transformer,
            scheduler=scheduler,
        )

        self.vae_scale_factor_temporal = 2 ** sum(self.vae.temperal_downsample) if getattr(self, "vae", None) else 4
        self.vae_scale_factor_spatial = 2 ** len(self.vae.temperal_downsample) if getattr(self, "vae", None) else 8
        self.video_processor = VideoProcessor(vae_scale_factor=self.vae_scale_factor_spatial)

    # Copied from diffusers.pipelines.wan.pipeline_wan.WanPipeline._get_t5_prompt_embeds
    def _get_t5_prompt_embeds(
        self,
        prompt: Union[str, List[str]] = None,
        num_videos_per_prompt: int = 1,
        max_sequence_length: int = 226,
        dtype: Optional[ms.Type] = None,
    ):
        dtype = dtype or self.text_encoder.dtype

        prompt = [prompt] if isinstance(prompt, str) else prompt
        prompt = [prompt_clean(u) for u in prompt]
        batch_size = len(prompt)

        text_inputs = self.tokenizer(
            prompt,
            padding="max_length",
            max_length=max_sequence_length,
            truncation=True,
            add_special_tokens=True,
            return_attention_mask=True,
            return_tensors="np",
        )
        text_input_ids, mask = ms.tensor(text_inputs.input_ids), ms.tensor(text_inputs.attention_mask)
        seq_lens = mask.gt(0).sum(dim=1).long()

        prompt_embeds = self.text_encoder(text_input_ids, mask)[0]
        prompt_embeds = prompt_embeds.to(dtype=dtype)
        prompt_embeds = [u[:v] for u, v in zip(prompt_embeds, seq_lens)]
        prompt_embeds = mint.stack(
            [mint.cat([u, u.new_zeros((max_sequence_length - u.shape[0], u.shape[1]))]) for u in prompt_embeds], dim=0
        )

        # duplicate text embeddings for each generation per prompt, using mps friendly method
        _, seq_len, _ = prompt_embeds.shape
        prompt_embeds = prompt_embeds.tile((1, num_videos_per_prompt, 1))
        prompt_embeds = prompt_embeds.view(batch_size * num_videos_per_prompt, seq_len, -1)

        return prompt_embeds

    # Copied from diffusers.pipelines.wan.pipeline_wan.WanPipeline.encode_prompt
    def encode_prompt(
        self,
        prompt: Union[str, List[str]],
        negative_prompt: Optional[Union[str, List[str]]] = None,
        do_classifier_free_guidance: bool = True,
        num_videos_per_prompt: int = 1,
        prompt_embeds: Optional[ms.Tensor] = None,
        negative_prompt_embeds: Optional[ms.Tensor] = None,
        max_sequence_length: int = 226,
        dtype: Optional[ms.Type] = None,
    ):
        r"""
        Encodes the prompt into text encoder hidden states.

        Args:
            prompt (`str` or `List[str]`, *optional*):
                prompt to be encoded
            negative_prompt (`str` or `List[str]`, *optional*):
                The prompt or prompts not to guide the image generation. If not defined, one has to pass
                `negative_prompt_embeds` instead. Ignored when not using guidance (i.e., ignored if `guidance_scale` is
                less than `1`).
            do_classifier_free_guidance (`bool`, *optional*, defaults to `True`):
                Whether to use classifier free guidance or not.
            num_videos_per_prompt (`int`, *optional*, defaults to 1):
                Number of videos that should be generated per prompt.
            prompt_embeds (`ms.Tensor`, *optional*):
                Pre-generated text embeddings. Can be used to easily tweak text inputs, *e.g.* prompt weighting. If not
                provided, text embeddings will be generated from `prompt` input argument.
            negative_prompt_embeds (`ms.Tensor`, *optional*):
                Pre-generated negative text embeddings. Can be used to easily tweak text inputs, *e.g.* prompt
                weighting. If not provided, negative_prompt_embeds will be generated from `negative_prompt` input
                argument.
            dtype: (`ms.Type`, *optional*):
                mindspore dtype
        """
        prompt = [prompt] if isinstance(prompt, str) else prompt
        if prompt is not None:
            batch_size = len(prompt)
        else:
            batch_size = prompt_embeds.shape[0]

        if prompt_embeds is None:
            prompt_embeds = self._get_t5_prompt_embeds(
                prompt=prompt,
                num_videos_per_prompt=num_videos_per_prompt,
                max_sequence_length=max_sequence_length,
                dtype=dtype,
            )

        if do_classifier_free_guidance and negative_prompt_embeds is None:
            negative_prompt = negative_prompt or ""
            negative_prompt = batch_size * [negative_prompt] if isinstance(negative_prompt, str) else negative_prompt

            if prompt is not None and type(prompt) is not type(negative_prompt):
                raise TypeError(
                    f"`negative_prompt` should be the same type to `prompt`, but got {type(negative_prompt)} !="
                    f" {type(prompt)}."
                )
            elif batch_size != len(negative_prompt):
                raise ValueError(
                    f"`negative_prompt`: {negative_prompt} has batch size {len(negative_prompt)}, but `prompt`:"
                    f" {prompt} has batch size {batch_size}. Please make sure that passed `negative_prompt` matches"
                    " the batch size of `prompt`."
                )

            negative_prompt_embeds = self._get_t5_prompt_embeds(
                prompt=negative_prompt,
                num_videos_per_prompt=num_videos_per_prompt,
                max_sequence_length=max_sequence_length,
                dtype=dtype,
            )

        return prompt_embeds, negative_prompt_embeds

    def check_inputs(
        self,
        prompt,
        negative_prompt,
        height,
        width,
        video=None,
        latents=None,
        prompt_embeds=None,
        negative_prompt_embeds=None,
        callback_on_step_end_tensor_inputs=None,
    ):
        if height % 16 != 0 or width % 16 != 0:
            raise ValueError(f"`height` and `width` have to be divisible by 16 but are {height} and {width}.")

        if callback_on_step_end_tensor_inputs is not None and not all(
            k in self._callback_tensor_inputs for k in callback_on_step_end_tensor_inputs
        ):
            raise ValueError(
                f"`callback_on_step_end_tensor_inputs` has to be in {self._callback_tensor_inputs}, but found {[k for k in callback_on_step_end_tensor_inputs if k not in self._callback_tensor_inputs]}"  # noqa: E501
            )

        if prompt is not None and prompt_embeds is not None:
            raise ValueError(
                f"Cannot forward both `prompt`: {prompt} and `prompt_embeds`: {prompt_embeds}. Please make sure to"
                " only forward one of the two."
            )
        elif negative_prompt is not None and negative_prompt_embeds is not None:
            raise ValueError(
                f"Cannot forward both `negative_prompt`: {negative_prompt} and `negative_prompt_embeds`: {negative_prompt_embeds}. Please make sure to"  # noqa: E501
                " only forward one of the two."
            )
        elif prompt is None and prompt_embeds is None:
            raise ValueError(
                "Provide either `prompt` or `prompt_embeds`. Cannot leave both `prompt` and `prompt_embeds` undefined."
            )
        elif prompt is not None and (not isinstance(prompt, str) and not isinstance(prompt, list)):
            raise ValueError(f"`prompt` has to be of type `str` or `list` but is {type(prompt)}")
        elif negative_prompt is not None and (
            not isinstance(negative_prompt, str) and not isinstance(negative_prompt, list)
        ):
            raise ValueError(f"`negative_prompt` has to be of type `str` or `list` but is {type(negative_prompt)}")

        if video is not None and latents is not None:
            raise ValueError("Only one of `video` or `latents` should be provided")

    def prepare_latents(
        self,
        video: Optional[ms.Tensor] = None,
        batch_size: int = 1,
        num_channels_latents: int = 16,
        height: int = 480,
        width: int = 832,
        dtype: Optional[ms.Type] = None,
        generator: Optional[np.random.Generator] = None,
        latents: Optional[ms.Tensor] = None,
        timestep: Optional[ms.Tensor] = None,
    ):
        if isinstance(generator, list) and len(generator) != batch_size:
            raise ValueError(
                f"You have passed a list of generators of length {len(generator)}, but requested an effective batch"
                f" size of {batch_size}. Make sure the batch size matches the length of the generators."
            )

        num_latent_frames = (
            (video.shape[2] - 1) // self.vae_scale_factor_temporal + 1 if latents is None else latents.shape[1]
        )
        shape = (
            batch_size,
            num_channels_latents,
            num_latent_frames,
            height // self.vae_scale_factor_spatial,
            width // self.vae_scale_factor_spatial,
        )

        if latents is None:
            # TODO: we use pynative mode here since cache in vae.encode which not supported in graph mode
            with pynative_context():
                init_latents = [
                    retrieve_latents(self.vae, self.vae.encode(vid.unsqueeze(0))[0], sample_mode="argmax")
                    for vid in video
                ]

            init_latents = mint.cat(init_latents, dim=0).to(dtype)

            latents_mean = ms.tensor(self.vae.config.latents_mean).view(1, self.vae.config.z_dim, 1, 1, 1).to(dtype)
            latents_std = 1.0 / ms.tensor(self.vae.config.latents_std).view(1, self.vae.config.z_dim, 1, 1, 1).to(dtype)

            init_latents = (init_latents - latents_mean) * latents_std

            noise = randn_tensor(shape, generator=generator, dtype=dtype)
            if hasattr(self.scheduler, "add_noise"):
                latents = self.scheduler.add_noise(init_latents, noise, timestep)
            else:
                latents = self.scheduler.scale_noise(init_latents, timestep, noise)

        return latents

    # Copied from diffusers.pipelines.animatediff.pipeline_animatediff_video2video.AnimateDiffVideoToVideoPipeline.get_timesteps
    def get_timesteps(self, num_inference_steps, timesteps, strength):
        # get the original timestep using init_timestep
        init_timestep = min(int(num_inference_steps * strength), num_inference_steps)

        t_start = max(num_inference_steps - init_timestep, 0)
        timesteps = timesteps[t_start * self.scheduler.order :]

        return timesteps, num_inference_steps - t_start

    @property
    def guidance_scale(self):
        return self._guidance_scale

    @property
    def do_classifier_free_guidance(self):
        return self._guidance_scale > 1.0

    @property
    def num_timesteps(self):
        return self._num_timesteps

    @property
    def current_timestep(self):
        return self._current_timestep

    @property
    def interrupt(self):
        return self._interrupt

    @property
    def attention_kwargs(self):
        return self._attention_kwargs

    def __call__(
        self,
        video: List[Image.Image] = None,
        prompt: Union[str, List[str]] = None,
        negative_prompt: Union[str, List[str]] = None,
        height: int = 480,
        width: int = 832,
        num_inference_steps: int = 50,
        timesteps: Optional[List[int]] = None,
        guidance_scale: float = 5.0,
        strength: float = 0.8,
        num_videos_per_prompt: Optional[int] = 1,
        generator: Optional[Union[np.random.Generator, List[np.random.Generator]]] = None,
        latents: Optional[ms.Tensor] = None,
        prompt_embeds: Optional[ms.Tensor] = None,
        negative_prompt_embeds: Optional[ms.Tensor] = None,
        output_type: Optional[str] = "np",
        return_dict: bool = False,
        attention_kwargs: Optional[Dict[str, Any]] = None,
        callback_on_step_end: Optional[
            Union[Callable[[int, int, Dict], None], PipelineCallback, MultiPipelineCallbacks]
        ] = None,
        callback_on_step_end_tensor_inputs: List[str] = ["latents"],
        max_sequence_length: int = 512,
    ):
        r"""
        The call function to the pipeline for generation.

        Args:
            prompt (`str` or `List[str]`, *optional*):
                The prompt or prompts to guide the image generation. If not defined, one has to pass `prompt_embeds`
                instead.
            height (`int`, defaults to `480`):
                The height in pixels of the generated image.
            width (`int`, defaults to `832`):
                The width in pixels of the generated image.
            num_frames (`int`, defaults to `81`):
                The number of frames in the generated video.
            num_inference_steps (`int`, defaults to `50`):
                The number of denoising steps. More denoising steps usually lead to a higher quality image at the
                expense of slower inference.
            guidance_scale (`float`, defaults to `5.0`):
                Guidance scale as defined in [Classifier-Free Diffusion
                Guidance](https://huggingface.co/papers/2207.12598). `guidance_scale` is defined as `w` of equation 2.
                of [Imagen Paper](https://huggingface.co/papers/2205.11487). Guidance scale is enabled by setting
                `guidance_scale > 1`. Higher guidance scale encourages to generate images that are closely linked to
                the text `prompt`, usually at the expense of lower image quality.
            strength (`float`, defaults to `0.8`):
                Higher strength leads to more differences between original image and generated video.
            num_videos_per_prompt (`int`, *optional*, defaults to 1):
                The number of images to generate per prompt.
            generator (`np.random.Generator` or `List[np.random.Generator]`, *optional*):
                A [`np.random.Generator`](https://numpy.org/doc/stable/reference/random/generator.html) to make
                generation deterministic.
            latents (`ms.Tensor`, *optional*):
                Pre-generated noisy latents sampled from a Gaussian distribution, to be used as inputs for image
                generation. Can be used to tweak the same generation with different prompts. If not provided, a latents
                tensor is generated by sampling using the supplied random `generator`.
            prompt_embeds (`ms.Tensor`, *optional*):
                Pre-generated text embeddings. Can be used to easily tweak text inputs (prompt weighting). If not
                provided, text embeddings are generated from the `prompt` input argument.
            output_type (`str`, *optional*, defaults to `"np"`):
                The output format of the generated image. Choose between `PIL.Image` or `np.array`.
            return_dict (`bool`, *optional*, defaults to `False`):
                Whether or not to return a [`WanPipelineOutput`] instead of a plain tuple.
            attention_kwargs (`dict`, *optional*):
                A kwargs dictionary that if specified is passed along to the `AttentionProcessor` as defined under
                `self.processor` in
                [diffusers.models.attention_processor](https://github.com/huggingface/diffusers/blob/main/src/diffusers/models/attention_processor.py).
            callback_on_step_end (`Callable`, `PipelineCallback`, `MultiPipelineCallbacks`, *optional*):
                A function or a subclass of `PipelineCallback` or `MultiPipelineCallbacks` that is called at the end of
                each denoising step during the inference. with the following arguments: `callback_on_step_end(self:
                DiffusionPipeline, step: int, timestep: int, callback_kwargs: Dict)`. `callback_kwargs` will include a
                list of all tensors as specified by `callback_on_step_end_tensor_inputs`.
            callback_on_step_end_tensor_inputs (`List`, *optional*):
                The list of tensor inputs for the `callback_on_step_end` function. The tensors specified in the list
                will be passed as `callback_kwargs` argument. You will only be able to include variables listed in the
                `._callback_tensor_inputs` attribute of your pipeline class.
            max_sequence_length (`int`, defaults to `512`):
                The maximum sequence length of the text encoder. If the prompt is longer than this, it will be
                truncated. If the prompt is shorter, it will be padded to this length.

        Examples:

        Returns:
            [`~WanPipelineOutput`] or `tuple`:
                If `return_dict` is `True`, [`WanPipelineOutput`] is returned, otherwise a `tuple` is returned where
                the first element is a list with the generated images and the second element is a list of `bool`s
                indicating whether the corresponding generated image contains "not-safe-for-work" (nsfw) content.
        """

        if isinstance(callback_on_step_end, (PipelineCallback, MultiPipelineCallbacks)):
            callback_on_step_end_tensor_inputs = callback_on_step_end.tensor_inputs

        height = height or self.transformer.config.sample_height * self.vae_scale_factor_spatial
        width = width or self.transformer.config.sample_width * self.vae_scale_factor_spatial
        num_videos_per_prompt = 1

        # 1. Check inputs. Raise error if not correct
        self.check_inputs(
            prompt,
            negative_prompt,
            height,
            width,
            video,
            latents,
            prompt_embeds,
            negative_prompt_embeds,
            callback_on_step_end_tensor_inputs,
        )

        self._guidance_scale = guidance_scale
        self._attention_kwargs = attention_kwargs
        self._current_timestep = None
        self._interrupt = False

        # 2. Define call parameters
        if prompt is not None and isinstance(prompt, str):
            batch_size = 1
        elif prompt is not None and isinstance(prompt, list):
            batch_size = len(prompt)
        else:
            batch_size = prompt_embeds.shape[0]

        # 3. Encode input prompt
        prompt_embeds, negative_prompt_embeds = self.encode_prompt(
            prompt=prompt,
            negative_prompt=negative_prompt,
            do_classifier_free_guidance=self.do_classifier_free_guidance,
            num_videos_per_prompt=num_videos_per_prompt,
            prompt_embeds=prompt_embeds,
            negative_prompt_embeds=negative_prompt_embeds,
            max_sequence_length=max_sequence_length,
        )

        transformer_dtype = self.transformer.dtype
        prompt_embeds = prompt_embeds.to(transformer_dtype)
        if negative_prompt_embeds is not None:
            negative_prompt_embeds = negative_prompt_embeds.to(transformer_dtype)

        # 4. Prepare timesteps
        timesteps, num_inference_steps = retrieve_timesteps(self.scheduler, num_inference_steps, timesteps)
        timesteps, num_inference_steps = self.get_timesteps(num_inference_steps, timesteps, strength)
        latent_timestep = timesteps[:1].repeat(batch_size * num_videos_per_prompt)
        self._num_timesteps = len(timesteps)

        if latents is None:
            video = self.video_processor.preprocess_video(video, height=height, width=width).to(dtype=ms.float32)

        # 5. Prepare latent variables
        num_channels_latents = self.transformer.config.in_channels
        latents = self.prepare_latents(
            video,
            batch_size * num_videos_per_prompt,
            num_channels_latents,
            height,
            width,
            ms.float32,
            generator,
            latents,
            latent_timestep,
        )

        # 6. Denoising loop
        num_warmup_steps = len(timesteps) - num_inference_steps * self.scheduler.order
        self._num_timesteps = len(timesteps)

        # we're popping the `scale` instead of getting it because otherwise `scale` will be propagated
        # to the transformer and will raise RuntimeError.
        lora_scale = attention_kwargs.pop("scale", None) if attention_kwargs is not None else None
        if lora_scale is not None:
            # weight the lora layers by setting `lora_scale` for each PEFT layer
            scale_lora_layers(self.transformer, lora_scale)

        with self.progress_bar(total=num_inference_steps) as progress_bar:
            for i, t in enumerate(timesteps):
                if self.interrupt:
                    continue

                self._current_timestep = t
                latent_model_input = latents.to(transformer_dtype)
                timestep = t.broadcast_to((latents.shape[0],))

                noise_pred = self.transformer(
                    hidden_states=latent_model_input,
                    timestep=timestep,
                    encoder_hidden_states=prompt_embeds,
                    attention_kwargs=attention_kwargs,
                    return_dict=False,
                )[0]

                if self.do_classifier_free_guidance:
                    noise_uncond = self.transformer(
                        hidden_states=latent_model_input,
                        timestep=timestep,
                        encoder_hidden_states=negative_prompt_embeds,
                        attention_kwargs=attention_kwargs,
                        return_dict=False,
                    )[0]
                    noise_pred = noise_uncond + guidance_scale * (noise_pred - noise_uncond)

                # compute the previous noisy sample x_t -> x_t-1
                latents = self.scheduler.step(noise_pred, t, latents, return_dict=False)[0]

                if callback_on_step_end is not None:
                    callback_kwargs = {}
                    for k in callback_on_step_end_tensor_inputs:
                        callback_kwargs[k] = locals()[k]
                    callback_outputs = callback_on_step_end(self, i, t, callback_kwargs)

                    latents = callback_outputs.pop("latents", latents)
                    prompt_embeds = callback_outputs.pop("prompt_embeds", prompt_embeds)
                    negative_prompt_embeds = callback_outputs.pop("negative_prompt_embeds", negative_prompt_embeds)

                # call the callback, if provided
                if i == len(timesteps) - 1 or ((i + 1) > num_warmup_steps and (i + 1) % self.scheduler.order == 0):
                    progress_bar.update()

        if lora_scale is not None:
            # remove `lora_scale` from each PEFT layer
            unscale_lora_layers(self.transformer, lora_scale)
            attention_kwargs["scale"] = lora_scale

        self._current_timestep = None

        if not output_type == "latent":
            latents = latents.to(self.vae.dtype)
            latents_mean = (
                ms.tensor(self.vae.config.latents_mean).view(1, self.vae.config.z_dim, 1, 1, 1).to(latents.dtype)
            )
            latents_std = 1.0 / ms.tensor(self.vae.config.latents_std).view(1, self.vae.config.z_dim, 1, 1, 1).to(
                latents.dtype
            )
            latents = latents / latents_std + latents_mean
            # TODO: we use pynative mode here since cache in vae.decode which not supported in graph mode
            with pynative_context():
                video = self.vae.decode(latents, return_dict=False)[0]
            video = self.video_processor.postprocess_video(video, output_type=output_type)
        else:
            video = latents

        if not return_dict:
            return (video,)

        return WanPipelineOutput(frames=video)

mindone.diffusers.WanVideoToVideoPipeline.__call__(video=None, prompt=None, negative_prompt=None, height=480, width=832, num_inference_steps=50, timesteps=None, guidance_scale=5.0, strength=0.8, num_videos_per_prompt=1, generator=None, latents=None, prompt_embeds=None, negative_prompt_embeds=None, output_type='np', return_dict=False, attention_kwargs=None, callback_on_step_end=None, callback_on_step_end_tensor_inputs=['latents'], max_sequence_length=512)

The call function to the pipeline for generation.

PARAMETER	DESCRIPTION
prompt	The prompt or prompts to guide the image generation. If not defined, one has to pass prompt_embeds instead. TYPE: `str` or `List[str]`, *optional* DEFAULT: None
height	The height in pixels of the generated image. TYPE: `int`, defaults to `480` DEFAULT: 480
width	The width in pixels of the generated image. TYPE: `int`, defaults to `832` DEFAULT: 832
num_frames	The number of frames in the generated video. TYPE: `int`, defaults to `81`
num_inference_steps	The number of denoising steps. More denoising steps usually lead to a higher quality image at the expense of slower inference. TYPE: `int`, defaults to `50` DEFAULT: 50
guidance_scale	Guidance scale as defined in Classifier-Free Diffusion Guidance. guidance_scale is defined as w of equation 2. of Imagen Paper. Guidance scale is enabled by setting guidance_scale > 1. Higher guidance scale encourages to generate images that are closely linked to the text prompt, usually at the expense of lower image quality. TYPE: `float`, defaults to `5.0` DEFAULT: 5.0
strength	Higher strength leads to more differences between original image and generated video. TYPE: `float`, defaults to `0.8` DEFAULT: 0.8
num_videos_per_prompt	The number of images to generate per prompt. TYPE: `int`, *optional*, defaults to 1 DEFAULT: 1
generator	A np.random.Generator to make generation deterministic. TYPE: `np.random.Generator` or `List[np.random.Generator]`, *optional* DEFAULT: None
latents	Pre-generated noisy latents sampled from a Gaussian distribution, to be used as inputs for image generation. Can be used to tweak the same generation with different prompts. If not provided, a latents tensor is generated by sampling using the supplied random generator. TYPE: `ms.Tensor`, *optional* DEFAULT: None
prompt_embeds	Pre-generated text embeddings. Can be used to easily tweak text inputs (prompt weighting). If not provided, text embeddings are generated from the prompt input argument. TYPE: `ms.Tensor`, *optional* DEFAULT: None
output_type	The output format of the generated image. Choose between PIL.Image or np.array. TYPE: `str`, *optional*, defaults to `"np"` DEFAULT: 'np'
return_dict	Whether or not to return a [WanPipelineOutput] instead of a plain tuple. TYPE: `bool`, *optional*, defaults to `False` DEFAULT: False
attention_kwargs	A kwargs dictionary that if specified is passed along to the AttentionProcessor as defined under self.processor in diffusers.models.attention_processor. TYPE: `dict`, *optional* DEFAULT: None
callback_on_step_end	A function or a subclass of PipelineCallback or MultiPipelineCallbacks that is called at the end of each denoising step during the inference. with the following arguments: callback_on_step_end(self: DiffusionPipeline, step: int, timestep: int, callback_kwargs: Dict). callback_kwargs will include a list of all tensors as specified by callback_on_step_end_tensor_inputs. TYPE: `Callable`, `PipelineCallback`, `MultiPipelineCallbacks`, *optional* DEFAULT: None
callback_on_step_end_tensor_inputs	The list of tensor inputs for the callback_on_step_end function. The tensors specified in the list will be passed as callback_kwargs argument. You will only be able to include variables listed in the ._callback_tensor_inputs attribute of your pipeline class. TYPE: `List`, *optional* DEFAULT: ['latents']
max_sequence_length	The maximum sequence length of the text encoder. If the prompt is longer than this, it will be truncated. If the prompt is shorter, it will be padded to this length. TYPE: `int`, defaults to `512` DEFAULT: 512

RETURNS	DESCRIPTION
	[~WanPipelineOutput] or tuple: If return_dict is True, [WanPipelineOutput] is returned, otherwise a tuple is returned where the first element is a list with the generated images and the second element is a list of bools indicating whether the corresponding generated image contains "not-safe-for-work" (nsfw) content.

Source code in mindone/diffusers/pipelines/wan/pipeline_wan_video2video.py

def __call__(
    self,
    video: List[Image.Image] = None,
    prompt: Union[str, List[str]] = None,
    negative_prompt: Union[str, List[str]] = None,
    height: int = 480,
    width: int = 832,
    num_inference_steps: int = 50,
    timesteps: Optional[List[int]] = None,
    guidance_scale: float = 5.0,
    strength: float = 0.8,
    num_videos_per_prompt: Optional[int] = 1,
    generator: Optional[Union[np.random.Generator, List[np.random.Generator]]] = None,
    latents: Optional[ms.Tensor] = None,
    prompt_embeds: Optional[ms.Tensor] = None,
    negative_prompt_embeds: Optional[ms.Tensor] = None,
    output_type: Optional[str] = "np",
    return_dict: bool = False,
    attention_kwargs: Optional[Dict[str, Any]] = None,
    callback_on_step_end: Optional[
        Union[Callable[[int, int, Dict], None], PipelineCallback, MultiPipelineCallbacks]
    ] = None,
    callback_on_step_end_tensor_inputs: List[str] = ["latents"],
    max_sequence_length: int = 512,
):
    r"""
    The call function to the pipeline for generation.

    Args:
        prompt (`str` or `List[str]`, *optional*):
            The prompt or prompts to guide the image generation. If not defined, one has to pass `prompt_embeds`
            instead.
        height (`int`, defaults to `480`):
            The height in pixels of the generated image.
        width (`int`, defaults to `832`):
            The width in pixels of the generated image.
        num_frames (`int`, defaults to `81`):
            The number of frames in the generated video.
        num_inference_steps (`int`, defaults to `50`):
            The number of denoising steps. More denoising steps usually lead to a higher quality image at the
            expense of slower inference.
        guidance_scale (`float`, defaults to `5.0`):
            Guidance scale as defined in [Classifier-Free Diffusion
            Guidance](https://huggingface.co/papers/2207.12598). `guidance_scale` is defined as `w` of equation 2.
            of [Imagen Paper](https://huggingface.co/papers/2205.11487). Guidance scale is enabled by setting
            `guidance_scale > 1`. Higher guidance scale encourages to generate images that are closely linked to
            the text `prompt`, usually at the expense of lower image quality.
        strength (`float`, defaults to `0.8`):
            Higher strength leads to more differences between original image and generated video.
        num_videos_per_prompt (`int`, *optional*, defaults to 1):
            The number of images to generate per prompt.
        generator (`np.random.Generator` or `List[np.random.Generator]`, *optional*):
            A [`np.random.Generator`](https://numpy.org/doc/stable/reference/random/generator.html) to make
            generation deterministic.
        latents (`ms.Tensor`, *optional*):
            Pre-generated noisy latents sampled from a Gaussian distribution, to be used as inputs for image
            generation. Can be used to tweak the same generation with different prompts. If not provided, a latents
            tensor is generated by sampling using the supplied random `generator`.
        prompt_embeds (`ms.Tensor`, *optional*):
            Pre-generated text embeddings. Can be used to easily tweak text inputs (prompt weighting). If not
            provided, text embeddings are generated from the `prompt` input argument.
        output_type (`str`, *optional*, defaults to `"np"`):
            The output format of the generated image. Choose between `PIL.Image` or `np.array`.
        return_dict (`bool`, *optional*, defaults to `False`):
            Whether or not to return a [`WanPipelineOutput`] instead of a plain tuple.
        attention_kwargs (`dict`, *optional*):
            A kwargs dictionary that if specified is passed along to the `AttentionProcessor` as defined under
            `self.processor` in
            [diffusers.models.attention_processor](https://github.com/huggingface/diffusers/blob/main/src/diffusers/models/attention_processor.py).
        callback_on_step_end (`Callable`, `PipelineCallback`, `MultiPipelineCallbacks`, *optional*):
            A function or a subclass of `PipelineCallback` or `MultiPipelineCallbacks` that is called at the end of
            each denoising step during the inference. with the following arguments: `callback_on_step_end(self:
            DiffusionPipeline, step: int, timestep: int, callback_kwargs: Dict)`. `callback_kwargs` will include a
            list of all tensors as specified by `callback_on_step_end_tensor_inputs`.
        callback_on_step_end_tensor_inputs (`List`, *optional*):
            The list of tensor inputs for the `callback_on_step_end` function. The tensors specified in the list
            will be passed as `callback_kwargs` argument. You will only be able to include variables listed in the
            `._callback_tensor_inputs` attribute of your pipeline class.
        max_sequence_length (`int`, defaults to `512`):
            The maximum sequence length of the text encoder. If the prompt is longer than this, it will be
            truncated. If the prompt is shorter, it will be padded to this length.

    Examples:

    Returns:
        [`~WanPipelineOutput`] or `tuple`:
            If `return_dict` is `True`, [`WanPipelineOutput`] is returned, otherwise a `tuple` is returned where
            the first element is a list with the generated images and the second element is a list of `bool`s
            indicating whether the corresponding generated image contains "not-safe-for-work" (nsfw) content.
    """

    if isinstance(callback_on_step_end, (PipelineCallback, MultiPipelineCallbacks)):
        callback_on_step_end_tensor_inputs = callback_on_step_end.tensor_inputs

    height = height or self.transformer.config.sample_height * self.vae_scale_factor_spatial
    width = width or self.transformer.config.sample_width * self.vae_scale_factor_spatial
    num_videos_per_prompt = 1

    # 1. Check inputs. Raise error if not correct
    self.check_inputs(
        prompt,
        negative_prompt,
        height,
        width,
        video,
        latents,
        prompt_embeds,
        negative_prompt_embeds,
        callback_on_step_end_tensor_inputs,
    )

    self._guidance_scale = guidance_scale
    self._attention_kwargs = attention_kwargs
    self._current_timestep = None
    self._interrupt = False

    # 2. Define call parameters
    if prompt is not None and isinstance(prompt, str):
        batch_size = 1
    elif prompt is not None and isinstance(prompt, list):
        batch_size = len(prompt)
    else:
        batch_size = prompt_embeds.shape[0]

    # 3. Encode input prompt
    prompt_embeds, negative_prompt_embeds = self.encode_prompt(
        prompt=prompt,
        negative_prompt=negative_prompt,
        do_classifier_free_guidance=self.do_classifier_free_guidance,
        num_videos_per_prompt=num_videos_per_prompt,
        prompt_embeds=prompt_embeds,
        negative_prompt_embeds=negative_prompt_embeds,
        max_sequence_length=max_sequence_length,
    )

    transformer_dtype = self.transformer.dtype
    prompt_embeds = prompt_embeds.to(transformer_dtype)
    if negative_prompt_embeds is not None:
        negative_prompt_embeds = negative_prompt_embeds.to(transformer_dtype)

    # 4. Prepare timesteps
    timesteps, num_inference_steps = retrieve_timesteps(self.scheduler, num_inference_steps, timesteps)
    timesteps, num_inference_steps = self.get_timesteps(num_inference_steps, timesteps, strength)
    latent_timestep = timesteps[:1].repeat(batch_size * num_videos_per_prompt)
    self._num_timesteps = len(timesteps)

    if latents is None:
        video = self.video_processor.preprocess_video(video, height=height, width=width).to(dtype=ms.float32)

    # 5. Prepare latent variables
    num_channels_latents = self.transformer.config.in_channels
    latents = self.prepare_latents(
        video,
        batch_size * num_videos_per_prompt,
        num_channels_latents,
        height,
        width,
        ms.float32,
        generator,
        latents,
        latent_timestep,
    )

    # 6. Denoising loop
    num_warmup_steps = len(timesteps) - num_inference_steps * self.scheduler.order
    self._num_timesteps = len(timesteps)

    # we're popping the `scale` instead of getting it because otherwise `scale` will be propagated
    # to the transformer and will raise RuntimeError.
    lora_scale = attention_kwargs.pop("scale", None) if attention_kwargs is not None else None
    if lora_scale is not None:
        # weight the lora layers by setting `lora_scale` for each PEFT layer
        scale_lora_layers(self.transformer, lora_scale)

    with self.progress_bar(total=num_inference_steps) as progress_bar:
        for i, t in enumerate(timesteps):
            if self.interrupt:
                continue

            self._current_timestep = t
            latent_model_input = latents.to(transformer_dtype)
            timestep = t.broadcast_to((latents.shape[0],))

            noise_pred = self.transformer(
                hidden_states=latent_model_input,
                timestep=timestep,
                encoder_hidden_states=prompt_embeds,
                attention_kwargs=attention_kwargs,
                return_dict=False,
            )[0]

            if self.do_classifier_free_guidance:
                noise_uncond = self.transformer(
                    hidden_states=latent_model_input,
                    timestep=timestep,
                    encoder_hidden_states=negative_prompt_embeds,
                    attention_kwargs=attention_kwargs,
                    return_dict=False,
                )[0]
                noise_pred = noise_uncond + guidance_scale * (noise_pred - noise_uncond)

            # compute the previous noisy sample x_t -> x_t-1
            latents = self.scheduler.step(noise_pred, t, latents, return_dict=False)[0]

            if callback_on_step_end is not None:
                callback_kwargs = {}
                for k in callback_on_step_end_tensor_inputs:
                    callback_kwargs[k] = locals()[k]
                callback_outputs = callback_on_step_end(self, i, t, callback_kwargs)

                latents = callback_outputs.pop("latents", latents)
                prompt_embeds = callback_outputs.pop("prompt_embeds", prompt_embeds)
                negative_prompt_embeds = callback_outputs.pop("negative_prompt_embeds", negative_prompt_embeds)

            # call the callback, if provided
            if i == len(timesteps) - 1 or ((i + 1) > num_warmup_steps and (i + 1) % self.scheduler.order == 0):
                progress_bar.update()

    if lora_scale is not None:
        # remove `lora_scale` from each PEFT layer
        unscale_lora_layers(self.transformer, lora_scale)
        attention_kwargs["scale"] = lora_scale

    self._current_timestep = None

    if not output_type == "latent":
        latents = latents.to(self.vae.dtype)
        latents_mean = (
            ms.tensor(self.vae.config.latents_mean).view(1, self.vae.config.z_dim, 1, 1, 1).to(latents.dtype)
        )
        latents_std = 1.0 / ms.tensor(self.vae.config.latents_std).view(1, self.vae.config.z_dim, 1, 1, 1).to(
            latents.dtype
        )
        latents = latents / latents_std + latents_mean
        # TODO: we use pynative mode here since cache in vae.decode which not supported in graph mode
        with pynative_context():
            video = self.vae.decode(latents, return_dict=False)[0]
        video = self.video_processor.postprocess_video(video, output_type=output_type)
    else:
        video = latents

    if not return_dict:
        return (video,)

    return WanPipelineOutput(frames=video)

mindone.diffusers.WanVideoToVideoPipeline.encode_prompt(prompt, negative_prompt=None, do_classifier_free_guidance=True, num_videos_per_prompt=1, prompt_embeds=None, negative_prompt_embeds=None, max_sequence_length=226, dtype=None)

Encodes the prompt into text encoder hidden states.

PARAMETER	DESCRIPTION
prompt	prompt to be encoded TYPE: `str` or `List[str]`, *optional*
negative_prompt	The prompt or prompts not to guide the image generation. If not defined, one has to pass negative_prompt_embeds instead. Ignored when not using guidance (i.e., ignored if guidance_scale is less than 1). TYPE: `str` or `List[str]`, *optional* DEFAULT: None
do_classifier_free_guidance	Whether to use classifier free guidance or not. TYPE: `bool`, *optional*, defaults to `True` DEFAULT: True
num_videos_per_prompt	Number of videos that should be generated per prompt. TYPE: `int`, *optional*, defaults to 1 DEFAULT: 1
prompt_embeds	Pre-generated text embeddings. Can be used to easily tweak text inputs, e.g. prompt weighting. If not provided, text embeddings will be generated from prompt input argument. TYPE: `ms.Tensor`, *optional* DEFAULT: None
negative_prompt_embeds	Pre-generated negative text embeddings. Can be used to easily tweak text inputs, e.g. prompt weighting. If not provided, negative_prompt_embeds will be generated from negative_prompt input argument. TYPE: `ms.Tensor`, *optional* DEFAULT: None
dtype	(ms.Type, optional): mindspore dtype TYPE: Optional[Type] DEFAULT: None

Source code in mindone/diffusers/pipelines/wan/pipeline_wan_video2video.py

def encode_prompt(
    self,
    prompt: Union[str, List[str]],
    negative_prompt: Optional[Union[str, List[str]]] = None,
    do_classifier_free_guidance: bool = True,
    num_videos_per_prompt: int = 1,
    prompt_embeds: Optional[ms.Tensor] = None,
    negative_prompt_embeds: Optional[ms.Tensor] = None,
    max_sequence_length: int = 226,
    dtype: Optional[ms.Type] = None,
):
    r"""
    Encodes the prompt into text encoder hidden states.

    Args:
        prompt (`str` or `List[str]`, *optional*):
            prompt to be encoded
        negative_prompt (`str` or `List[str]`, *optional*):
            The prompt or prompts not to guide the image generation. If not defined, one has to pass
            `negative_prompt_embeds` instead. Ignored when not using guidance (i.e., ignored if `guidance_scale` is
            less than `1`).
        do_classifier_free_guidance (`bool`, *optional*, defaults to `True`):
            Whether to use classifier free guidance or not.
        num_videos_per_prompt (`int`, *optional*, defaults to 1):
            Number of videos that should be generated per prompt.
        prompt_embeds (`ms.Tensor`, *optional*):
            Pre-generated text embeddings. Can be used to easily tweak text inputs, *e.g.* prompt weighting. If not
            provided, text embeddings will be generated from `prompt` input argument.
        negative_prompt_embeds (`ms.Tensor`, *optional*):
            Pre-generated negative text embeddings. Can be used to easily tweak text inputs, *e.g.* prompt
            weighting. If not provided, negative_prompt_embeds will be generated from `negative_prompt` input
            argument.
        dtype: (`ms.Type`, *optional*):
            mindspore dtype
    """
    prompt = [prompt] if isinstance(prompt, str) else prompt
    if prompt is not None:
        batch_size = len(prompt)
    else:
        batch_size = prompt_embeds.shape[0]

    if prompt_embeds is None:
        prompt_embeds = self._get_t5_prompt_embeds(
            prompt=prompt,
            num_videos_per_prompt=num_videos_per_prompt,
            max_sequence_length=max_sequence_length,
            dtype=dtype,
        )

    if do_classifier_free_guidance and negative_prompt_embeds is None:
        negative_prompt = negative_prompt or ""
        negative_prompt = batch_size * [negative_prompt] if isinstance(negative_prompt, str) else negative_prompt

        if prompt is not None and type(prompt) is not type(negative_prompt):
            raise TypeError(
                f"`negative_prompt` should be the same type to `prompt`, but got {type(negative_prompt)} !="
                f" {type(prompt)}."
            )
        elif batch_size != len(negative_prompt):
            raise ValueError(
                f"`negative_prompt`: {negative_prompt} has batch size {len(negative_prompt)}, but `prompt`:"
                f" {prompt} has batch size {batch_size}. Please make sure that passed `negative_prompt` matches"
                " the batch size of `prompt`."
            )

        negative_prompt_embeds = self._get_t5_prompt_embeds(
            prompt=negative_prompt,
            num_videos_per_prompt=num_videos_per_prompt,
            max_sequence_length=max_sequence_length,
            dtype=dtype,
        )

    return prompt_embeds, negative_prompt_embeds

mindone.diffusers.pipelines.wan.pipeline_output.WanPipelineOutput dataclass

Bases: BaseOutput

Output class for Wan pipelines.

PARAMETER	DESCRIPTION
frames	List of video outputs - It can be a nested list of length batch_size, with each sub-list containing denoised PIL image sequences of length num_frames. It can also be a NumPy array or MindSpore tensor of shape (batch_size, num_frames, channels, height, width). TYPE: `ms.Tensor`, `np.ndarray`, or List[List[PIL.Image.Image]]

Source code in mindone/diffusers/pipelines/wan/pipeline_output.py

@dataclass
class WanPipelineOutput(BaseOutput):
    r"""
    Output class for Wan pipelines.

    Args:
        frames (`ms.Tensor`, `np.ndarray`, or List[List[PIL.Image.Image]]):
            List of video outputs - It can be a nested list of length `batch_size,` with each sub-list containing
            denoised PIL image sequences of length `num_frames.` It can also be a NumPy array or MindSpore tensor of
            shape `(batch_size, num_frames, channels, height, width)`.
    """

    frames: ms.Tensor