examples/community/pipeline_animatediff_controlnet.py

# Copyright 2024 The HuggingFace Team. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
#     http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.

import inspect
from typing import Any, Callable, Dict, List, Optional, Tuple, Union

import numpy as np
import torch
import torch.nn.functional as F
from PIL import Image
from transformers import CLIPImageProcessor, CLIPTextModel, CLIPTokenizer, CLIPVisionModelWithProjection

from diffusers.image_processor import PipelineImageInput, VaeImageProcessor
from diffusers.loaders import IPAdapterMixin, LoraLoaderMixin, TextualInversionLoaderMixin
from diffusers.models import AutoencoderKL, ControlNetModel, ImageProjection, UNet2DConditionModel, UNetMotionModel
from diffusers.models.lora import adjust_lora_scale_text_encoder
from diffusers.models.unets.unet_motion_model import MotionAdapter
from diffusers.pipelines.animatediff.pipeline_output import AnimateDiffPipelineOutput
from diffusers.pipelines.controlnet.multicontrolnet import MultiControlNetModel
from diffusers.pipelines.pipeline_utils import DiffusionPipeline, StableDiffusionMixin
from diffusers.schedulers import (
    DDIMScheduler,
    DPMSolverMultistepScheduler,
    EulerAncestralDiscreteScheduler,
    EulerDiscreteScheduler,
    LMSDiscreteScheduler,
    PNDMScheduler,
)
from diffusers.utils import USE_PEFT_BACKEND, deprecate, logging, scale_lora_layers, unscale_lora_layers
from diffusers.utils.torch_utils import is_compiled_module, randn_tensor


logger = logging.get_logger(__name__)  # pylint: disable=invalid-name

EXAMPLE_DOC_STRING = """
    Examples:
        ```py
        >>> import torch
        >>> from diffusers import AutoencoderKL, ControlNetModel, MotionAdapter
        >>> from diffusers.pipelines import DiffusionPipeline
        >>> from diffusers.schedulers import DPMSolverMultistepScheduler
        >>> from PIL import Image

        >>> motion_id = "guoyww/animatediff-motion-adapter-v1-5-2"
        >>> adapter = MotionAdapter.from_pretrained(motion_id)
        >>> controlnet = ControlNetModel.from_pretrained("lllyasviel/control_v11p_sd15_openpose", torch_dtype=torch.float16)
        >>> vae = AutoencoderKL.from_pretrained("stabilityai/sd-vae-ft-mse", torch_dtype=torch.float16)

        >>> model_id = "SG161222/Realistic_Vision_V5.1_noVAE"
        >>> pipe = DiffusionPipeline.from_pretrained(
        ...     model_id,
        ...     motion_adapter=adapter,
        ...     controlnet=controlnet,
        ...     vae=vae,
        ...     custom_pipeline="pipeline_animatediff_controlnet",
        ... ).to(device="cuda", dtype=torch.float16)
        >>> pipe.scheduler = DPMSolverMultistepScheduler.from_pretrained(
        ...     model_id, subfolder="scheduler", clip_sample=False, timestep_spacing="linspace", steps_offset=1, beta_schedule="linear",
        ... )
        >>> pipe.enable_vae_slicing()

        >>> conditioning_frames = []
        >>> for i in range(1, 16 + 1):
        ...     conditioning_frames.append(Image.open(f"frame_{i}.png"))

        >>> prompt = "astronaut in space, dancing"
        >>> negative_prompt = "bad quality, worst quality, jpeg artifacts, ugly"
        >>> result = pipe(
        ...     prompt=prompt,
        ...     negative_prompt=negative_prompt,
        ...     width=512,
        ...     height=768,
        ...     conditioning_frames=conditioning_frames,
        ...     num_inference_steps=12,
        ... )

        >>> from diffusers.utils import export_to_gif
        >>> export_to_gif(result.frames[0], "result.gif")
        ```
"""


# Copied from diffusers.pipelines.animatediff.pipeline_animatediff.tensor2vid
def tensor2vid(video: torch.Tensor, processor, output_type="np"):
    batch_size, channels, num_frames, height, width = video.shape
    outputs = []
    for batch_idx in range(batch_size):
        batch_vid = video[batch_idx].permute(1, 0, 2, 3)
        batch_output = processor.postprocess(batch_vid, output_type)

        outputs.append(batch_output)

    if output_type == "np":
        outputs = np.stack(outputs)

    elif output_type == "pt":
        outputs = torch.stack(outputs)

    elif not output_type == "pil":
        raise ValueError(f"{output_type} does not exist. Please choose one of ['np', 'pt', 'pil']")

    return outputs


class AnimateDiffControlNetPipeline(
    DiffusionPipeline, StableDiffusionMixin, TextualInversionLoaderMixin, IPAdapterMixin, LoraLoaderMixin
):
    r"""
    Pipeline for text-to-video generation.

    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.).

    The pipeline also inherits the following loading methods:
        - [`~loaders.TextualInversionLoaderMixin.load_textual_inversion`] for loading textual inversion embeddings
        - [`~loaders.LoraLoaderMixin.load_lora_weights`] for loading LoRA weights
        - [`~loaders.LoraLoaderMixin.save_lora_weights`] for saving LoRA weights
        - [`~loaders.IPAdapterMixin.load_ip_adapter`] for loading IP Adapters

    Args:
        vae ([`AutoencoderKL`]):
            Variational Auto-Encoder (VAE) Model to encode and decode images to and from latent representations.
        text_encoder ([`CLIPTextModel`]):
            Frozen text-encoder ([clip-vit-large-patch14](https://huggingface.co/openai/clip-vit-large-patch14)).
        tokenizer (`CLIPTokenizer`):
            A [`~transformers.CLIPTokenizer`] to tokenize text.
        unet ([`UNet2DConditionModel`]):
            A [`UNet2DConditionModel`] used to create a UNetMotionModel to denoise the encoded video latents.
        motion_adapter ([`MotionAdapter`]):
            A [`MotionAdapter`] to be used in combination with `unet` to denoise the encoded video latents.
        scheduler ([`SchedulerMixin`]):
            A scheduler to be used in combination with `unet` to denoise the encoded image latents. Can be one of
            [`DDIMScheduler`], [`LMSDiscreteScheduler`], or [`PNDMScheduler`].
    """

    model_cpu_offload_seq = "text_encoder->unet->vae"
    _optional_components = ["feature_extractor", "image_encoder"]
    _callback_tensor_inputs = ["latents", "prompt_embeds", "negative_prompt_embeds"]

    def __init__(
        self,
        vae: AutoencoderKL,
        text_encoder: CLIPTextModel,
        tokenizer: CLIPTokenizer,
        unet: UNet2DConditionModel,
        motion_adapter: MotionAdapter,
        controlnet: Union[ControlNetModel, List[ControlNetModel], Tuple[ControlNetModel], MultiControlNetModel],
        scheduler: Union[
            DDIMScheduler,
            PNDMScheduler,
            LMSDiscreteScheduler,
            EulerDiscreteScheduler,
            EulerAncestralDiscreteScheduler,
            DPMSolverMultistepScheduler,
        ],
        feature_extractor: Optional[CLIPImageProcessor] = None,
        image_encoder: Optional[CLIPVisionModelWithProjection] = None,
    ):
        super().__init__()
        unet = UNetMotionModel.from_unet2d(unet, motion_adapter)

        if isinstance(controlnet, (list, tuple)):
            controlnet = MultiControlNetModel(controlnet)

        self.register_modules(
            vae=vae,
            text_encoder=text_encoder,
            tokenizer=tokenizer,
            unet=unet,
            motion_adapter=motion_adapter,
            controlnet=controlnet,
            scheduler=scheduler,
            feature_extractor=feature_extractor,
            image_encoder=image_encoder,
        )
        self.vae_scale_factor = 2 ** (len(self.vae.config.block_out_channels) - 1)
        self.image_processor = VaeImageProcessor(vae_scale_factor=self.vae_scale_factor)
        self.control_image_processor = VaeImageProcessor(
            vae_scale_factor=self.vae_scale_factor, do_convert_rgb=True, do_normalize=False
        )

    # Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.StableDiffusionPipeline.encode_prompt with num_images_per_prompt -> num_videos_per_prompt
    def encode_prompt(
        self,
        prompt,
        device,
        num_images_per_prompt,
        do_classifier_free_guidance,
        negative_prompt=None,
        prompt_embeds: Optional[torch.Tensor] = None,
        negative_prompt_embeds: Optional[torch.Tensor] = None,
        lora_scale: Optional[float] = None,
        clip_skip: Optional[int] = None,
    ):
        r"""
        Encodes the prompt into text encoder hidden states.

        Args:
            prompt (`str` or `List[str]`, *optional*):
                prompt to be encoded
            device: (`torch.device`):
                torch device
            num_images_per_prompt (`int`):
                number of images that should be generated per prompt
            do_classifier_free_guidance (`bool`):
                whether to use classifier free guidance or not
            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`).
            prompt_embeds (`torch.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 (`torch.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.
            lora_scale (`float`, *optional*):
                A LoRA scale that will be applied to all LoRA layers of the text encoder if LoRA layers are loaded.
            clip_skip (`int`, *optional*):
                Number of layers to be skipped from CLIP while computing the prompt embeddings. A value of 1 means that
                the output of the pre-final layer will be used for computing the prompt embeddings.
        """
        # set lora scale so that monkey patched LoRA
        # function of text encoder can correctly access it
        if lora_scale is not None and isinstance(self, LoraLoaderMixin):
            self._lora_scale = lora_scale

            # dynamically adjust the LoRA scale
            if not USE_PEFT_BACKEND:
                adjust_lora_scale_text_encoder(self.text_encoder, lora_scale)
            else:
                scale_lora_layers(self.text_encoder, lora_scale)

        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]

        if prompt_embeds is None:
            # textual inversion: process multi-vector tokens if necessary
            if isinstance(self, TextualInversionLoaderMixin):
                prompt = self.maybe_convert_prompt(prompt, self.tokenizer)

            text_inputs = self.tokenizer(
                prompt,
                padding="max_length",
                max_length=self.tokenizer.model_max_length,
                truncation=True,
                return_tensors="pt",
            )
            text_input_ids = text_inputs.input_ids
            untruncated_ids = self.tokenizer(prompt, padding="longest", return_tensors="pt").input_ids

            if untruncated_ids.shape[-1] >= text_input_ids.shape[-1] and not torch.equal(
                text_input_ids, untruncated_ids
            ):
                removed_text = self.tokenizer.batch_decode(
                    untruncated_ids[:, self.tokenizer.model_max_length - 1 : -1]
                )
                logger.warning(
                    "The following part of your input was truncated because CLIP can only handle sequences up to"
                    f" {self.tokenizer.model_max_length} tokens: {removed_text}"
                )

            if hasattr(self.text_encoder.config, "use_attention_mask") and self.text_encoder.config.use_attention_mask:
                attention_mask = text_inputs.attention_mask.to(device)
            else:
                attention_mask = None

            if clip_skip is None:
                prompt_embeds = self.text_encoder(text_input_ids.to(device), attention_mask=attention_mask)
                prompt_embeds = prompt_embeds[0]
            else:
                prompt_embeds = self.text_encoder(
                    text_input_ids.to(device), attention_mask=attention_mask, output_hidden_states=True
                )
                # Access the `hidden_states` first, that contains a tuple of
                # all the hidden states from the encoder layers. Then index into
                # the tuple to access the hidden states from the desired layer.
                prompt_embeds = prompt_embeds[-1][-(clip_skip + 1)]
                # We also need to apply the final LayerNorm here to not mess with the
                # representations. The `last_hidden_states` that we typically use for
                # obtaining the final prompt representations passes through the LayerNorm
                # layer.
                prompt_embeds = self.text_encoder.text_model.final_layer_norm(prompt_embeds)

        if self.text_encoder is not None:
            prompt_embeds_dtype = self.text_encoder.dtype
        elif self.unet is not None:
            prompt_embeds_dtype = self.unet.dtype
        else:
            prompt_embeds_dtype = prompt_embeds.dtype

        prompt_embeds = prompt_embeds.to(dtype=prompt_embeds_dtype, device=device)

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

        # get unconditional embeddings for classifier free guidance
        if do_classifier_free_guidance and negative_prompt_embeds is None:
            uncond_tokens: List[str]
            if negative_prompt is None:
                uncond_tokens = [""] * batch_size
            elif 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 isinstance(negative_prompt, str):
                uncond_tokens = [negative_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`."
                )
            else:
                uncond_tokens = negative_prompt

            # textual inversion: process multi-vector tokens if necessary
            if isinstance(self, TextualInversionLoaderMixin):
                uncond_tokens = self.maybe_convert_prompt(uncond_tokens, self.tokenizer)

            max_length = prompt_embeds.shape[1]
            uncond_input = self.tokenizer(
                uncond_tokens,
                padding="max_length",
                max_length=max_length,
                truncation=True,
                return_tensors="pt",
            )

            if hasattr(self.text_encoder.config, "use_attention_mask") and self.text_encoder.config.use_attention_mask:
                attention_mask = uncond_input.attention_mask.to(device)
            else:
                attention_mask = None

            negative_prompt_embeds = self.text_encoder(
                uncond_input.input_ids.to(device),
                attention_mask=attention_mask,
            )
            negative_prompt_embeds = negative_prompt_embeds[0]

        if do_classifier_free_guidance:
            # duplicate unconditional embeddings for each generation per prompt, using mps friendly method
            seq_len = negative_prompt_embeds.shape[1]

            negative_prompt_embeds = negative_prompt_embeds.to(dtype=prompt_embeds_dtype, device=device)

            negative_prompt_embeds = negative_prompt_embeds.repeat(1, num_images_per_prompt, 1)
            negative_prompt_embeds = negative_prompt_embeds.view(batch_size * num_images_per_prompt, seq_len, -1)

        if isinstance(self, LoraLoaderMixin) and USE_PEFT_BACKEND:
            # Retrieve the original scale by scaling back the LoRA layers
            unscale_lora_layers(self.text_encoder, lora_scale)

        return prompt_embeds, negative_prompt_embeds

    # Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.StableDiffusionPipeline.encode_image
    def encode_image(self, image, device, num_images_per_prompt, output_hidden_states=None):
        dtype = next(self.image_encoder.parameters()).dtype

        if not isinstance(image, torch.Tensor):
            image = self.feature_extractor(image, return_tensors="pt").pixel_values

        image = image.to(device=device, dtype=dtype)
        if output_hidden_states:
            image_enc_hidden_states = self.image_encoder(image, output_hidden_states=True).hidden_states[-2]
            image_enc_hidden_states = image_enc_hidden_states.repeat_interleave(num_images_per_prompt, dim=0)
            uncond_image_enc_hidden_states = self.image_encoder(
                torch.zeros_like(image), output_hidden_states=True
            ).hidden_states[-2]
            uncond_image_enc_hidden_states = uncond_image_enc_hidden_states.repeat_interleave(
                num_images_per_prompt, dim=0
            )
            return image_enc_hidden_states, uncond_image_enc_hidden_states
        else:
            image_embeds = self.image_encoder(image).image_embeds
            image_embeds = image_embeds.repeat_interleave(num_images_per_prompt, dim=0)
            uncond_image_embeds = torch.zeros_like(image_embeds)

            return image_embeds, uncond_image_embeds

    # Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.StableDiffusionPipeline.prepare_ip_adapter_image_embeds
    def prepare_ip_adapter_image_embeds(
        self, ip_adapter_image, ip_adapter_image_embeds, device, num_images_per_prompt
    ):
        if ip_adapter_image_embeds is None:
            if not isinstance(ip_adapter_image, list):
                ip_adapter_image = [ip_adapter_image]

            if len(ip_adapter_image) != len(self.unet.encoder_hid_proj.image_projection_layers):
                raise ValueError(
                    f"`ip_adapter_image` must have same length as the number of IP Adapters. Got {len(ip_adapter_image)} images and {len(self.unet.encoder_hid_proj.image_projection_layers)} IP Adapters."
                )

            image_embeds = []
            for single_ip_adapter_image, image_proj_layer in zip(
                ip_adapter_image, self.unet.encoder_hid_proj.image_projection_layers
            ):
                output_hidden_state = not isinstance(image_proj_layer, ImageProjection)
                single_image_embeds, single_negative_image_embeds = self.encode_image(
                    single_ip_adapter_image, device, 1, output_hidden_state
                )
                single_image_embeds = torch.stack([single_image_embeds] * num_images_per_prompt, dim=0)
                single_negative_image_embeds = torch.stack(
                    [single_negative_image_embeds] * num_images_per_prompt, dim=0
                )

                if self.do_classifier_free_guidance:
                    single_image_embeds = torch.cat([single_negative_image_embeds, single_image_embeds])
                    single_image_embeds = single_image_embeds.to(device)

                image_embeds.append(single_image_embeds)
        else:
            image_embeds = ip_adapter_image_embeds
        return image_embeds

    # Copied from diffusers.pipelines.text_to_video_synthesis/pipeline_text_to_video_synth.TextToVideoSDPipeline.decode_latents
    def decode_latents(self, latents):
        latents = 1 / self.vae.config.scaling_factor * latents

        batch_size, channels, num_frames, height, width = latents.shape
        latents = latents.permute(0, 2, 1, 3, 4).reshape(batch_size * num_frames, channels, height, width)

        image = self.vae.decode(latents).sample
        video = (
            image[None, :]
            .reshape(
                (
                    batch_size,
                    num_frames,
                    -1,
                )
                + image.shape[2:]
            )
            .permute(0, 2, 1, 3, 4)
        )
        # we always cast to float32 as this does not cause significant overhead and is compatible with bfloat16
        video = video.float()
        return video

    # Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.StableDiffusionPipeline.prepare_extra_step_kwargs
    def prepare_extra_step_kwargs(self, generator, eta):
        # prepare extra kwargs for the scheduler step, since not all schedulers have the same signature
        # eta (η) is only used with the DDIMScheduler, it will be ignored for other schedulers.
        # eta corresponds to η in DDIM paper: https://arxiv.org/abs/2010.02502
        # and should be between [0, 1]

        accepts_eta = "eta" in set(inspect.signature(self.scheduler.step).parameters.keys())
        extra_step_kwargs = {}
        if accepts_eta:
            extra_step_kwargs["eta"] = eta

        # check if the scheduler accepts generator
        accepts_generator = "generator" in set(inspect.signature(self.scheduler.step).parameters.keys())
        if accepts_generator:
            extra_step_kwargs["generator"] = generator
        return extra_step_kwargs

    # Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.StableDiffusionPipeline.check_inputs
    def check_inputs(
        self,
        prompt,
        height,
        width,
        num_frames,
        callback_steps,
        negative_prompt=None,
        prompt_embeds=None,
        negative_prompt_embeds=None,
        callback_on_step_end_tensor_inputs=None,
        image=None,
        controlnet_conditioning_scale=1.0,
        control_guidance_start=0.0,
        control_guidance_end=1.0,
    ):
        if height % 8 != 0 or width % 8 != 0:
            raise ValueError(f"`height` and `width` have to be divisible by 8 but are {height} and {width}.")

        if callback_steps is not None and (not isinstance(callback_steps, int) or callback_steps <= 0):
            raise ValueError(
                f"`callback_steps` has to be a positive integer but is {callback_steps} of type"
                f" {type(callback_steps)}."
            )
        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]}"
            )

        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 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)}")

        if 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`:"
                f" {negative_prompt_embeds}. Please make sure to only forward one of the two."
            )

        if prompt_embeds is not None and negative_prompt_embeds is not None:
            if prompt_embeds.shape != negative_prompt_embeds.shape:
                raise ValueError(
                    "`prompt_embeds` and `negative_prompt_embeds` must have the same shape when passed directly, but"
                    f" got: `prompt_embeds` {prompt_embeds.shape} != `negative_prompt_embeds`"
                    f" {negative_prompt_embeds.shape}."
                )

        # `prompt` needs more sophisticated handling when there are multiple
        # conditionings.
        if isinstance(self.controlnet, MultiControlNetModel):
            if isinstance(prompt, list):
                logger.warning(
                    f"You have {len(self.controlnet.nets)} ControlNets and you have passed {len(prompt)}"
                    " prompts. The conditionings will be fixed across the prompts."
                )

        # Check `image`
        is_compiled = hasattr(F, "scaled_dot_product_attention") and isinstance(
            self.controlnet, torch._dynamo.eval_frame.OptimizedModule
        )
        if (
            isinstance(self.controlnet, ControlNetModel)
            or is_compiled
            and isinstance(self.controlnet._orig_mod, ControlNetModel)
        ):
            if not isinstance(image, list):
                raise TypeError(f"For single controlnet, `image` must be of type `list` but got {type(image)}")
            if len(image) != num_frames:
                raise ValueError(f"Excepted image to have length {num_frames} but got {len(image)=}")
        elif (
            isinstance(self.controlnet, MultiControlNetModel)
            or is_compiled
            and isinstance(self.controlnet._orig_mod, MultiControlNetModel)
        ):
            if not isinstance(image, list) or not isinstance(image[0], list):
                raise TypeError(f"For multiple controlnets: `image` must be type list of lists but got {type(image)=}")
            if len(image[0]) != num_frames:
                raise ValueError(f"Expected length of image sublist as {num_frames} but got {len(image[0])=}")
            if any(len(img) != len(image[0]) for img in image):
                raise ValueError("All conditioning frame batches for multicontrolnet must be same size")
        else:
            assert False

        # Check `controlnet_conditioning_scale`
        if (
            isinstance(self.controlnet, ControlNetModel)
            or is_compiled
            and isinstance(self.controlnet._orig_mod, ControlNetModel)
        ):
            if not isinstance(controlnet_conditioning_scale, float):
                raise TypeError("For single controlnet: `controlnet_conditioning_scale` must be type `float`.")
        elif (
            isinstance(self.controlnet, MultiControlNetModel)
            or is_compiled
            and isinstance(self.controlnet._orig_mod, MultiControlNetModel)
        ):
            if isinstance(controlnet_conditioning_scale, list):
                if any(isinstance(i, list) for i in controlnet_conditioning_scale):
                    raise ValueError("A single batch of multiple conditionings are supported at the moment.")
            elif isinstance(controlnet_conditioning_scale, list) and len(controlnet_conditioning_scale) != len(
                self.controlnet.nets
            ):
                raise ValueError(
                    "For multiple controlnets: When `controlnet_conditioning_scale` is specified as `list`, it must have"
                    " the same length as the number of controlnets"
                )
        else:
            assert False

        if not isinstance(control_guidance_start, (tuple, list)):
            control_guidance_start = [control_guidance_start]

        if not isinstance(control_guidance_end, (tuple, list)):
            control_guidance_end = [control_guidance_end]

        if len(control_guidance_start) != len(control_guidance_end):
            raise ValueError(
                f"`control_guidance_start` has {len(control_guidance_start)} elements, but `control_guidance_end` has {len(control_guidance_end)} elements. Make sure to provide the same number of elements to each list."
            )

        if isinstance(self.controlnet, MultiControlNetModel):
            if len(control_guidance_start) != len(self.controlnet.nets):
                raise ValueError(
                    f"`control_guidance_start`: {control_guidance_start} has {len(control_guidance_start)} elements but there are {len(self.controlnet.nets)} controlnets available. Make sure to provide {len(self.controlnet.nets)}."
                )

        for start, end in zip(control_guidance_start, control_guidance_end):
            if start >= end:
                raise ValueError(
                    f"control guidance start: {start} cannot be larger or equal to control guidance end: {end}."
                )
            if start < 0.0:
                raise ValueError(f"control guidance start: {start} can't be smaller than 0.")
            if end > 1.0:
                raise ValueError(f"control guidance end: {end} can't be larger than 1.0.")

    # Copied from diffusers.pipelines.controlnet.pipeline_controlnet.StableDiffusionControlNetPipeline.check_image
    def check_image(self, image, prompt, prompt_embeds):
        image_is_pil = isinstance(image, Image.Image)
        image_is_tensor = isinstance(image, torch.Tensor)
        image_is_np = isinstance(image, np.ndarray)
        image_is_pil_list = isinstance(image, list) and isinstance(image[0], Image.Image)
        image_is_tensor_list = isinstance(image, list) and isinstance(image[0], torch.Tensor)
        image_is_np_list = isinstance(image, list) and isinstance(image[0], np.ndarray)

        if (
            not image_is_pil
            and not image_is_tensor
            and not image_is_np
            and not image_is_pil_list
            and not image_is_tensor_list
            and not image_is_np_list
        ):
            raise TypeError(
                f"image must be passed and be one of PIL image, numpy array, torch tensor, list of PIL images, list of numpy arrays or list of torch tensors, but is {type(image)}"
            )

        if image_is_pil:
            image_batch_size = 1
        else:
            image_batch_size = len(image)

        if prompt is not None and isinstance(prompt, str):
            prompt_batch_size = 1
        elif prompt is not None and isinstance(prompt, list):
            prompt_batch_size = len(prompt)
        elif prompt_embeds is not None:
            prompt_batch_size = prompt_embeds.shape[0]

        if image_batch_size != 1 and image_batch_size != prompt_batch_size:
            raise ValueError(
                f"If image batch size is not 1, image batch size must be same as prompt batch size. image batch size: {image_batch_size}, prompt batch size: {prompt_batch_size}"
            )

    # Copied from diffusers.pipelines.text_to_video_synthesis.pipeline_text_to_video_synth.TextToVideoSDPipeline.prepare_latents
    def prepare_latents(
        self, batch_size, num_channels_latents, num_frames, height, width, dtype, device, generator, latents=None
    ):
        shape = (
            batch_size,
            num_channels_latents,
            num_frames,
            height // self.vae_scale_factor,
            width // self.vae_scale_factor,
        )
        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, device=device, dtype=dtype)
        else:
            latents = latents.to(device)

        # scale the initial noise by the standard deviation required by the scheduler
        latents = latents * self.scheduler.init_noise_sigma
        return latents

    # Copied from diffusers.pipelines.controlnet.pipeline_controlnet.StableDiffusionControlNetPipeline.prepare_image
    def prepare_image(
        self,
        image,
        width,
        height,
        batch_size,
        num_images_per_prompt,
        device,
        dtype,
        do_classifier_free_guidance=False,
        guess_mode=False,
    ):
        image = self.control_image_processor.preprocess(image, height=height, width=width).to(dtype=torch.float32)
        image_batch_size = image.shape[0]

        if image_batch_size == 1:
            repeat_by = batch_size
        else:
            # image batch size is the same as prompt batch size
            repeat_by = num_images_per_prompt

        image = image.repeat_interleave(repeat_by, dim=0)

        image = image.to(device=device, dtype=dtype)

        if do_classifier_free_guidance and not guess_mode:
            image = torch.cat([image] * 2)

        return image

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

    @property
    def clip_skip(self):
        return self._clip_skip

    # here `guidance_scale` is defined analog to the guidance weight `w` of equation (2)
    # of the Imagen paper: https://arxiv.org/pdf/2205.11487.pdf . `guidance_scale = 1`
    # corresponds to doing no classifier free guidance.
    @property
    def do_classifier_free_guidance(self):
        return self._guidance_scale > 1

    @property
    def cross_attention_kwargs(self):
        return self._cross_attention_kwargs

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

    @torch.no_grad()
    def __call__(
        self,
        prompt: Union[str, List[str]] = None,
        num_frames: Optional[int] = 16,
        height: Optional[int] = None,
        width: Optional[int] = None,
        num_inference_steps: int = 50,
        guidance_scale: float = 7.5,
        negative_prompt: Optional[Union[str, List[str]]] = None,
        num_videos_per_prompt: Optional[int] = 1,
        eta: float = 0.0,
        generator: Optional[Union[torch.Generator, List[torch.Generator]]] = None,
        latents: Optional[torch.Tensor] = None,
        prompt_embeds: Optional[torch.Tensor] = None,
        negative_prompt_embeds: Optional[torch.Tensor] = None,
        ip_adapter_image: Optional[PipelineImageInput] = None,
        ip_adapter_image_embeds: Optional[PipelineImageInput] = None,
        conditioning_frames: Optional[List[PipelineImageInput]] = None,
        output_type: Optional[str] = "pil",
        return_dict: bool = True,
        cross_attention_kwargs: Optional[Dict[str, Any]] = None,
        controlnet_conditioning_scale: Union[float, List[float]] = 1.0,
        guess_mode: bool = False,
        control_guidance_start: Union[float, List[float]] = 0.0,
        control_guidance_end: Union[float, List[float]] = 1.0,
        clip_skip: Optional[int] = None,
        callback_on_step_end: Optional[Callable[[int, int, Dict], None]] = None,
        callback_on_step_end_tensor_inputs: List[str] = ["latents"],
        **kwargs,
    ):
        r"""
        The call function to the pipeline for generation.

        Args:
            prompt (`str` or `List[str]`, *optional*):
                The prompt or prompts to guide image generation. If not defined, you need to pass `prompt_embeds`.
            height (`int`, *optional*, defaults to `self.unet.config.sample_size * self.vae_scale_factor`):
                The height in pixels of the generated video.
            width (`int`, *optional*, defaults to `self.unet.config.sample_size * self.vae_scale_factor`):
                The width in pixels of the generated video.
            num_frames (`int`, *optional*, defaults to 16):
                The number of video frames that are generated. Defaults to 16 frames which at 8 frames per seconds
                amounts to 2 seconds of video.
            num_inference_steps (`int`, *optional*, defaults to 50):
                The number of denoising steps. More denoising steps usually lead to a higher quality videos at the
                expense of slower inference.
            guidance_scale (`float`, *optional*, defaults to 7.5):
                A higher guidance scale value encourages the model to generate images closely linked to the text
                `prompt` at the expense of lower image quality. Guidance scale is enabled when `guidance_scale > 1`.
            negative_prompt (`str` or `List[str]`, *optional*):
                The prompt or prompts to guide what to not include in image generation. If not defined, you need to
                pass `negative_prompt_embeds` instead. Ignored when not using guidance (`guidance_scale < 1`).
            eta (`float`, *optional*, defaults to 0.0):
                Corresponds to parameter eta (η) from the [DDIM](https://arxiv.org/abs/2010.02502) paper. Only applies
                to the [`~schedulers.DDIMScheduler`], and is ignored in other schedulers.
            generator (`torch.Generator` or `List[torch.Generator]`, *optional*):
                A [`torch.Generator`](https://pytorch.org/docs/stable/generated/torch.Generator.html) to make
                generation deterministic.
            latents (`torch.Tensor`, *optional*):
                Pre-generated noisy latents sampled from a Gaussian distribution, to be used as inputs for video
                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`. Latents should be of shape
                `(batch_size, num_channel, num_frames, height, width)`.
            prompt_embeds (`torch.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 (`torch.Tensor`, *optional*):
                Pre-generated negative text embeddings. Can be used to easily tweak text inputs (prompt weighting). If
                not provided, `negative_prompt_embeds` are generated from the `negative_prompt` input argument.
            ip_adapter_image (`PipelineImageInput`, *optional*):
                Optional image input to work with IP Adapters.
            ip_adapter_image_embeds (`List[torch.Tensor]`, *optional*):
                Pre-generated image embeddings for IP-Adapter. It should be a list of length same as number of IP-adapters.
                Each element should be a tensor of shape `(batch_size, num_images, emb_dim)`. It should contain the negative image embedding
                if `do_classifier_free_guidance` is set to `True`.
                If not provided, embeddings are computed from the `ip_adapter_image` input argument.
            conditioning_frames (`List[PipelineImageInput]`, *optional*):
                The ControlNet input condition to provide guidance to the `unet` for generation. If multiple ControlNets
                are specified, images must be passed as a list such that each element of the list can be correctly
                batched for input to a single ControlNet.
            output_type (`str`, *optional*, defaults to `"pil"`):
                The output format of the generated video. Choose between `torch.Tensor`, `PIL.Image` or
                `np.array`.
            return_dict (`bool`, *optional*, defaults to `True`):
                Whether or not to return a [`~pipelines.text_to_video_synthesis.TextToVideoSDPipelineOutput`] instead
                of a plain tuple.
            cross_attention_kwargs (`dict`, *optional*):
                A kwargs dictionary that if specified is passed along to the [`AttentionProcessor`] as defined in
                [`self.processor`](https://github.com/huggingface/diffusers/blob/main/src/diffusers/models/attention_processor.py).
            controlnet_conditioning_scale (`float` or `List[float]`, *optional*, defaults to 1.0):
                The outputs of the ControlNet are multiplied by `controlnet_conditioning_scale` before they are added
                to the residual in the original `unet`. If multiple ControlNets are specified in `init`, you can set
                the corresponding scale as a list.
            guess_mode (`bool`, *optional*, defaults to `False`):
                The ControlNet encoder tries to recognize the content of the input image even if you remove all
                prompts. A `guidance_scale` value between 3.0 and 5.0 is recommended.
            control_guidance_start (`float` or `List[float]`, *optional*, defaults to 0.0):
                The percentage of total steps at which the ControlNet starts applying.
            control_guidance_end (`float` or `List[float]`, *optional*, defaults to 1.0):
                The percentage of total steps at which the ControlNet stops applying.
            clip_skip (`int`, *optional*):
                Number of layers to be skipped from CLIP while computing the prompt embeddings. A value of 1 means that
                the output of the pre-final layer will be used for computing the prompt embeddings.
            callback_on_step_end (`Callable`, *optional*):
                A function that calls at the end of each denoising steps during the inference. The function is called
                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.

        Examples:

        Returns:
            [`~pipelines.animatediff.pipeline_output.AnimateDiffPipelineOutput`] or `tuple`:
                If `return_dict` is `True`, [`~pipelines.animatediff.pipeline_output.AnimateDiffPipelineOutput`] is
                returned, otherwise a `tuple` is returned where the first element is a list with the generated frames.
        """

        callback = kwargs.pop("callback", None)
        callback_steps = kwargs.pop("callback_steps", None)

        if callback is not None:
            deprecate(
                "callback",
                "1.0.0",
                "Passing `callback` as an input argument to `__call__` is deprecated, consider using `callback_on_step_end`",
            )
        if callback_steps is not None:
            deprecate(
                "callback_steps",
                "1.0.0",
                "Passing `callback_steps` as an input argument to `__call__` is deprecated, consider using `callback_on_step_end`",
            )

        controlnet = self.controlnet._orig_mod if is_compiled_module(self.controlnet) else self.controlnet

        # align format for control guidance
        if not isinstance(control_guidance_start, list) and isinstance(control_guidance_end, list):
            control_guidance_start = len(control_guidance_end) * [control_guidance_start]
        elif not isinstance(control_guidance_end, list) and isinstance(control_guidance_start, list):
            control_guidance_end = len(control_guidance_start) * [control_guidance_end]
        elif not isinstance(control_guidance_start, list) and not isinstance(control_guidance_end, list):
            mult = len(controlnet.nets) if isinstance(controlnet, MultiControlNetModel) else 1
            control_guidance_start, control_guidance_end = (
                mult * [control_guidance_start],
                mult * [control_guidance_end],
            )

        # 0. Default height and width to unet
        height = height or self.unet.config.sample_size * self.vae_scale_factor
        width = width or self.unet.config.sample_size * self.vae_scale_factor

        num_videos_per_prompt = 1

        # 1. Check inputs. Raise error if not correct
        self.check_inputs(
            prompt=prompt,
            height=height,
            width=width,
            num_frames=num_frames,
            callback_steps=callback_steps,
            negative_prompt=negative_prompt,
            callback_on_step_end_tensor_inputs=callback_on_step_end_tensor_inputs,
            prompt_embeds=prompt_embeds,
            negative_prompt_embeds=negative_prompt_embeds,
            image=conditioning_frames,
            controlnet_conditioning_scale=controlnet_conditioning_scale,
            control_guidance_start=control_guidance_start,
            control_guidance_end=control_guidance_end,
        )

        self._guidance_scale = guidance_scale
        self._clip_skip = clip_skip
        self._cross_attention_kwargs = cross_attention_kwargs

        # 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]

        device = self._execution_device

        if isinstance(controlnet, MultiControlNetModel) and isinstance(controlnet_conditioning_scale, float):
            controlnet_conditioning_scale = [controlnet_conditioning_scale] * len(controlnet.nets)

        global_pool_conditions = (
            controlnet.config.global_pool_conditions
            if isinstance(controlnet, ControlNetModel)
            else controlnet.nets[0].config.global_pool_conditions
        )
        guess_mode = guess_mode or global_pool_conditions

        # 3. Encode input prompt
        text_encoder_lora_scale = (
            cross_attention_kwargs.get("scale", None) if cross_attention_kwargs is not None else None
        )
        prompt_embeds, negative_prompt_embeds = self.encode_prompt(
            prompt,
            device,
            num_videos_per_prompt,
            self.do_classifier_free_guidance,
            negative_prompt,
            prompt_embeds=prompt_embeds,
            negative_prompt_embeds=negative_prompt_embeds,
            lora_scale=text_encoder_lora_scale,
            clip_skip=self.clip_skip,
        )
        # For classifier free guidance, we need to do two forward passes.
        # Here we concatenate the unconditional and text embeddings into a single batch
        # to avoid doing two forward passes
        if self.do_classifier_free_guidance:
            prompt_embeds = torch.cat([negative_prompt_embeds, prompt_embeds])

        if ip_adapter_image is not None:
            image_embeds = self.prepare_ip_adapter_image_embeds(
                ip_adapter_image, ip_adapter_image_embeds, device, batch_size * num_videos_per_prompt
            )

        if isinstance(controlnet, ControlNetModel):
            conditioning_frames = self.prepare_image(
                image=conditioning_frames,
                width=width,
                height=height,
                batch_size=batch_size * num_videos_per_prompt * num_frames,
                num_images_per_prompt=num_videos_per_prompt,
                device=device,
                dtype=controlnet.dtype,
                do_classifier_free_guidance=self.do_classifier_free_guidance,
                guess_mode=guess_mode,
            )
        elif isinstance(controlnet, MultiControlNetModel):
            cond_prepared_frames = []
            for frame_ in conditioning_frames:
                prepared_frame = self.prepare_image(
                    image=frame_,
                    width=width,
                    height=height,
                    batch_size=batch_size * num_videos_per_prompt * num_frames,
                    num_images_per_prompt=num_videos_per_prompt,
                    device=device,
                    dtype=controlnet.dtype,
                    do_classifier_free_guidance=self.do_classifier_free_guidance,
                    guess_mode=guess_mode,
                )
                cond_prepared_frames.append(prepared_frame)
            conditioning_frames = cond_prepared_frames
        else:
            assert False

        # 4. Prepare timesteps
        self.scheduler.set_timesteps(num_inference_steps, device=device)
        timesteps = self.scheduler.timesteps
        self._num_timesteps = len(timesteps)

        # 5. Prepare latent variables
        num_channels_latents = self.unet.config.in_channels
        latents = self.prepare_latents(
            batch_size * num_videos_per_prompt,
            num_channels_latents,
            num_frames,
            height,
            width,
            prompt_embeds.dtype,
            device,
            generator,
            latents,
        )

        # 6. Prepare extra step kwargs. TODO: Logic should ideally just be moved out of the pipeline
        extra_step_kwargs = self.prepare_extra_step_kwargs(generator, eta)

        # 7. Add image embeds for IP-Adapter
        added_cond_kwargs = (
            {"image_embeds": image_embeds}
            if ip_adapter_image is not None or ip_adapter_image_embeds is not None
            else None
        )

        # 7.1 Create tensor stating which controlnets to keep
        controlnet_keep = []
        for i in range(len(timesteps)):
            keeps = [
                1.0 - float(i / len(timesteps) < s or (i + 1) / len(timesteps) > e)
                for s, e in zip(control_guidance_start, control_guidance_end)
            ]
            controlnet_keep.append(keeps[0] if isinstance(controlnet, ControlNetModel) else keeps)

        # 8. Denoising loop
        num_warmup_steps = len(timesteps) - num_inference_steps * self.scheduler.order
        with self.progress_bar(total=num_inference_steps) as progress_bar:
            for i, t in enumerate(timesteps):
                # expand the latents if we are doing classifier free guidance
                latent_model_input = torch.cat([latents] * 2) if self.do_classifier_free_guidance else latents
                latent_model_input = self.scheduler.scale_model_input(latent_model_input, t)

                if guess_mode and self.do_classifier_free_guidance:
                    # Infer ControlNet only for the conditional batch.
                    control_model_input = latents
                    control_model_input = self.scheduler.scale_model_input(control_model_input, t)
                    controlnet_prompt_embeds = prompt_embeds.chunk(2)[1]
                else:
                    control_model_input = latent_model_input
                    controlnet_prompt_embeds = prompt_embeds
                controlnet_prompt_embeds = controlnet_prompt_embeds.repeat_interleave(num_frames, dim=0)

                if isinstance(controlnet_keep[i], list):
                    cond_scale = [c * s for c, s in zip(controlnet_conditioning_scale, controlnet_keep[i])]
                else:
                    controlnet_cond_scale = controlnet_conditioning_scale
                    if isinstance(controlnet_cond_scale, list):
                        controlnet_cond_scale = controlnet_cond_scale[0]
                    cond_scale = controlnet_cond_scale * controlnet_keep[i]

                control_model_input = torch.transpose(control_model_input, 1, 2)
                control_model_input = control_model_input.reshape(
                    (-1, control_model_input.shape[2], control_model_input.shape[3], control_model_input.shape[4])
                )

                down_block_res_samples, mid_block_res_sample = self.controlnet(
                    control_model_input,
                    t,
                    encoder_hidden_states=controlnet_prompt_embeds,
                    controlnet_cond=conditioning_frames,
                    conditioning_scale=cond_scale,
                    guess_mode=guess_mode,
                    return_dict=False,
                )

                # predict the noise residual
                noise_pred = self.unet(
                    latent_model_input,
                    t,
                    encoder_hidden_states=prompt_embeds,
                    cross_attention_kwargs=self.cross_attention_kwargs,
                    added_cond_kwargs=added_cond_kwargs,
                    down_block_additional_residuals=down_block_res_samples,
                    mid_block_additional_residual=mid_block_res_sample,
                ).sample

                # perform guidance
                if self.do_classifier_free_guidance:
                    noise_pred_uncond, noise_pred_text = noise_pred.chunk(2)
                    noise_pred = noise_pred_uncond + guidance_scale * (noise_pred_text - noise_pred_uncond)

                # compute the previous noisy sample x_t -> x_t-1
                latents = self.scheduler.step(noise_pred, t, latents, **extra_step_kwargs).prev_sample

                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 callback is not None and i % callback_steps == 0:
                        callback(i, t, latents)

        # 9. Post processing
        if output_type == "latent":
            video = latents
        else:
            video_tensor = self.decode_latents(latents)
            video = tensor2vid(video_tensor, self.image_processor, output_type=output_type)

        # 10. Offload all models
        self.maybe_free_model_hooks()

        if not return_dict:
            return (video,)

        return AnimateDiffPipelineOutput(frames=video)