# Copyright 2025 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 torch import torch.nn.functional as F from transformers import CLIPImageProcessor, CLIPTextModel, CLIPTokenizer, CLIPVisionModelWithProjection from ...image_processor import PipelineImageInput from ...loaders import FromSingleFileMixin, IPAdapterMixin, StableDiffusionLoraLoaderMixin, TextualInversionLoaderMixin from ...models import ( AutoencoderKL, ControlNetModel, ImageProjection, MultiControlNetModel, UNet2DConditionModel, UNetMotionModel, ) from ...models.lora import adjust_lora_scale_text_encoder from ...models.unets.unet_motion_model import MotionAdapter from ...schedulers import KarrasDiffusionSchedulers from ...utils import USE_PEFT_BACKEND, is_torch_xla_available, logging, scale_lora_layers, unscale_lora_layers from ...utils.torch_utils import is_compiled_module, randn_tensor from ...video_processor import VideoProcessor from ..free_init_utils import FreeInitMixin from ..free_noise_utils import AnimateDiffFreeNoiseMixin from ..pipeline_utils import DiffusionPipeline, StableDiffusionMixin from .pipeline_output import AnimateDiffPipelineOutput if is_torch_xla_available(): import torch_xla.core.xla_model as xm XLA_AVAILABLE = True else: XLA_AVAILABLE = False logger = logging.get_logger(__name__) # pylint: disable=invalid-name EXAMPLE_DOC_STRING = """ Examples: ```py >>> import torch >>> from diffusers import ( ... AnimateDiffControlNetPipeline, ... AutoencoderKL, ... ControlNetModel, ... MotionAdapter, ... LCMScheduler, ... ) >>> from diffusers.utils import export_to_gif, load_video >>> # Additionally, you will need a preprocess videos before they can be used with the ControlNet >>> # HF maintains just the right package for it: `pip install controlnet_aux` >>> from controlnet_aux.processor import ZoeDetector >>> # Download controlnets from https://huggingface.co/lllyasviel/ControlNet-v1-1 to use .from_single_file >>> # Download Diffusers-format controlnets, such as https://huggingface.co/lllyasviel/sd-controlnet-depth, to use .from_pretrained() >>> controlnet = ControlNetModel.from_single_file("control_v11f1p_sd15_depth.pth", torch_dtype=torch.float16) >>> # We use AnimateLCM for this example but one can use the original motion adapters as well (for example, https://huggingface.co/guoyww/animatediff-motion-adapter-v1-5-3) >>> motion_adapter = MotionAdapter.from_pretrained("wangfuyun/AnimateLCM") >>> vae = AutoencoderKL.from_pretrained("stabilityai/sd-vae-ft-mse", torch_dtype=torch.float16) >>> pipe: AnimateDiffControlNetPipeline = AnimateDiffControlNetPipeline.from_pretrained( ... "SG161222/Realistic_Vision_V5.1_noVAE", ... motion_adapter=motion_adapter, ... controlnet=controlnet, ... vae=vae, ... ).to(device="cuda", dtype=torch.float16) >>> pipe.scheduler = LCMScheduler.from_config(pipe.scheduler.config, beta_schedule="linear") >>> pipe.load_lora_weights( ... "wangfuyun/AnimateLCM", weight_name="AnimateLCM_sd15_t2v_lora.safetensors", adapter_name="lcm-lora" ... ) >>> pipe.set_adapters(["lcm-lora"], [0.8]) >>> depth_detector = ZoeDetector.from_pretrained("lllyasviel/Annotators").to("cuda") >>> video = load_video( ... "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/animatediff-vid2vid-input-1.gif" ... ) >>> conditioning_frames = [] >>> with pipe.progress_bar(total=len(video)) as progress_bar: ... for frame in video: ... conditioning_frames.append(depth_detector(frame)) ... progress_bar.update() >>> prompt = "a panda, playing a guitar, sitting in a pink boat, in the ocean, mountains in background, realistic, high quality" >>> negative_prompt = "bad quality, worst quality" >>> video = pipe( ... prompt=prompt, ... negative_prompt=negative_prompt, ... num_frames=len(video), ... num_inference_steps=10, ... guidance_scale=2.0, ... conditioning_frames=conditioning_frames, ... generator=torch.Generator().manual_seed(42), ... ).frames[0] >>> export_to_gif(video, "animatediff_controlnet.gif", fps=8) ``` """ class AnimateDiffControlNetPipeline( DiffusionPipeline, StableDiffusionMixin, TextualInversionLoaderMixin, IPAdapterMixin, StableDiffusionLoraLoaderMixin, FreeInitMixin, AnimateDiffFreeNoiseMixin, FromSingleFileMixin, ): r""" Pipeline for text-to-video generation with ControlNet guidance. 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.StableDiffusionLoraLoaderMixin.load_lora_weights`] for loading LoRA weights - [`~loaders.StableDiffusionLoraLoaderMixin.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: Union[UNet2DConditionModel, UNetMotionModel], motion_adapter: MotionAdapter, controlnet: Union[ControlNetModel, List[ControlNetModel], Tuple[ControlNetModel], MultiControlNetModel], scheduler: KarrasDiffusionSchedulers, feature_extractor: Optional[CLIPImageProcessor] = None, image_encoder: Optional[CLIPVisionModelWithProjection] = None, ): super().__init__() if isinstance(unet, UNet2DConditionModel): 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) if getattr(self, "vae", None) else 8 self.video_processor = VideoProcessor(vae_scale_factor=self.vae_scale_factor) self.control_video_processor = VideoProcessor( 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, StableDiffusionLoraLoaderMixin): 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 self.text_encoder is not None: if isinstance(self, StableDiffusionLoraLoaderMixin) 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, do_classifier_free_guidance ): image_embeds = [] if do_classifier_free_guidance: negative_image_embeds = [] 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." ) 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 ) image_embeds.append(single_image_embeds[None, :]) if do_classifier_free_guidance: negative_image_embeds.append(single_negative_image_embeds[None, :]) else: for single_image_embeds in ip_adapter_image_embeds: if do_classifier_free_guidance: single_negative_image_embeds, single_image_embeds = single_image_embeds.chunk(2) negative_image_embeds.append(single_negative_image_embeds) image_embeds.append(single_image_embeds) ip_adapter_image_embeds = [] for i, single_image_embeds in enumerate(image_embeds): single_image_embeds = torch.cat([single_image_embeds] * num_images_per_prompt, dim=0) if do_classifier_free_guidance: single_negative_image_embeds = torch.cat([negative_image_embeds[i]] * num_images_per_prompt, dim=0) single_image_embeds = torch.cat([single_negative_image_embeds, single_image_embeds], dim=0) single_image_embeds = single_image_embeds.to(device=device) ip_adapter_image_embeds.append(single_image_embeds) return ip_adapter_image_embeds # Copied from diffusers.pipelines.animatediff.pipeline_animatediff.AnimateDiffPipeline.decode_latents def decode_latents(self, latents, decode_chunk_size: int = 16): 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) video = [] for i in range(0, latents.shape[0], decode_chunk_size): batch_latents = latents[i : i + decode_chunk_size] batch_latents = self.vae.decode(batch_latents).sample video.append(batch_latents) video = torch.cat(video) video = video[None, :].reshape((batch_size, num_frames, -1) + video.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://huggingface.co/papers/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 def check_inputs( self, prompt, height, width, num_frames, negative_prompt=None, prompt_embeds=None, negative_prompt_embeds=None, callback_on_step_end_tensor_inputs=None, video=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_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, list, dict)): raise ValueError(f"`prompt` has to be of type `str`, `list` or `dict` 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(video, list): raise TypeError(f"For single controlnet, `image` must be of type `list` but got {type(video)}") if len(video) != num_frames: raise ValueError(f"Excepted image to have length {num_frames} but got {len(video)=}") elif ( isinstance(self.controlnet, MultiControlNetModel) or is_compiled and isinstance(self.controlnet._orig_mod, MultiControlNetModel) ): if not isinstance(video, list) or not isinstance(video[0], list): raise TypeError(f"For multiple controlnets: `image` must be type list of lists but got {type(video)=}") if len(video[0]) != num_frames: raise ValueError(f"Expected length of image sublist as {num_frames} but got {len(video[0])=}") if any(len(img) != len(video[0]) for img in video): 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.animatediff.pipeline_animatediff.AnimateDiffPipeline.prepare_latents def prepare_latents( self, batch_size, num_channels_latents, num_frames, height, width, dtype, device, generator, latents=None ): # If FreeNoise is enabled, generate latents as described in Equation (7) of [FreeNoise](https://huggingface.co/papers/2310.15169) if self.free_noise_enabled: latents = self._prepare_latents_free_noise( batch_size, num_channels_latents, num_frames, height, width, dtype, device, generator, latents ) 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." ) shape = ( batch_size, num_channels_latents, num_frames, height // self.vae_scale_factor, width // self.vae_scale_factor, ) 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 def prepare_video( self, video, width, height, batch_size, num_videos_per_prompt, device, dtype, do_classifier_free_guidance=False, guess_mode=False, ): video = self.control_video_processor.preprocess_video(video, height=height, width=width).to( dtype=torch.float32 ) video = video.permute(0, 2, 1, 3, 4).flatten(0, 1) video_batch_size = video.shape[0] if video_batch_size == 1: repeat_by = batch_size else: # image batch size is the same as prompt batch size repeat_by = num_videos_per_prompt video = video.repeat_interleave(repeat_by, dim=0) video = video.to(device=device, dtype=dtype) if do_classifier_free_guidance and not guess_mode: video = torch.cat([video] * 2) return video @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://huggingface.co/papers/2205.11487 . `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 @property def interrupt(self): return self._interrupt @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"], decode_chunk_size: int = 16, ): 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://huggingface.co/papers/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. """ 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, 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, video=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 self._interrupt = False # 2. Define call parameters if prompt is not None and isinstance(prompt, (str, dict)): 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 ) if self.free_noise_enabled: prompt_embeds, negative_prompt_embeds = self._encode_prompt_free_noise( prompt=prompt, num_frames=num_frames, device=device, num_videos_per_prompt=num_videos_per_prompt, do_classifier_free_guidance=self.do_classifier_free_guidance, negative_prompt=negative_prompt, prompt_embeds=prompt_embeds, negative_prompt_embeds=negative_prompt_embeds, lora_scale=text_encoder_lora_scale, clip_skip=self.clip_skip, ) else: 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]) prompt_embeds = prompt_embeds.repeat_interleave(repeats=num_frames, dim=0) if ip_adapter_image is not None or ip_adapter_image_embeds 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, self.do_classifier_free_guidance, ) if isinstance(controlnet, ControlNetModel): conditioning_frames = self.prepare_video( video=conditioning_frames, width=width, height=height, batch_size=batch_size * num_videos_per_prompt * num_frames, num_videos_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_videos = [] for frame_ in conditioning_frames: prepared_video = self.prepare_video( video=frame_, width=width, height=height, batch_size=batch_size * num_videos_per_prompt * num_frames, num_videos_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_videos.append(prepared_video) conditioning_frames = cond_prepared_videos else: assert False # 4. Prepare timesteps self.scheduler.set_timesteps(num_inference_steps, device=device) timesteps = self.scheduler.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) num_free_init_iters = self._free_init_num_iters if self.free_init_enabled else 1 for free_init_iter in range(num_free_init_iters): if self.free_init_enabled: latents, timesteps = self._apply_free_init( latents, free_init_iter, num_inference_steps, device, latents.dtype, generator ) self._num_timesteps = len(timesteps) num_warmup_steps = len(timesteps) - num_inference_steps * self.scheduler.order # 8. Denoising loop with self.progress_bar(total=self._num_timesteps) as progress_bar: for i, t in enumerate(timesteps): if self.interrupt: continue # 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 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 XLA_AVAILABLE: xm.mark_step() # 9. Post processing if output_type == "latent": video = latents else: video_tensor = self.decode_latents(latents, decode_chunk_size) video = self.video_processor.postprocess_video(video=video_tensor, output_type=output_type) # 10. Offload all models self.maybe_free_model_hooks() if not return_dict: return (video,) return AnimateDiffPipelineOutput(frames=video)