easyanimate/pipeline/pipeline_easyanimate_inpaint.py

# Copyright 2023 PixArt-Alpha Authors and 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 html
import inspect
import re
import copy
import urllib.parse as ul
from dataclasses import dataclass
from typing import Callable, List, Optional, Tuple, Union

import numpy as np
import torch
from diffusers import DiffusionPipeline, ImagePipelineOutput
from diffusers.image_processor import VaeImageProcessor
from diffusers.models import AutoencoderKL
from diffusers.schedulers import DPMSolverMultistepScheduler
from diffusers.utils import (BACKENDS_MAPPING, BaseOutput, deprecate,
                             is_bs4_available, is_ftfy_available, logging,
                             replace_example_docstring)
from diffusers.utils.torch_utils import randn_tensor
from einops import rearrange
from tqdm import tqdm
from transformers import T5EncoderModel, T5Tokenizer

from ..models.transformer3d import Transformer3DModel

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

if is_bs4_available():
    from bs4 import BeautifulSoup

if is_ftfy_available():
    import ftfy


EXAMPLE_DOC_STRING = """
    Examples:
        ```py
        >>> import torch
        >>> from diffusers import EasyAnimatePipeline

        >>> # You can replace the checkpoint id with "PixArt-alpha/PixArt-XL-2-512x512" too.
        >>> pipe = EasyAnimatePipeline.from_pretrained("PixArt-alpha/PixArt-XL-2-1024-MS", torch_dtype=torch.float16)
        >>> # Enable memory optimizations.
        >>> pipe.enable_model_cpu_offload()

        >>> prompt = "A small cactus with a happy face in the Sahara desert."
        >>> image = pipe(prompt).images[0]
        ```
"""
# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion_img2img.retrieve_latents
def retrieve_latents(encoder_output, generator):
    if hasattr(encoder_output, "latent_dist"):
        return encoder_output.latent_dist.sample(generator)
    elif hasattr(encoder_output, "latents"):
        return encoder_output.latents
    else:
        raise AttributeError("Could not access latents of provided encoder_output")
    
@dataclass
class EasyAnimatePipelineOutput(BaseOutput):
    videos: Union[torch.Tensor, np.ndarray]

class EasyAnimateInpaintPipeline(DiffusionPipeline):
    r"""
    Pipeline for text-to-image generation using PixArt-Alpha.

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

    Args:
        vae ([`AutoencoderKL`]):
            Variational Auto-Encoder (VAE) Model to encode and decode images to and from latent representations.
        text_encoder ([`T5EncoderModel`]):
            Frozen text-encoder. PixArt-Alpha uses
            [T5](https://huggingface.co/docs/transformers/model_doc/t5#transformers.T5EncoderModel), specifically the
            [t5-v1_1-xxl](https://huggingface.co/PixArt-alpha/PixArt-alpha/tree/main/t5-v1_1-xxl) variant.
        tokenizer (`T5Tokenizer`):
            Tokenizer of class
            [T5Tokenizer](https://huggingface.co/docs/transformers/model_doc/t5#transformers.T5Tokenizer).
        transformer ([`Transformer3DModel`]):
            A text conditioned `Transformer3DModel` to denoise the encoded image latents.
        scheduler ([`SchedulerMixin`]):
            A scheduler to be used in combination with `transformer` to denoise the encoded image latents.
    """
    bad_punct_regex = re.compile(
        r"[" + "#®•©™&@·º½¾¿¡§~" + "\)" + "\(" + "\]" + "\[" + "\}" + "\{" + "\|" + "\\" + "\/" + "\*" + r"]{1,}"
    )  # noqa

    _optional_components = ["tokenizer", "text_encoder"]
    model_cpu_offload_seq = "text_encoder->transformer->vae"

    def __init__(
        self,
        tokenizer: T5Tokenizer,
        text_encoder: T5EncoderModel,
        vae: AutoencoderKL,
        transformer: Transformer3DModel,
        scheduler: DPMSolverMultistepScheduler,
    ):
        super().__init__()

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

        self.vae_scale_factor = 2 ** (len(self.vae.config.block_out_channels) - 1)
        self.image_processor = VaeImageProcessor(vae_scale_factor=self.vae_scale_factor, do_normalize=True)
        self.mask_processor = VaeImageProcessor(
            vae_scale_factor=self.vae_scale_factor, do_normalize=False, do_binarize=True, do_convert_grayscale=True
        )

    # Adapted from https://github.com/PixArt-alpha/PixArt-alpha/blob/master/diffusion/model/utils.py
    def mask_text_embeddings(self, emb, mask):
        if emb.shape[0] == 1:
            keep_index = mask.sum().item()
            return emb[:, :, :keep_index, :], keep_index
        else:
            masked_feature = emb * mask[:, None, :, None]
            return masked_feature, emb.shape[2]

    # Adapted from diffusers.pipelines.deepfloyd_if.pipeline_if.encode_prompt
    def encode_prompt(
        self,
        prompt: Union[str, List[str]],
        do_classifier_free_guidance: bool = True,
        negative_prompt: str = "",
        num_images_per_prompt: int = 1,
        device: Optional[torch.device] = None,
        prompt_embeds: Optional[torch.FloatTensor] = None,
        negative_prompt_embeds: Optional[torch.FloatTensor] = None,
        prompt_attention_mask: Optional[torch.FloatTensor] = None,
        negative_prompt_attention_mask: Optional[torch.FloatTensor] = None,
        clean_caption: bool = False,
        max_sequence_length: int = 120,
        **kwargs,
    ):
        r"""
        Encodes the prompt into text encoder hidden states.

        Args:
            prompt (`str` or `List[str]`, *optional*):
                prompt to be encoded
            negative_prompt (`str` or `List[str]`, *optional*):
                The prompt 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`). For
                PixArt-Alpha, this should be "".
            do_classifier_free_guidance (`bool`, *optional*, defaults to `True`):
                whether to use classifier free guidance or not
            num_images_per_prompt (`int`, *optional*, defaults to 1):
                number of images that should be generated per prompt
            device: (`torch.device`, *optional*):
                torch device to place the resulting embeddings on
            prompt_embeds (`torch.FloatTensor`, *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.FloatTensor`, *optional*):
                Pre-generated negative text embeddings. For PixArt-Alpha, it's should be the embeddings of the ""
                string.
            clean_caption (`bool`, defaults to `False`):
                If `True`, the function will preprocess and clean the provided caption before encoding.
            max_sequence_length (`int`, defaults to 120): Maximum sequence length to use for the prompt.
        """

        if "mask_feature" in kwargs:
            deprecation_message = "The use of `mask_feature` is deprecated. It is no longer used in any computation and that doesn't affect the end results. It will be removed in a future version."
            deprecate("mask_feature", "1.0.0", deprecation_message, standard_warn=False)

        if device is None:
            device = self._execution_device

        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]

        # See Section 3.1. of the paper.
        max_length = max_sequence_length

        if prompt_embeds is None:
            prompt = self._text_preprocessing(prompt, clean_caption=clean_caption)
            text_inputs = self.tokenizer(
                prompt,
                padding="max_length",
                max_length=max_length,
                truncation=True,
                add_special_tokens=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[:, max_length - 1 : -1])
                logger.warning(
                    "The following part of your input was truncated because CLIP can only handle sequences up to"
                    f" {max_length} tokens: {removed_text}"
                )

            prompt_attention_mask = text_inputs.attention_mask
            prompt_attention_mask = prompt_attention_mask.to(device)

            prompt_embeds = self.text_encoder(text_input_ids.to(device), attention_mask=prompt_attention_mask)
            prompt_embeds = prompt_embeds[0]

        if self.text_encoder is not None:
            dtype = self.text_encoder.dtype
        elif self.transformer is not None:
            dtype = self.transformer.dtype
        else:
            dtype = None

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

        bs_embed, seq_len, _ = prompt_embeds.shape
        # duplicate text embeddings and attention mask 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)
        prompt_attention_mask = prompt_attention_mask.view(bs_embed, -1)
        prompt_attention_mask = prompt_attention_mask.repeat(num_images_per_prompt, 1)

        # get unconditional embeddings for classifier free guidance
        if do_classifier_free_guidance and negative_prompt_embeds is None:
            uncond_tokens = [negative_prompt] * batch_size
            uncond_tokens = self._text_preprocessing(uncond_tokens, clean_caption=clean_caption)
            max_length = prompt_embeds.shape[1]
            uncond_input = self.tokenizer(
                uncond_tokens,
                padding="max_length",
                max_length=max_length,
                truncation=True,
                return_attention_mask=True,
                add_special_tokens=True,
                return_tensors="pt",
            )
            negative_prompt_attention_mask = uncond_input.attention_mask
            negative_prompt_attention_mask = negative_prompt_attention_mask.to(device)

            negative_prompt_embeds = self.text_encoder(
                uncond_input.input_ids.to(device), attention_mask=negative_prompt_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=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)

            negative_prompt_attention_mask = negative_prompt_attention_mask.view(bs_embed, -1)
            negative_prompt_attention_mask = negative_prompt_attention_mask.repeat(num_images_per_prompt, 1)
        else:
            negative_prompt_embeds = None
            negative_prompt_attention_mask = None

        return prompt_embeds, prompt_attention_mask, negative_prompt_embeds, negative_prompt_attention_mask

    # 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

    def check_inputs(
        self,
        prompt,
        height,
        width,
        negative_prompt,
        callback_steps,
        prompt_embeds=None,
        negative_prompt_embeds=None,
    ):
        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 None) or (
            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 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 prompt is not None and negative_prompt_embeds is not None:
            raise ValueError(
                f"Cannot forward both `prompt`: {prompt} and `negative_prompt_embeds`:"
                f" {negative_prompt_embeds}. Please make sure to only forward one of the two."
            )

        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}."
                )

    # Copied from diffusers.pipelines.deepfloyd_if.pipeline_if.IFPipeline._text_preprocessing
    def _text_preprocessing(self, text, clean_caption=False):
        if clean_caption and not is_bs4_available():
            logger.warn(BACKENDS_MAPPING["bs4"][-1].format("Setting `clean_caption=True`"))
            logger.warn("Setting `clean_caption` to False...")
            clean_caption = False

        if clean_caption and not is_ftfy_available():
            logger.warn(BACKENDS_MAPPING["ftfy"][-1].format("Setting `clean_caption=True`"))
            logger.warn("Setting `clean_caption` to False...")
            clean_caption = False

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

        def process(text: str):
            if clean_caption:
                text = self._clean_caption(text)
                text = self._clean_caption(text)
            else:
                text = text.lower().strip()
            return text

        return [process(t) for t in text]

    # Copied from diffusers.pipelines.deepfloyd_if.pipeline_if.IFPipeline._clean_caption
    def _clean_caption(self, caption):
        caption = str(caption)
        caption = ul.unquote_plus(caption)
        caption = caption.strip().lower()
        caption = re.sub("<person>", "person", caption)
        # urls:
        caption = re.sub(
            r"\b((?:https?:(?:\/{1,3}|[a-zA-Z0-9%])|[a-zA-Z0-9.\-]+[.](?:com|co|ru|net|org|edu|gov|it)[\w/-]*\b\/?(?!@)))",  # noqa
            "",
            caption,
        )  # regex for urls
        caption = re.sub(
            r"\b((?:www:(?:\/{1,3}|[a-zA-Z0-9%])|[a-zA-Z0-9.\-]+[.](?:com|co|ru|net|org|edu|gov|it)[\w/-]*\b\/?(?!@)))",  # noqa
            "",
            caption,
        )  # regex for urls
        # html:
        caption = BeautifulSoup(caption, features="html.parser").text

        # @<nickname>
        caption = re.sub(r"@[\w\d]+\b", "", caption)

        # 31C0—31EF CJK Strokes
        # 31F0—31FF Katakana Phonetic Extensions
        # 3200—32FF Enclosed CJK Letters and Months
        # 3300—33FF CJK Compatibility
        # 3400—4DBF CJK Unified Ideographs Extension A
        # 4DC0—4DFF Yijing Hexagram Symbols
        # 4E00—9FFF CJK Unified Ideographs
        caption = re.sub(r"[\u31c0-\u31ef]+", "", caption)
        caption = re.sub(r"[\u31f0-\u31ff]+", "", caption)
        caption = re.sub(r"[\u3200-\u32ff]+", "", caption)
        caption = re.sub(r"[\u3300-\u33ff]+", "", caption)
        caption = re.sub(r"[\u3400-\u4dbf]+", "", caption)
        caption = re.sub(r"[\u4dc0-\u4dff]+", "", caption)
        caption = re.sub(r"[\u4e00-\u9fff]+", "", caption)
        #######################################################

        # все виды тире / all types of dash --> "-"
        caption = re.sub(
            r"[\u002D\u058A\u05BE\u1400\u1806\u2010-\u2015\u2E17\u2E1A\u2E3A\u2E3B\u2E40\u301C\u3030\u30A0\uFE31\uFE32\uFE58\uFE63\uFF0D]+",  # noqa
            "-",
            caption,
        )

        # кавычки к одному стандарту
        caption = re.sub(r"[`´«»“”¨]", '"', caption)
        caption = re.sub(r"[‘’]", "'", caption)

        # &quot;
        caption = re.sub(r"&quot;?", "", caption)
        # &amp
        caption = re.sub(r"&amp", "", caption)

        # ip adresses:
        caption = re.sub(r"\d{1,3}\.\d{1,3}\.\d{1,3}\.\d{1,3}", " ", caption)

        # article ids:
        caption = re.sub(r"\d:\d\d\s+$", "", caption)

        # \n
        caption = re.sub(r"\\n", " ", caption)

        # "#123"
        caption = re.sub(r"#\d{1,3}\b", "", caption)
        # "#12345.."
        caption = re.sub(r"#\d{5,}\b", "", caption)
        # "123456.."
        caption = re.sub(r"\b\d{6,}\b", "", caption)
        # filenames:
        caption = re.sub(r"[\S]+\.(?:png|jpg|jpeg|bmp|webp|eps|pdf|apk|mp4)", "", caption)

        #
        caption = re.sub(r"[\"\']{2,}", r'"', caption)  # """AUSVERKAUFT"""
        caption = re.sub(r"[\.]{2,}", r" ", caption)  # """AUSVERKAUFT"""

        caption = re.sub(self.bad_punct_regex, r" ", caption)  # ***AUSVERKAUFT***, #AUSVERKAUFT
        caption = re.sub(r"\s+\.\s+", r" ", caption)  # " . "

        # this-is-my-cute-cat / this_is_my_cute_cat
        regex2 = re.compile(r"(?:\-|\_)")
        if len(re.findall(regex2, caption)) > 3:
            caption = re.sub(regex2, " ", caption)

        caption = ftfy.fix_text(caption)
        caption = html.unescape(html.unescape(caption))

        caption = re.sub(r"\b[a-zA-Z]{1,3}\d{3,15}\b", "", caption)  # jc6640
        caption = re.sub(r"\b[a-zA-Z]+\d+[a-zA-Z]+\b", "", caption)  # jc6640vc
        caption = re.sub(r"\b\d+[a-zA-Z]+\d+\b", "", caption)  # 6640vc231

        caption = re.sub(r"(worldwide\s+)?(free\s+)?shipping", "", caption)
        caption = re.sub(r"(free\s)?download(\sfree)?", "", caption)
        caption = re.sub(r"\bclick\b\s(?:for|on)\s\w+", "", caption)
        caption = re.sub(r"\b(?:png|jpg|jpeg|bmp|webp|eps|pdf|apk|mp4)(\simage[s]?)?", "", caption)
        caption = re.sub(r"\bpage\s+\d+\b", "", caption)

        caption = re.sub(r"\b\d*[a-zA-Z]+\d+[a-zA-Z]+\d+[a-zA-Z\d]*\b", r" ", caption)  # j2d1a2a...

        caption = re.sub(r"\b\d+\.?\d*[xх×]\d+\.?\d*\b", "", caption)

        caption = re.sub(r"\b\s+\:\s+", r": ", caption)
        caption = re.sub(r"(\D[,\./])\b", r"\1 ", caption)
        caption = re.sub(r"\s+", " ", caption)

        caption.strip()

        caption = re.sub(r"^[\"\']([\w\W]+)[\"\']$", r"\1", caption)
        caption = re.sub(r"^[\'\_,\-\:;]", r"", caption)
        caption = re.sub(r"[\'\_,\-\:\-\+]$", r"", caption)
        caption = re.sub(r"^\.\S+$", "", caption)

        return caption.strip()

    def prepare_latents(
        self, 
        batch_size,
        num_channels_latents,
        height,
        width,
        video_length,
        dtype,
        device,
        generator,
        latents=None,
        video=None,
        timestep=None,
        is_strength_max=True,
        return_noise=False,
        return_video_latents=False,
    ):
        if self.vae.quant_conv.weight.ndim==5:
            shape = (batch_size, num_channels_latents, int(video_length // 5 * 2) if video_length != 1 else 1, height // self.vae_scale_factor, width // self.vae_scale_factor)
        else:
            shape = (batch_size, num_channels_latents, video_length, 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 return_video_latents or (latents is None and not is_strength_max):
            video = video.to(device=device, dtype=dtype)

            if video.shape[1] == 4:
                video_latents = video
            else:
                video_length = video.shape[2]
                video = rearrange(video, "b c f h w -> (b f) c h w")
                video_latents = self._encode_vae_image(image=video, generator=generator)
                video_latents = rearrange(video_latents, "(b f) c h w -> b c f h w", f=video_length)
            video_latents = video_latents.repeat(batch_size // video_latents.shape[0], 1, 1, 1, 1)

        if latents is None:
            rand_device = "cpu" if device.type == "mps" else device

            noise = torch.randn(shape, generator=generator, device=rand_device, dtype=dtype).to(device)
            # if strength is 1. then initialise the latents to noise, else initial to image + noise
            latents = noise if is_strength_max else self.scheduler.add_noise(video_latents, noise, timestep)
        else:
            noise = latents.to(device)
            if latents.shape != shape:
                raise ValueError(f"Unexpected latents shape, got {latents.shape}, expected {shape}")
            latents = latents.to(device)

        # scale the initial noise by the standard deviation required by the scheduler
        latents = latents * self.scheduler.init_noise_sigma
        outputs = (latents,)

        if return_noise:
            outputs += (noise,)

        if return_video_latents:
            outputs += (video_latents,)

        return outputs

    def decode_latents(self, latents):
        video_length = latents.shape[2]
        latents = 1 / 0.18215 * latents
        if self.vae.quant_conv.weight.ndim==5:
            mini_batch_decoder = 2
            # Decoder
            video = []
            for i in range(0, latents.shape[2], mini_batch_decoder):
                with torch.no_grad():
                    start_index = i
                    end_index = i + mini_batch_decoder
                    latents_bs = self.vae.decode(latents[:, :, start_index:end_index, :, :])[0]
                    video.append(latents_bs)

            # Smooth
            mini_batch_encoder = 5
            video = torch.cat(video, 2).cpu()
            for i in range(mini_batch_encoder, video.shape[2], mini_batch_encoder):
                origin_before = copy.deepcopy(video[:, :, i - 1, :, :])
                origin_after = copy.deepcopy(video[:, :, i, :, :])

                video[:, :, i - 1, :, :] = origin_before * 0.75 + origin_after * 0.25
                video[:, :, i, :, :] = origin_before * 0.25 + origin_after * 0.75
            video = video.clamp(-1, 1)
        else:
            latents = rearrange(latents, "b c f h w -> (b f) c h w")
            # video = self.vae.decode(latents).sample
            video = []
            for frame_idx in tqdm(range(latents.shape[0])):
                video.append(self.vae.decode(latents[frame_idx:frame_idx+1]).sample)
            video = torch.cat(video)
            video = rearrange(video, "(b f) c h w -> b c f h w", f=video_length)
        video = (video / 2 + 0.5).clamp(0, 1)
        # we always cast to float32 as this does not cause significant overhead and is compatible with bfloa16
        video = video.cpu().float().numpy()
        return video

    def _encode_vae_image(self, image: torch.Tensor, generator: torch.Generator):
        if isinstance(generator, list):
            image_latents = [
                retrieve_latents(self.vae.encode(image[i : i + 1]), generator=generator[i])
                for i in range(image.shape[0])
            ]
            image_latents = torch.cat(image_latents, dim=0)
        else:
            image_latents = retrieve_latents(self.vae.encode(image), generator=generator)

        image_latents = self.vae.config.scaling_factor * image_latents

        return image_latents

    def prepare_mask_latents(
        self, mask, masked_image, batch_size, height, width, dtype, device, generator, do_classifier_free_guidance
    ):
        # resize the mask to latents shape as we concatenate the mask to the latents
        # we do that before converting to dtype to avoid breaking in case we're using cpu_offload
        # and half precision
        video_length = mask.shape[2]

        mask = mask.to(device=device, dtype=self.vae.dtype)
        if self.vae.quant_conv.weight.ndim==5:
            bs = 1
            new_mask = []
            for i in range(0, mask.shape[0], bs):
                mini_batch = 5
                new_mask_mini_batch = []
                for j in range(0, mask.shape[2], mini_batch):
                    mask_bs = mask[i : i + bs, :, j: j + mini_batch, :, :]
                    mask_bs = self.vae.encode(mask_bs)[0]
                    mask_bs = mask_bs.sample()
                    new_mask_mini_batch.append(mask_bs)
                new_mask_mini_batch = torch.cat(new_mask_mini_batch, dim = 2)
                new_mask.append(new_mask_mini_batch)
            mask = torch.cat(new_mask, dim = 0)
            mask = mask * 0.1825

        else:
            if mask.shape[1] == 4:
                mask = mask
            else:
                video_length = mask.shape[2]
                mask = rearrange(mask, "b c f h w -> (b f) c h w")
                mask = self._encode_vae_image(mask, generator=generator)
                mask = rearrange(mask, "(b f) c h w -> b c f h w", f=video_length)

        masked_image = masked_image.to(device=device, dtype=self.vae.dtype)
        if self.vae.quant_conv.weight.ndim==5:
            bs = 1
            new_mask_pixel_values = []
            for i in range(0, masked_image.shape[0], bs):
                mini_batch = 5
                new_mask_pixel_values_mini_batch = []
                for j in range(0, masked_image.shape[2], mini_batch):
                    mask_pixel_values_bs = masked_image[i : i + bs, :, j: j + mini_batch, :, :]
                    mask_pixel_values_bs = self.vae.encode(mask_pixel_values_bs)[0]
                    mask_pixel_values_bs = mask_pixel_values_bs.sample()
                    new_mask_pixel_values_mini_batch.append(mask_pixel_values_bs)
                new_mask_pixel_values_mini_batch = torch.cat(new_mask_pixel_values_mini_batch, dim = 2)
                new_mask_pixel_values.append(new_mask_pixel_values_mini_batch)
            masked_image_latents = torch.cat(new_mask_pixel_values, dim = 0)
            masked_image_latents = masked_image_latents * 0.1825

        else:
            if masked_image.shape[1] == 4:
                masked_image_latents = masked_image
            else:
                video_length = mask.shape[2]
                masked_image = rearrange(masked_image, "b c f h w -> (b f) c h w")
                masked_image_latents = self._encode_vae_image(masked_image, generator=generator)
                masked_image_latents = rearrange(masked_image_latents, "(b f) c h w -> b c f h w", f=video_length)

        # aligning device to prevent device errors when concating it with the latent model input
        masked_image_latents = masked_image_latents.to(device=device, dtype=dtype)
        return mask, masked_image_latents
    
    @torch.no_grad()
    @replace_example_docstring(EXAMPLE_DOC_STRING)
    def __call__(
        self,
        prompt: Union[str, List[str]] = None,
        video_length: Optional[int] = None,
        video: Union[torch.FloatTensor] = None,
        mask_video: Union[torch.FloatTensor] = None,
        masked_video_latents: Union[torch.FloatTensor] = None,
        negative_prompt: str = "",
        num_inference_steps: int = 20,
        timesteps: List[int] = None,
        guidance_scale: float = 4.5,
        num_images_per_prompt: Optional[int] = 1,
        height: Optional[int] = None,
        width: Optional[int] = None,
        strength: float = 1.0,
        eta: float = 0.0,
        generator: Optional[Union[torch.Generator, List[torch.Generator]]] = None,
        latents: Optional[torch.FloatTensor] = None,
        prompt_embeds: Optional[torch.FloatTensor] = None,
        prompt_attention_mask: Optional[torch.FloatTensor] = None,
        negative_prompt_embeds: Optional[torch.FloatTensor] = None,
        negative_prompt_attention_mask: Optional[torch.FloatTensor] = None,
        output_type: Optional[str] = "latent",
        return_dict: bool = True,
        callback: Optional[Callable[[int, int, torch.FloatTensor], None]] = None,
        callback_steps: int = 1,
        clean_caption: bool = True,
        mask_feature: bool = True,
        max_sequence_length: int = 120
    ) -> Union[EasyAnimatePipelineOutput, Tuple]:
        """
        Function invoked when calling the pipeline for generation.

        Args:
            prompt (`str` or `List[str]`, *optional*):
                The prompt or prompts to guide the image generation. If not defined, one has to pass `prompt_embeds`.
                instead.
            negative_prompt (`str` or `List[str]`, *optional*):
                The prompt or prompts not to guide the image generation. If not defined, one has to pass
                `negative_prompt_embeds` instead. Ignored when not using guidance (i.e., ignored if `guidance_scale` is
                less than `1`).
            num_inference_steps (`int`, *optional*, defaults to 100):
                The number of denoising steps. More denoising steps usually lead to a higher quality image at the
                expense of slower inference.
            timesteps (`List[int]`, *optional*):
                Custom timesteps to use for the denoising process. If not defined, equal spaced `num_inference_steps`
                timesteps are used. Must be in descending order.
            guidance_scale (`float`, *optional*, defaults to 7.0):
                Guidance scale as defined in [Classifier-Free Diffusion Guidance](https://arxiv.org/abs/2207.12598).
                `guidance_scale` is defined as `w` of equation 2. of [Imagen
                Paper](https://arxiv.org/pdf/2205.11487.pdf). Guidance scale is enabled by setting `guidance_scale >
                1`. Higher guidance scale encourages to generate images that are closely linked to the text `prompt`,
                usually at the expense of lower image quality.
            num_images_per_prompt (`int`, *optional*, defaults to 1):
                The number of images to generate per prompt.
            height (`int`, *optional*, defaults to self.unet.config.sample_size):
                The height in pixels of the generated image.
            width (`int`, *optional*, defaults to self.unet.config.sample_size):
                The width in pixels of the generated image.
            eta (`float`, *optional*, defaults to 0.0):
                Corresponds to parameter eta (η) in the DDIM paper: https://arxiv.org/abs/2010.02502. Only applies to
                [`schedulers.DDIMScheduler`], will be ignored for others.
            generator (`torch.Generator` or `List[torch.Generator]`, *optional*):
                One or a list of [torch generator(s)](https://pytorch.org/docs/stable/generated/torch.Generator.html)
                to make generation deterministic.
            latents (`torch.FloatTensor`, *optional*):
                Pre-generated noisy latents, sampled from a Gaussian distribution, to be used as inputs for image
                generation. Can be used to tweak the same generation with different prompts. If not provided, a latents
                tensor will ge generated by sampling using the supplied random `generator`.
            prompt_embeds (`torch.FloatTensor`, *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.FloatTensor`, *optional*):
                Pre-generated negative text embeddings. For PixArt-Alpha this negative prompt should be "". If not
                provided, negative_prompt_embeds will be generated from `negative_prompt` input argument.
            output_type (`str`, *optional*, defaults to `"pil"`):
                The output format of the generate image. Choose between
                [PIL](https://pillow.readthedocs.io/en/stable/): `PIL.Image.Image` or `np.array`.
            return_dict (`bool`, *optional*, defaults to `True`):
                Whether or not to return a [`~pipelines.stable_diffusion.IFPipelineOutput`] instead of a plain tuple.
            callback (`Callable`, *optional*):
                A function that will be called every `callback_steps` steps during inference. The function will be
                called with the following arguments: `callback(step: int, timestep: int, latents: torch.FloatTensor)`.
            callback_steps (`int`, *optional*, defaults to 1):
                The frequency at which the `callback` function will be called. If not specified, the callback will be
                called at every step.
            clean_caption (`bool`, *optional*, defaults to `True`):
                Whether or not to clean the caption before creating embeddings. Requires `beautifulsoup4` and `ftfy` to
                be installed. If the dependencies are not installed, the embeddings will be created from the raw
                prompt.
            mask_feature (`bool` defaults to `True`): If set to `True`, the text embeddings will be masked.

        Examples:

        Returns:
            [`~pipelines.ImagePipelineOutput`] or `tuple`:
                If `return_dict` is `True`, [`~pipelines.ImagePipelineOutput`] is returned, otherwise a `tuple` is
                returned where the first element is a list with the generated images
        """
        # 1. Check inputs. Raise error if not correct
        height = height or self.transformer.config.sample_size * self.vae_scale_factor
        width = width or self.transformer.config.sample_size * self.vae_scale_factor

        # 2. Default height and width to transformer
        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

        # 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.
        do_classifier_free_guidance = guidance_scale > 1.0

        # 3. Encode input prompt
        (
            prompt_embeds,
            prompt_attention_mask,
            negative_prompt_embeds,
            negative_prompt_attention_mask,
        ) = self.encode_prompt(
            prompt,
            do_classifier_free_guidance,
            negative_prompt=negative_prompt,
            num_images_per_prompt=num_images_per_prompt,
            device=device,
            prompt_embeds=prompt_embeds,
            negative_prompt_embeds=negative_prompt_embeds,
            prompt_attention_mask=prompt_attention_mask,
            negative_prompt_attention_mask=negative_prompt_attention_mask,
            clean_caption=clean_caption,
            max_sequence_length=max_sequence_length,
        )
        if do_classifier_free_guidance:
            prompt_embeds = torch.cat([negative_prompt_embeds, prompt_embeds], dim=0)
            prompt_attention_mask = torch.cat([negative_prompt_attention_mask, prompt_attention_mask], dim=0)

        # 4. Prepare timesteps
        self.scheduler.set_timesteps(num_inference_steps, device=device)
        timesteps = self.scheduler.timesteps
        # at which timestep to set the initial noise (n.b. 50% if strength is 0.5)
        latent_timestep = timesteps[:1].repeat(batch_size)
        # create a boolean to check if the strength is set to 1. if so then initialise the latents with pure noise
        is_strength_max = strength == 1.0

        if video is not None:
            video_length = video.shape[2]
            init_video = self.image_processor.preprocess(rearrange(video, "b c f h w -> (b f) c h w"), height=height, width=width) 
            init_video = init_video.to(dtype=torch.float32)
            init_video = rearrange(init_video, "(b f) c h w -> b c f h w", f=video_length)
        else:
            init_video = None

        # Prepare latent variables
        num_channels_latents = self.vae.config.latent_channels
        num_channels_transformer = self.transformer.config.in_channels
        return_image_latents = num_channels_transformer == 4

        # 5. Prepare latents.
        latents_outputs = self.prepare_latents(
            batch_size * num_images_per_prompt,
            num_channels_latents,
            height,
            width,
            video_length,
            prompt_embeds.dtype,
            device,
            generator,
            latents,
            video=init_video,
            timestep=latent_timestep,
            is_strength_max=is_strength_max,
            return_noise=True,
            return_video_latents=return_image_latents,
        )
        if return_image_latents:
            latents, noise, image_latents = latents_outputs
        else:
            latents, noise = latents_outputs
        latents_dtype = latents.dtype

        if mask_video is not None:
            # Prepare mask latent variables
            video_length = video.shape[2]
            mask_condition = self.mask_processor.preprocess(rearrange(mask_video, "b c f h w -> (b f) c h w"), height=height, width=width) 
            mask_condition = mask_condition.to(dtype=torch.float32)
            mask_condition = rearrange(mask_condition, "(b f) c h w -> b c f h w", f=video_length)

            if masked_video_latents is None:
                masked_video = init_video * (mask_condition < 0.5) + torch.ones_like(init_video) * (mask_condition > 0.5) * -1
            else:
                masked_video = masked_video_latents

            mask, masked_video_latents = self.prepare_mask_latents(
                mask_condition,
                masked_video,
                batch_size,
                height,
                width,
                prompt_embeds.dtype,
                device,
                generator,
                do_classifier_free_guidance,
            )
        else:
            mask = torch.zeros_like(latents).to(latents.device, latents.dtype)
            masked_video_latents = torch.zeros_like(latents).to(latents.device, latents.dtype)

        # Check that sizes of mask, masked image and latents match
        if num_channels_transformer == 12:
            # default case for runwayml/stable-diffusion-inpainting
            num_channels_mask = mask.shape[1]
            num_channels_masked_image = masked_video_latents.shape[1]
            if num_channels_latents + num_channels_mask + num_channels_masked_image != self.transformer.config.in_channels:
                raise ValueError(
                    f"Incorrect configuration settings! The config of `pipeline.transformer`: {self.transformer.config} expects"
                    f" {self.transformer.config.in_channels} but received `num_channels_latents`: {num_channels_latents} +"
                    f" `num_channels_mask`: {num_channels_mask} + `num_channels_masked_image`: {num_channels_masked_image}"
                    f" = {num_channels_latents+num_channels_masked_image+num_channels_mask}. Please verify the config of"
                    " `pipeline.transformer` or your `mask_image` or `image` input."
                )
        elif num_channels_transformer == 4:
            raise ValueError(
                f"The transformer {self.transformer.__class__} should have 9 input channels, not {self.transformer.config.in_channels}."
            )
        
        # 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)

        # 6.1 Prepare micro-conditions.
        added_cond_kwargs = {"resolution": None, "aspect_ratio": None}
        if self.transformer.config.sample_size == 128:
            resolution = torch.tensor([height, width]).repeat(batch_size * num_images_per_prompt, 1)
            aspect_ratio = torch.tensor([float(height / width)]).repeat(batch_size * num_images_per_prompt, 1)
            resolution = resolution.to(dtype=prompt_embeds.dtype, device=device)
            aspect_ratio = aspect_ratio.to(dtype=prompt_embeds.dtype, device=device)

            if do_classifier_free_guidance:
                resolution = torch.cat([resolution, resolution], dim=0)
                aspect_ratio = torch.cat([aspect_ratio, aspect_ratio], dim=0)
            
            added_cond_kwargs = {"resolution": resolution, "aspect_ratio": aspect_ratio}

        # 7. Denoising loop
        num_warmup_steps = max(len(timesteps) - num_inference_steps * self.scheduler.order, 0)

        with self.progress_bar(total=num_inference_steps) as progress_bar:
            for i, t in enumerate(timesteps):
                latent_model_input = torch.cat([latents] * 2) if do_classifier_free_guidance else latents
                latent_model_input = self.scheduler.scale_model_input(latent_model_input, t)

                if num_channels_transformer == 12:
                    mask_input = torch.cat([mask] * 2) if do_classifier_free_guidance else mask
                    masked_video_latents_input = (
                        torch.cat([masked_video_latents] * 2) if do_classifier_free_guidance else masked_video_latents
                    )
                    inpaint_latents = torch.cat([mask_input, masked_video_latents_input], dim=1)
    
                current_timestep = t
                if not torch.is_tensor(current_timestep):
                    # TODO: this requires sync between CPU and GPU. So try to pass timesteps as tensors if you can
                    # This would be a good case for the `match` statement (Python 3.10+)
                    is_mps = latent_model_input.device.type == "mps"
                    if isinstance(current_timestep, float):
                        dtype = torch.float32 if is_mps else torch.float64
                    else:
                        dtype = torch.int32 if is_mps else torch.int64
                    current_timestep = torch.tensor([current_timestep], dtype=dtype, device=latent_model_input.device)
                elif len(current_timestep.shape) == 0:
                    current_timestep = current_timestep[None].to(latent_model_input.device)
                # broadcast to batch dimension in a way that's compatible with ONNX/Core ML
                current_timestep = current_timestep.expand(latent_model_input.shape[0])

                # predict noise model_output
                noise_pred = self.transformer(
                    latent_model_input,
                    encoder_hidden_states=prompt_embeds,
                    encoder_attention_mask=prompt_attention_mask,
                    timestep=current_timestep,
                    added_cond_kwargs=added_cond_kwargs,
                    inpaint_latents=inpaint_latents.to(latent_model_input.dtype),
                    return_dict=False,
                )[0]

                # perform guidance
                if 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)

                # learned sigma
                noise_pred = noise_pred.chunk(2, dim=1)[0]

                # compute previous image: x_t -> x_t-1
                latents = self.scheduler.step(noise_pred, t, latents, **extra_step_kwargs, return_dict=False)[0]

                # 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:
                        step_idx = i // getattr(self.scheduler, "order", 1)
                        callback(step_idx, t, latents)

        # Post-processing
        video = self.decode_latents(latents)

        # Convert to tensor
        if output_type == "latent":
            video = torch.from_numpy(video)

        if not return_dict:
            return video

        return EasyAnimatePipelineOutput(videos=video)