FateZero-main/test_fatezero.py

import cv2
import numpy as np
import os
from glob import glob
import copy
from typing import Optional, Dict
from tqdm.auto import tqdm
from omegaconf import OmegaConf
import click
import json

import torch
import torch.utils.data
import torch.utils.checkpoint

from accelerate import Accelerator
from accelerate.logging import get_logger
from accelerate.utils import set_seed
from diffusers import (
    AutoencoderKL,
    DDIMScheduler,
)
from diffusers.utils.import_utils import is_xformers_available
from transformers import AutoTokenizer, CLIPTextModel
from einops import rearrange

from video_diffusion.models.unet_3d_condition import UNetPseudo3DConditionModel
from video_diffusion.data.dataset import ImageSequenceDataset
from video_diffusion.common.util import get_time_string, get_function_args
from video_diffusion.common.logger import get_logger_config_path
from video_diffusion.common.image_util import log_train_samples
from video_diffusion.common.instantiate_from_config import instantiate_from_config
from video_diffusion.pipelines.p2p_validation_loop import P2pSampleLogger

def collate_fn(examples):
    batch = {
        "prompt_ids": torch.cat([example["prompt_ids"] for example in examples], dim=0),
        "images": torch.stack([example["images"] for example in examples]),
    }
    return batch

class VideoDataset(torch.utils.data.Dataset):
    def __init__(self, video_path, prompt_ids, frame_stride=1, target_size=(512, 512)):
        self.cap = cv2.VideoCapture(video_path)
        self.frame_stride = frame_stride
        self.target_size = target_size
        self.prompt_ids = prompt_ids
        self.frames = []
        self._preprocess_video()

    def _preprocess_video(self):
        frame_count = 0
        while True:
            ret, frame = self.cap.read()
            if not ret:
                break
            if frame_count % self.frame_stride == 0:
                frame = cv2.cvtColor(frame, cv2.COLOR_BGR2RGB)
                frame = cv2.resize(frame, self.target_size)
                frame = torch.from_numpy(frame).float() / 127.5 - 1.0
                self.frames.append(frame.permute(2, 0, 1))
            frame_count += 1
        self.cap.release()

    def __len__(self):
        return 1

    def __getitem__(self, idx):
        return {
            "prompt_ids": self.prompt_ids[0],
            "images": torch.stack(self.frames).permute(1, 0, 2, 3)
        }

def test(
    config: str,
    pretrained_model_path: str,
    dataset_config: Dict,
    logdir: str = None,
    editing_config: Optional[Dict] = None,
    test_pipeline_config: Optional[Dict] = None,
    gradient_accumulation_steps: int = 1,
    seed: Optional[int] = None,
    mixed_precision: Optional[str] = "fp16",
    batch_size: int = 1,
    model_config: dict={},
    verbose: bool=True,
    **kwargs
):
    args = get_function_args()
    time_string = get_time_string()
    if logdir is None:
        logdir = config.replace('config', 'result').replace('.yml', '').replace('.yaml', '')
    logdir += f"_{time_string}"
    accelerator = Accelerator(
        gradient_accumulation_steps=gradient_accumulation_steps,
        mixed_precision=mixed_precision,
    )
    if accelerator.is_main_process:
        os.makedirs(logdir, exist_ok=True)
        OmegaConf.save(args, os.path.join(logdir, "config.yml"))
    logger = get_logger_config_path(logdir)
    if seed is not None:
        set_seed(seed)
    tokenizer = AutoTokenizer.from_pretrained(
        pretrained_model_path,
        subfolder="tokenizer",
        use_fast=False,
    )
    text_encoder = CLIPTextModel.from_pretrained(
        pretrained_model_path,
        subfolder="text_encoder",
    )
    vae = AutoencoderKL.from_pretrained(
        pretrained_model_path,
        subfolder="vae",
    )
    unet = UNetPseudo3DConditionModel.from_2d_model(
        os.path.join(pretrained_model_path, "unet"), model_config=model_config
    )
    if 'target' not in test_pipeline_config:
        test_pipeline_config['target'] = 'video_diffusion.pipelines.stable_diffusion.SpatioTemporalStableDiffusionPipeline'
    pipeline = instantiate_from_config(
        test_pipeline_config,
        vae=vae,
        text_encoder=text_encoder,
        tokenizer=tokenizer,
        unet=unet,
        scheduler=DDIMScheduler.from_pretrained(
            pretrained_model_path,
            subfolder="scheduler",
        ),
        disk_store=kwargs.get('disk_store', False)
    )
    pipeline.scheduler.set_timesteps(editing_config['num_inference_steps'])
    pipeline.set_progress_bar_config(disable=True)
    pipeline.print_pipeline(logger)
    if is_xformers_available():
        try:
            pipeline.enable_xformers_memory_efficient_attention()
        except Exception as e:
            logger.warning(
                "Could not enable memory efficient attention. Make sure xformers is installed"
                f" correctly and a GPU is available: {e}"
            )
    vae.requires_grad_(False)
    unet.requires_grad_(False)
    text_encoder.requires_grad_(False)
    prompt_ids = tokenizer(
        dataset_config["prompt"],
        truncation=True,
        padding="max_length",
        max_length=tokenizer.model_max_length,
        return_tensors="pt",
    ).input_ids
    video_dataset = VideoDataset(
        video_path=dataset_config["video_path"],
        prompt_ids=prompt_ids,
        frame_stride=dataset_config.get("frame_stride", 1),
        target_size=tuple(dataset_config.get("target_size", [512, 512]))
    )
    train_dataloader = torch.utils.data.DataLoader(
        video_dataset,
        batch_size=batch_size,
        shuffle=True,
        num_workers=4,
        collate_fn=collate_fn,
    )
    train_sample_save_path = os.path.join(logdir, "train_samples.gif")
    log_train_samples(save_path=train_sample_save_path, train_dataloader=train_dataloader)
    unet, train_dataloader = accelerator.prepare(
        unet, train_dataloader
    )
    weight_dtype = torch.float32
    if accelerator.mixed_precision == "fp16":
        weight_dtype = torch.float16
    elif accelerator.mixed_precision == "bf16":
        weight_dtype = torch.bfloat16
    vae.to(accelerator.device, dtype=weight_dtype)
    text_encoder.to(accelerator.device, dtype=weight_dtype)
    if accelerator.is_main_process:
        accelerator.init_trackers("video")
    logger.info("***** wait to fix the logger path *****")
    if editing_config is not None and accelerator.is_main_process:
        validation_sample_logger = P2pSampleLogger(**editing_config, logdir=logdir, source_prompt=dataset_config['prompt'])
    def make_data_yielder(dataloader):
        while True:
            for batch in dataloader:
                yield batch
            accelerator.wait_for_everyone()
    train_data_yielder = make_data_yielder(train_dataloader)
    batch = next(train_data_yielder)
    if editing_config.get('use_invertion_latents', False):
        assert batch["images"].shape[0] == 1, "Only support, overfitting on a single video"
        vae.eval()
        text_encoder.eval()
        unet.eval()
        text_embeddings = pipeline._encode_prompt(
            dataset_config["prompt"],
            device=accelerator.device,
            num_images_per_prompt=1,
            do_classifier_free_guidance=True,
            negative_prompt=None
        )
        use_inversion_attention = editing_config.get('use_inversion_attention', False)
        batch['latents_all_step'] = pipeline.prepare_latents_ddim_inverted(
            rearrange(batch["images"].to(dtype=weight_dtype), "b c f h w -> (b f) c h w"),
            batch_size=1,
            num_images_per_prompt=1,
            text_embeddings=text_embeddings,
            prompt=dataset_config["prompt"],
            store_attention=use_inversion_attention,
            LOW_RESOURCE=True,
            save_path=logdir if verbose else None
        )
        batch['ddim_init_latents'] = batch['latents_all_step'][-1]
    else:
        batch['ddim_init_latents'] = None
    vae.eval()
    text_encoder.eval()
    unet.eval()
    images = batch["images"].to(dtype=weight_dtype)
    images = rearrange(images, "b c f h w -> (b f) c h w")
    if accelerator.is_main_process:
        if validation_sample_logger is not None:
            unet.eval()
            validation_sample_logger.log_sample_images(
                image=images,
                pipeline=pipeline,
                device=accelerator.device,
                step=0,
                latents=batch['ddim_init_latents'],
                save_dir=logdir if verbose else None
            )
    accelerator.end_training()

@click.command()
@click.option("--config", type=str, default="config/sample.yml")
@click.option("--jsonl_path", type=str, required=True, help="Path to the JSONL file containing video information")
def run(config, jsonl_path):
    base_config = OmegaConf.load(config) 
    root_logdir = base_config.get('logdir', 'results')  
    with open(jsonl_path, 'r') as f:
        video_infos = [json.loads(line) for line in f]
    for video_info in video_infos:
        video_path = os.path.join("/home/wangjuntong/video_editing_dataset/all_sourse/", video_info['video'])
        prompt = video_info['prompt']
        editing_prompts = video_info['edit_prompt']
        
        
        Omegadict = copy.deepcopy(base_config)
        Omegadict['dataset_config']['video_path'] = video_path
        Omegadict['dataset_config']['prompt'] = prompt
        Omegadict['editing_config']['editing_prompts'] = [editing_prompts] if isinstance(editing_prompts, str) else editing_prompts
        
      
        video_name = os.path.basename(video_path).split('.')[0]
        logdir = os.path.join(root_logdir, video_name)
        Omegadict['logdir'] = logdir
        
        test(config=config, **Omegadict)

if __name__ == "__main__":
    run()