text2world_hf.py

import os
import argparse
import torch
from transformers import PreTrainedModel, PretrainedConfig

from .cosmos1.models.diffusion.inference.inference_utils import add_common_arguments, validate_args
from .cosmos1.models.diffusion.inference.world_generation_pipeline import DiffusionText2WorldGenerationPipeline
import .cosmos1.utils.log as log 
import .cosmos1.utils.misc as misc
from .cosmos1.utils.io import read_prompts_from_file, save_video

class DiffusionText2WorldConfig(PretrainedConfig):
    model_type = "DiffusionText2World" 
    def __init__(self, **kwargs):
        super().__init__(**kwargs)
        self.diffusion_transformer_dir = kwargs.get("diffusion_transformer_dir", "Cosmos-1.0-Diffusion-7B-Text2World")
        self.prompt_upsampler_dir = kwargs.get("prompt_upsampler_dir", "Cosmos-1.0-Prompt-Upsampler-12B-Text2World")
        self.word_limit_to_skip_upsampler = kwargs.get("word_limit_to_skip_upsampler", 250)
        self.checkpoint_dir = kwargs.get("checkpoint_dir", "checkpoints")
        self.tokenizer_dir = kwargs.get("tokenizer_dir", "Cosmos-1.0-Tokenizer-CV8x8x8")
        self.video_save_name = kwargs.get("video_save_name", "output")
        self.video_save_folder = kwargs.get("video_save_folder", "outputs/")
        self.prompt = kwargs.get("prompt", None)
        self.batch_input_path = kwargs.get("batch_input_path", None)
        self.negative_prompt = kwargs.get("negative_prompt", None)
        self.num_steps = kwargs.get("num_steps", 35)
        self.guidance = kwargs.get("guidance", 7)
        self.num_video_frames = kwargs.get("num_video_frames", 121)
        self.height = kwargs.get("height", 704)
        self.width = kwargs.get("width", 1280)
        self.fps = kwargs.get("fps", 24)
        self.seed = kwargs.get("seed", 1)
        self.disable_prompt_upsampler = kwargs.get("disable_prompt_upsampler", False)
        self.offload_diffusion_transformer = kwargs.get("offload_diffusion_transformer", False)
        self.offload_tokenizer = kwargs.get("offload_tokenizer", False)
        self.offload_text_encoder_model = kwargs.get("offload_text_encoder_model", False)
        self.offload_prompt_upsampler = kwargs.get("offload_prompt_upsampler", False)
        self.offload_guardrail_models = kwargs.get("offload_guardrail_models", False)


class DiffusionText2World(PreTrainedModel):
    config_class = DiffusionText2WorldConfig

    def __init__(self, config=DiffusionText2WorldConfig()):
        super().__init__(config)
        torch.enable_grad(False)   # TODO: do we need this?
        self.config = config
        inference_type = "text2world"
        config.prompt = 1          # TODO: this is to hack args validation, maybe find a better way 
        validate_args(config, inference_type) 
        del config.prompt
        self.pipeline = DiffusionText2WorldGenerationPipeline(
            inference_type=inference_type,
            checkpoint_dir=config.checkpoint_dir,
            checkpoint_name=config.diffusion_transformer_dir,
            prompt_upsampler_dir=config.prompt_upsampler_dir,
            enable_prompt_upsampler=not config.disable_prompt_upsampler,
            offload_network=config.offload_diffusion_transformer,
            offload_tokenizer=config.offload_tokenizer,
            offload_text_encoder_model=config.offload_text_encoder_model,
            offload_prompt_upsampler=config.offload_prompt_upsampler,
            offload_guardrail_models=config.offload_guardrail_models,
            guidance=config.guidance,
            num_steps=config.num_steps,
            height=config.height,
            width=config.width,
            fps=config.fps,
            num_video_frames=config.num_video_frames,
            seed=config.seed,
        )

    def forward(self, prompt):
        cfg = self.config
        # Handle multiple prompts if prompt file is provided
        if cfg.batch_input_path:
            log.info(f"Reading batch inputs from path: {cfg.batch_input_path}")
            prompts = read_prompts_from_file(cfg.batch_input_path)
        else:
            # Single prompt case
            prompts = [{"prompt": cfg.prompt}]

        os.makedirs(cfg.video_save_folder, exist_ok=True)
        for i, input_dict in enumerate(prompts):
            current_prompt = input_dict.get("prompt", None)
            if current_prompt is None:
                log.critical("Prompt is missing, skipping world generation.")
                continue

            # Generate video
            generated_output = self.pipeline.generate(current_prompt, cfg.negative_prompt, cfg.word_limit_to_skip_upsampler)
            if generated_output is None:
                log.critical("Guardrail blocked text2world generation.")
                continue
            video, prompt = generated_output

            if cfg.batch_input_path:
                video_save_path = os.path.join(cfg.video_save_folder, f"{i}.mp4")
                prompt_save_path = os.path.join(cfg.video_save_folder, f"{i}.txt")
            else:
                video_save_path = os.path.join(cfg.video_save_folder, f"{cfg.video_save_name}.mp4")
                prompt_save_path = os.path.join(cfg.video_save_folder, f"{cfg.video_save_name}.txt")

            # Save video
            save_video(
                video=video,
                fps=cfg.fps,
                H=cfg.height,
                W=cfg.width,
                video_save_quality=5,
                video_save_path=video_save_path,
            )

            # Save prompt to text file alongside video
            with open(prompt_save_path, "wb") as f:
                f.write(prompt.encode("utf-8"))

            log.info(f"Saved video to {video_save_path}")
            log.info(f"Saved prompt to {prompt_save_path}")
    
    def save_pretrained(self, save_directory, **kwargs): 
        # We don't save anything
        pass

    @classmethod
    def from_pretrained(cls, pretrained_model_name_or_path, *model_args, **kwargs):
        config = kwargs["config"]
        model = cls(config)
        return model