import gradio as gr import torch import os import spaces import uuid from diffusers import AnimateDiffPipeline, MotionAdapter, EulerDiscreteScheduler from diffusers.utils import export_to_video from huggingface_hub import hf_hub_download from safetensors.torch import load_file from PIL import Image # Constants base = "frankjoshua/toonyou_beta6" repo = "ByteDance/AnimateDiff-Lightning" checkpoints = { "1-Step" : ["animatediff_lightning_1step_diffusers.safetensors", 1], "2-Step" : ["animatediff_lightning_2step_diffusers.safetensors", 2], "4-Step" : ["animatediff_lightning_4step_diffusers.safetensors", 4], "8-Step" : ["animatediff_lightning_8step_diffusers.safetensors", 8], } loaded = None # Ensure model and scheduler are initialized in GPU-enabled function if torch.cuda.is_available(): device = "cuda" dtype = torch.float16 adapter = MotionAdapter().to(device, dtype) pipe = AnimateDiffPipeline.from_pretrained(base, motion_adapter=adapter, torch_dtype=dtype).to(device) pipe.scheduler = EulerDiscreteScheduler.from_config(pipe.scheduler.config, timestep_spacing="trailing", beta_schedule="linear") else: raise NotImplementedError("No GPU detected!") # Function @spaces.GPU(enable_queue=True) def generate_image(prompt, ckpt): global loaded print(prompt, ckpt) checkpoint = checkpoints[ckpt][0] num_inference_steps = checkpoints[ckpt][1] if loaded != num_inference_steps: pipe.unet.load_state_dict(load_file(hf_hub_download(repo, checkpoint), device=device), strict=False) loaded = num_inference_steps output = pipe(prompt=prompt, guidance_scale=1.0, num_inference_steps=num_inference_steps) name = str(uuid.uuid4()).replace("-", "") path = f"/tmp/{name}.mp4" export_to_video(output.frames[0], path, fps=10) return path # Gradio Interface with gr.Blocks(css="style.css") as demo: gr.HTML("