| import torch | |
| import os | |
| import gradio as gr | |
| from diffusers import DiffusionPipeline, DPMSolverMultistepScheduler | |
| from diffusers.utils import export_to_video | |
| from IPython.display import HTML | |
| from base64 import b64encode | |
| pipe = DiffusionPipeline.from_pretrained("damo-vilab/text-to-video-ms-1.7b", torch_dtype=torch.float16, variant="fp16") | |
| pipe.scheduler = DPMSolverMultistepScheduler.from_config(pipe.scheduler.config) | |
| pipe.enable_model_cpu_offload() | |
| pipe.enable_vae_slicing() | |
| def model(txt, time): | |
| prompt = txt | |
| video_duration_seconds = time | |
| num_frames = video_duration_seconds * 10 | |
| video_frames = pipe(prompt, negative_prompt="low quality", | |
| num_inference_steps=25, num_frames=num_frames).frames | |
| video_path = export_to_video(video_frames) | |
| return video_path | |
| demo = gr.Interface( | |
| fn=model, | |
| inputs=["text", gr.Slider(1, 10, step=1)], | |
| outputs=gr.Video(label="Out",output_width=400, output_height=300) | |
| ) | |
| demo.launch() | |