Spaces:
Running
on
Zero
Running
on
Zero
| import gradio as gr | |
| import spaces | |
| import torch | |
| import numpy as np | |
| import os | |
| import random | |
| from ltx_video.inference import infer, InferenceConfig | |
| from functools import partial | |
| import warnings | |
| warnings.filterwarnings("ignore", category=FutureWarning) | |
| def create( | |
| prompt, | |
| negative_prompt="worst quality, inconsistent motion, blurry, jittery, distorted", | |
| input_image_filepath=None, | |
| input_video_filepath=None, | |
| height_ui=512, | |
| width_ui=704, | |
| duration_ui=2.0, | |
| ui_frames_to_use=16, | |
| seed_ui=42, | |
| randomize_seed=True, | |
| ui_guidance_scale=3.0, | |
| improve_texture_flag=True, | |
| fps=8, | |
| progress=gr.Progress(track_tqdm=True), | |
| mode="text-to-video" | |
| ): | |
| """ | |
| Generate videos using the LTX Video model. | |
| """ | |
| # pick seed | |
| used_seed = seed_ui | |
| output_path = f"output_{mode}_{used_seed}.mp4" | |
| # Validate mode-specific required parameters | |
| if mode == "image-to-video": | |
| if not input_image_filepath: | |
| raise gr.Error(f"input_image_filepath, {input_image_filepath} is required for image-to-video mode") | |
| elif mode == "video-to-video": | |
| if not input_video_filepath: | |
| raise gr.Error(f"input_video_filepath, {input_video_filepath} is required for video-to-video mode") | |
| elif mode == "text-to-video": | |
| # No additional file inputs required for text-to-video | |
| pass | |
| else: | |
| raise gr.Error(f"Invalid mode: {mode}. Must be one of: text-to-video, image-to-video, video-to-video") | |
| config = InferenceConfig( | |
| pipeline_config="configs/ltxv-2b-0.9.6-dev.yaml", | |
| prompt=prompt, | |
| negative_prompt=negative_prompt, | |
| height=height_ui, | |
| width=width_ui, | |
| num_frames=ui_frames_to_use, | |
| seed=used_seed, | |
| output_path=output_path | |
| ) | |
| # attach initial image or video if mode requires | |
| if mode == "image-to-video" and input_image_filepath: | |
| config.input_media_path = input_image_filepath | |
| elif mode == "video-to-video" and input_video_filepath: | |
| config.input_media_path = input_video_filepath | |
| # run inference | |
| infer(config) | |
| return output_path, f"✅ Done! Seed: {used_seed}" | |
| # ---- Gradio Blocks & UI ---- | |
| with gr.Blocks(title="AI Video Converter", theme=gr.themes.Soft()) as demo: | |
| gr.Markdown("# 🎬 AI Video Converter") | |
| gr.Markdown("Convert text, images, and videos into stunning AI-generated videos!") | |
| with gr.Tabs(): | |
| # --- Text to Video --- | |
| with gr.Tab("📝 Text to Video"): | |
| gr.Markdown("### Generate videos from text descriptions") | |
| with gr.Row(): | |
| with gr.Column(): | |
| text_prompt = gr.Textbox( | |
| label="Text Prompt", | |
| placeholder="Describe the video you want to create...", | |
| value="A Nigerian woman dancing on the streets of Lagos, Nigeria", | |
| lines=3 | |
| ) | |
| text_num_frames = gr.Slider(minimum=8, maximum=32, value=16, step=1,label="Number of Frames") | |
| text_fps = gr.Slider(minimum=4, maximum=30, value=8, step=1,label="Frames Per Second") | |
| text_generate_video_btn = gr.Button("Generate Video", variant="primary") | |
| with gr.Column(): | |
| text_output_video = gr.Video(label="Generated Video") | |
| text_status = gr.Textbox(label="Status", interactive=False) | |
| # --- Image to Video --- | |
| with gr.Tab("🖼️ Image to Video"): | |
| gr.Markdown("### Animate images into videos") | |
| with gr.Row(): | |
| with gr.Column(): | |
| image_input = gr.Image(label="Input Image",type="filepath", sources=["upload", "webcam", "clipboard"]) | |
| image_text_prompt = gr.Textbox( | |
| label="Text Prompt", | |
| placeholder="Describe the video you want to create...", | |
| value="The creature from the image starts to move", | |
| lines=3 | |
| ) | |
| image_num_frames = gr.Slider(minimum=8, maximum=50, value=25, step=1,label="Number of Frames") | |
| image_fps = gr.Slider(minimum=4, maximum=30, value=8, step=1,label="Frames Per Second") | |
| image_generate_video_btn = gr.Button("Generate Video", variant="primary") | |
| with gr.Column(): | |
| image_output_video = gr.Video(label="Generated Video") | |
| image_status = gr.Textbox(label="Status", interactive=False) | |
| # --- Video to Video --- | |
| with gr.Tab("🎥 Video to Video"): | |
| gr.Markdown("### Transform videos with AI") | |
| with gr.Row(): | |
| with gr.Column(): | |
| video_input = gr.Video(label="Input Video") | |
| video_prompt = gr.Textbox( | |
| label="Transformation Prompt", | |
| placeholder="Describe how you want to transform the video...", | |
| lines=3 | |
| ) | |
| video_strength = gr.Slider(minimum=0.1, maximum=1.0, value=0.8, step=0.1,label="Transformation Strength") | |
| video_generate_video_btn = gr.Button("Transform Video", variant="primary") | |
| with gr.Column(): | |
| video_output_video = gr.Video(label="Transformed Video") | |
| video_status = gr.Textbox(label="Status", interactive=False) | |
| # --- Inputs --- | |
| tgv_inputs = [text_prompt, gr.State(None), gr.State(None), text_num_frames, text_fps] | |
| igv_inputs = [image_text_prompt, image_input, gr.State(None), image_num_frames, image_fps] | |
| vgv_inputs = [video_prompt, gr.State(None), video_input, video_strength] | |
| # --- Outputs --- | |
| tgv_outputs = [text_output_video, text_status] | |
| igv_outputs = [image_output_video, image_status] | |
| vgv_outputs = [video_output_video, video_status] | |
| # --- Button Logic --- | |
| text_generate_video_btn.click( | |
| fn=partial(create, mode="text-to-video"), | |
| inputs=tgv_inputs, | |
| outputs=tgv_outputs | |
| ) | |
| image_generate_video_btn.click( | |
| fn=partial(create, mode="image-to-video"), | |
| inputs=igv_inputs, | |
| outputs=igv_outputs | |
| ) | |
| video_generate_video_btn.click( | |
| fn=partial(create, mode="video-to-video"), | |
| inputs=vgv_inputs, | |
| outputs=vgv_outputs | |
| ) | |
| if __name__ == "__main__": | |
| demo.launch(debug=True, share=False) |