app.py

import gradio as gr
from diffusers import StableDiffusionPipeline, DPMSolverMultistepScheduler
import torch
from moviepy import ImageSequenceClip
import numpy as np
import random

# Load model
device = "cuda" if torch.cuda.is_available() else "cpu"
pipe = StableDiffusionPipeline.from_pretrained(
    "runwayml/stable-diffusion-v1-5",
    variant="fp16"
).to(device)
pipe.scheduler = DPMSolverMultistepScheduler.from_config(pipe.scheduler.config)

def generate_frames(prompt, num_frames, guidance_scale, num_inference_steps, seed):
    frames = []
    seed = seed if seed is not None else random.randint(0, 1000000)
    generator = torch.Generator(device=device).manual_seed(seed)

    with torch.inference_mode():
        for i in range(num_frames):
            # Variasikan prompt sedikit untuk setiap frame
            prompt_with_variation = f"{prompt} --v {i}"  
            image = pipe(prompt_with_variation, 
                         num_images=1, 
                         guidance_scale=guidance_scale, 
                         num_inference_steps=num_inference_steps, 
                         generator=generator).images[0]
            frames.append(np.array(image))
    return frames

def create_video(frames, output_filename="output.mp4", fps=24):
    clip = ImageSequenceClip(frames, fps=fps)
    clip.write_videofile(output_filename, codec='libx264', audio=False, bitrate="5000k")
    return output_filename

def process_video(prompt, num_frames, guidance_scale, num_inference_steps, seed):
    frames = generate_frames(prompt, num_frames, guidance_scale, num_inference_steps, seed)
    output_filename = create_video(frames)
    return gr.File(output_filename)

iface = gr.Interface(
    fn=process_video,
    inputs=[
        gr.Textbox(lines=1, placeholder="Masukkan prompt Anda di sini...", label="Prompt"),
        gr.Slider(minimum=5, maximum=30, step=1, value=10, label="Jumlah Frame"),
        gr.Slider(minimum=1, maximum=20, step=0.5, value=7.5, label="Guidance Scale"),
        gr.Slider(minimum=10, maximum=50, step=5, value=20, label="Inference Steps"),
        gr.Number(label="Seed (opsional)", value=None),
    ],
    outputs=gr.File(label="Video Output"),
    title="Stable Diffusion Video Generator",
    description="Buat video pendek dari text prompt menggunakan Stable Diffusion!",
)

iface.launch(share=True)