app.py

import gradio as gr
import torch
from diffusers import StableDiffusionImg2ImgPipeline
from PIL import Image
import numpy as np

# Load pre-trained model (no fine-tuning required)
pipe = StableDiffusionImg2ImgPipeline.from_pretrained(
    "runwayml/stable-diffusion-v1-5",
    torch_dtype=torch.float16
).to("cuda")

def generate_thumbnail(prompt, input_image):
    if input_image is None or prompt.strip() == "":
        return None

    # Convert image to 512x512
    input_image = input_image.resize((512, 512))
    
    result = pipe(prompt=prompt, image=input_image, strength=0.75, guidance_scale=7.5).images[0]
    return result

gr.Interface(
    fn=generate_thumbnail,
    inputs=[
        gr.Textbox(label="Prompt (e.g. 'NOOB vs PRO Minecraft epic battle')"),
        gr.Image(label="Background Image", type="pil")
    ],
    outputs=gr.Image(label="Generated Thumbnail"),
    title="🖼️ AI Thumbnail Generator (Stable Diffusion)",
    description="Upload a background + write your prompt to generate a Minecraft-style thumbnail using Stable Diffusion v1.5"
).launch()