app.py

# app.py
import os
import io
import base64
from PIL import Image
import gradio as gr
from diffusers import StableDiffusionXLPipeline
import torch

# Check if CUDA is available, otherwise use CPU
device = "cuda" if torch.cuda.is_available() else "cpu"

# Initialize the SDXL pipeline
model_id = "stabilityai/stable-diffusion-xl-base-1.0"
pipe = StableDiffusionXLPipeline.from_pretrained(
    model_id,
    torch_dtype=torch.float16 if device == "cuda" else torch.float32,
    use_safetensors=True,
    variant="fp16" if device == "cuda" else None
)
pipe = pipe.to(device)

# Enable memory efficient attention if running on CUDA
if device == "cuda":
    pipe.enable_attention_slicing()

def generate_image(prompt, negative_prompt="", height=512, width=512, num_inference_steps=30, guidance_scale=7.5):
    """Generate an image from a text prompt"""
    # Validate inputs
    if height % 8 != 0 or width % 8 != 0:
        raise ValueError("Height and width must be divisible by 8")
    
    # Generate the image
    image = pipe(
        prompt=prompt,
        negative_prompt=negative_prompt,
        height=height,
        width=width,
        num_inference_steps=num_inference_steps,
        guidance_scale=guidance_scale,
    ).images[0]
    
    # Convert PIL Image to base64 string
    buffered = io.BytesIO()
    image.save(buffered, format="PNG")
    img_str = base64.b64encode(buffered.getvalue()).decode()
    
    return image, f"data:image/png;base64,{img_str}"

# Define the Gradio interface
with gr.Blocks() as demo:
    gr.Markdown("# Text-to-Image Generator API")
    
    with gr.Row():
        with gr.Column():
            prompt = gr.Textbox(label="Prompt", placeholder="Enter your prompt here...")
            negative_prompt = gr.Textbox(label="Negative Prompt", placeholder="Things you don't want in the image...")
            
            with gr.Row():
                height = gr.Slider(minimum=256, maximum=1024, step=8, value=512, label="Height")
                width = gr.Slider(minimum=256, maximum=1024, step=8, value=512, label="Width")
            
            with gr.Row():
                steps = gr.Slider(minimum=10, maximum=50, step=1, value=30, label="Inference Steps")
                guidance = gr.Slider(minimum=1, maximum=15, step=0.1, value=7.5, label="Guidance Scale")
            
            generate_btn = gr.Button("Generate")
        
        with gr.Column():
            output_image = gr.Image(label="Generated Image")
            output_json = gr.Textbox(label="Image Base64", show_copy_button=True)
    
    generate_btn.click(
        fn=generate_image,
        inputs=[prompt, negative_prompt, height, width, steps, guidance],
        outputs=[output_image, output_json]
    )
    
    gr.Markdown("""
    ## API Usage
    
    You can use this as an API with this curl command:
    ```bash
    curl -X POST "https://your-username-text-to-image-api.hf.space/api/predict" \\
         -H "Content-Type: application/json" \\
         -d '{
               "data": [
                 "A beautiful sunset over mountains",
                 "",
                 512,
                 512,
                 30,
                 7.5
               ]
             }'
    ```
    """)

# Create FastAPI app for direct API usage
from fastapi import FastAPI, HTTPException
from pydantic import BaseModel
from typing import Optional, List, Union
import nest_asyncio

nest_asyncio.apply()
app = FastAPI()

class ImageRequest(BaseModel):
    prompt: str
    negative_prompt: Optional[str] = ""
    height: Optional[int] = 512
    width: Optional[int] = 512
    num_inference_steps: Optional[int] = 30
    guidance_scale: Optional[float] = 7.5

class ImageResponse(BaseModel):
    image_base64: str

@app.post("/generate", response_model=ImageResponse)
async def generate_image_api(request: ImageRequest):
    try:
        _, base64_string = generate_image(
            prompt=request.prompt,
            negative_prompt=request.negative_prompt,
            height=request.height,
            width=request.width,
            num_inference_steps=request.num_inference_steps,
            guidance_scale=request.guidance_scale
        )
        return ImageResponse(image_base64=base64_string)
    except Exception as e:
        raise HTTPException(status_code=500, detail=str(e))

# Mount the FastAPI app to the Gradio app
demo.queue().launch(share=True)