app.py

import os
import gradio as gr
import numpy as np
import random
import spaces  # ZeroGPU integration
from diffusers import DiffusionPipeline
import torch

# Get Hugging Face token from environment variable
HF_TOKEN = os.environ.get("HF_TOKEN") if os.environ.get("HF_TOKEN") else None
if not HF_TOKEN:
    raise ValueError("Hugging Face token not found. Please set the 'HF_TOKEN' environment variable.")

device = "cuda" if torch.cuda.is_available() else "cpu"
model_repo_id = "CompVis/stable-diffusion-v1-4"  # Replace with the model you would like to use

if torch.cuda.is_available():
    torch_dtype = torch.float16
else:
    torch_dtype = torch.float32

pipe = DiffusionPipeline.from_pretrained(
    model_repo_id, torch_dtype=torch_dtype, token=HF_TOKEN
)
pipe = pipe.to(device)


MAX_SEED = np.iinfo(np.int32).max
MAX_IMAGE_SIZE = 1024


@spaces.GPU  # ZeroGPU decorator
def infer(
    prompt,
    negative_prompt,
    seed,
    randomize_seed,
    width,
    height,
    guidance_scale,
    num_inference_steps,
    progress=gr.Progress(track_tqdm=True),
):
    # Seed Handling
    if randomize_seed:
        seed = random.randint(0, MAX_SEED)

    generator = torch.Generator().manual_seed(seed)

    # Generate Image
    image = pipe(
        prompt=prompt,
        negative_prompt=negative_prompt,
        guidance_scale=guidance_scale,
        num_inference_steps=num_inference_steps,
        width=width,
        height=height,
        generator=generator,
    ).images[0]

    return image, seed


examples = [
    "Astronaut in a jungle, cold color palette, muted colors, detailed, 8k",
    "An astronaut riding a green horse",
    "A delicious ceviche cheesecake slice",
]

css = """
/* CSS Styling (remains unchanged from earlier examples) */
"""

# Higher Defaults for Advanced Settings
DEFAULT_STEPS = 90
DEFAULT_GUIDANCE = 10

with gr.Blocks(css=css) as demo:
    with gr.Column(elem_id="col-container"):
        gr.Markdown("<div id='header'><h1 id='title'>Veshon: Veshup's Image Generation AI</h1><p id='subtitle'>Create stunning images with just a prompt. Powered by cutting-edge AI technology.</p></div>")

        with gr.Row():
            prompt = gr.Text(
                label="Your Creative Prompt",
                show_label=False,
                max_lines=1,
                placeholder="Enter your prompt here...",
                container=False,
            )

            run_button = gr.Button("Generate Image", scale=0, variant="primary", elem_classes="gradio-button")

        result = gr.Image(label="Generated Image", show_label=False)

        with gr.Accordion("Advanced Settings", open=False):
            negative_prompt = gr.Text(
                label="Negative Prompt",
                max_lines=1,
                placeholder="Enter a negative prompt if needed",
                visible=False,
            )

            seed = gr.Slider(
                label="Seed",
                minimum=0,
                maximum=MAX_SEED,
                step=1,
                value=0,
            )

            randomize_seed = gr.Checkbox(label="Randomize Seed", value=True)

            with gr.Row():
                width = gr.Slider(
                    label="Width",
                    minimum=256,
                    maximum=MAX_IMAGE_SIZE,
                    step=32,
                    value=720,  # Higher default resolution
                )

                height = gr.Slider(
                    label="Height",
                    minimum=256,
                    maximum=MAX_IMAGE_SIZE,
                    step=32,
                    value=720,  # Higher default resolution
                )

            with gr.Row():
                guidance_scale = gr.Slider(
                    label="Guidance Scale",
                    minimum=0.0,
                    maximum=15.0,
                    step=0.1,
                    value=DEFAULT_GUIDANCE,  # Higher guidance by default
                )

                num_inference_steps = gr.Slider(
                    label="Number of Inference Steps",
                    minimum=1,
                    maximum=150,  # Increased maximum steps
                    step=1,
                    value=DEFAULT_STEPS,  # Higher inference steps for quality
                )

        gr.Examples(examples=examples, inputs=[prompt])
    gr.on(
        triggers=[run_button.click, prompt.submit],
        fn=infer,
        inputs=[
            prompt,
            negative_prompt,
            seed,
            randomize_seed,
            width,
            height,
            guidance_scale,
            num_inference_steps,
        ],
        outputs=[result, seed],
    )

if __name__ == "__main__":
    demo.launch()