app.py

import os
import random
import uuid
import json

import gradio as gr
import numpy as np
from PIL import Image
import spaces
import torch
from diffusers import DiffusionPipeline

DESCRIPTION = """# SDXL Flash"""
if not torch.cuda.is_available():
    DESCRIPTION += "\n<p>Running on CPU 🥶 This demo may not work on CPU.</p>"

MAX_SEED = np.iinfo(np.int32).max
CACHE_EXAMPLES = torch.cuda.is_available() and os.getenv("CACHE_EXAMPLES", "0") == "1"
MAX_IMAGE_SIZE = int(os.getenv("MAX_IMAGE_SIZE", "4096"))
USE_TORCH_COMPILE = os.getenv("USE_TORCH_COMPILE", "0") == "1"
ENABLE_CPU_OFFLOAD = os.getenv("ENABLE_CPU_OFFLOAD", "0") == "1"

device = torch.device("cuda:0" if torch.cuda.is_available() else "cpu")

NUM_IMAGES_PER_PROMPT = 1

if torch.cuda.is_available():
    pipe = DiffusionPipeline.from_pretrained(
        "sd-community/sdxl-flash",
        torch_dtype=torch.float16,
        use_safetensors=True,
        add_watermarker=False,
        variant="fp16"
    )
    if ENABLE_CPU_OFFLOAD:
        pipe.enable_model_cpu_offload()
    else:
        pipe.to(device)       
        print("Loaded on Device!")
    
    if USE_TORCH_COMPILE:
        pipe.unet = torch.compile(pipe.unet, mode="reduce-overhead", fullgraph=True)
        pipe2.unet = torch.compile(pipe2.unet, mode="reduce-overhead", fullgraph=True)
        print("Model Compiled!")


def save_image(img):
    unique_name = str(uuid.uuid4()) + ".png"
    img.save(unique_name)
    return unique_name


def randomize_seed_fn(seed: int, randomize_seed: bool) -> int:
    if randomize_seed:
        seed = random.randint(0, MAX_SEED)
    return seed

@spaces.GPU(enable_queue=False)
def generate(
    prompt: str,
    negative_prompt: str = "",
    use_negative_prompt: bool = False,
    seed: int = 0,
    width: int = 1024,
    height: int = 1024,
    guidance_scale: float = 3,
    num_inference_steps: int = 9,
    randomize_seed: bool = False,
    use_resolution_binning: bool = True,
    progress=gr.Progress(track_tqdm=True),
):
    pipe.to(device)
    seed = int(randomize_seed_fn(seed, randomize_seed))
    generator = torch.Generator().manual_seed(seed)

    if not use_negative_prompt:
        negative_prompt = ""  # type: ignore
    negative_prompt += default_negative    

    options = {
        "prompt":prompt,
        "negative_prompt":negative_prompt,
        "width":width,
        "height":height,
        "guidance_scale":guidance_scale,
        "num_inference_steps":num_inference_steps,
        "generator":generator,
        "num_images_per_prompt":NUM_IMAGES_PER_PROMPT,
        "use_resolution_binning":use_resolution_binning,
        "output_type":"pil",

    }
    
    images = pipe(**options).images

    image_paths = [save_image(img) for img in images]
    return image_paths, seed


examples = [
    "neon holography crystal cat",
    "a cat eating a piece of cheese",
    "an astronaut riding a horse in space",
    "a cartoon of a boy playing with a tiger",
    "a cute robot artist painting on an easel, concept art",
    #"a close up of a woman wearing a transparent, prismatic, elaborate nemeses headdress, over the should pose, brown skin-tone"
]

css = '''
.gradio-container{max-width: 560px !important}
h1{text-align:center}
'''
with gr.Blocks(css=css) as demo:
    gr.Markdown(DESCRIPTION)
    with gr.Group():
        with gr.Row():
            prompt = gr.Text(
                label="Prompt",
                show_label=False,
                max_lines=1,
                placeholder="Enter your prompt",
                container=False,
            )
            run_button = gr.Button("Run", scale=0)
        result = gr.Gallery(label="Result", columns=NUM_IMAGES_PER_PROMPT, show_label=False)
    with gr.Accordion("Advanced options", open=False):
        with gr.Row():
            use_negative_prompt = gr.Checkbox(label="Use negative prompt", value=True)
            negative_prompt = gr.Text(
                label="Negative prompt",
                max_lines=1,
                placeholder="Enter a negative prompt",
                value="(deformed, distorted, disfigured:1.3), poorly drawn, bad anatomy, wrong anatomy, extra limb, missing limb, floating limbs, (mutated hands and fingers:1.4), disconnected limbs, mutation, mutated, ugly, disgusting, blurry, amputation, NSFW",
                visible=True,
            )
        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(visible=True):
            width = gr.Slider(
                label="Width",
                minimum=256,
                maximum=MAX_IMAGE_SIZE,
                step=32,
                value=1024,
            )
            height = gr.Slider(
                label="Height",
                minimum=256,
                maximum=MAX_IMAGE_SIZE,
                step=32,
                value=1024,
            )
        with gr.Row():
            guidance_scale = gr.Slider(
                label="Guidance Scale",
                minimum=0.1,
                maximum=6,
                step=0.1,
                value=3.0,
            )
            num_inference_steps = gr.Slider(
                label="Number of inference steps",
                minimum=1,
                maximum=15,
                step=1,
                value=5,
            )

    gr.Examples(
        examples=examples,
        inputs=prompt,
        outputs=[result, seed],
        fn=generate,
        cache_examples=CACHE_EXAMPLES,
    )

    use_negative_prompt.change(
        fn=lambda x: gr.update(visible=x),
        inputs=use_negative_prompt,
        outputs=negative_prompt,
        api_name=False,
    )

    gr.on(
        triggers=[
            prompt.submit,
            negative_prompt.submit,
            run_button.click,
        ],
        fn=generate,
        inputs=[
            prompt,
            negative_prompt,
            use_negative_prompt,
            seed,
            width,
            height,
            guidance_scale,
            num_inference_steps,
            randomize_seed,
        ],
        outputs=[result, seed],
        api_name="run",
    )

if __name__ == "__main__":
    demo.queue(max_size=20).launch()