app.py

# Imports
import gradio as gr
import random
import spaces
import torch
import uuid
import os
from transformers import pipeline

from diffusers import StableDiffusionXLPipeline, ControlNetModel, EulerAncestralDiscreteScheduler
from diffusers.models import AutoencoderKL
from PIL import Image

# Pre-Initialize
DEVICE = "auto"
if DEVICE == "auto":
    DEVICE = "cuda" if torch.cuda.is_available() else "cpu"
print(f"[SYSTEM] | Using {DEVICE} type compute device.")

# Variables
MAX_SEED = 9007199254740991
DEFAULT_INPUT = ""
DEFAULT_NEGATIVE_INPUT = "(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, (exposed, explicit, porn, nude, nudity, naked, nsfw:1.25)"
DEFAULT_MODEL = "Default"
DEFAULT_HEIGHT = 1024
DEFAULT_WIDTH = 1024

css = '''
.gradio-container{max-width: 560px !important}
h1{text-align:center}
footer {
    visibility: hidden
}
'''

repo_nsfw_classifier = pipeline("image-classification", model="Falconsai/nsfw_image_detection")

vae = AutoencoderKL.from_pretrained("madebyollin/sdxl-vae-fp16-fix", torch_dtype=torch.float16)
controlnet = ControlNetModel.from_pretrained("MakiPan/controlnet-encoded-hands-130k", torch_dtype=torch.float16)

repo_default = StableDiffusionXLPipeline.from_pretrained("fluently/Fluently-XL-Final", vae=vae, controlnet=controlnet, torch_dtype=torch.float16, use_safetensors=True, add_watermarker=False)

repo_default.load_lora_weights("ehristoforu/dalle-3-xl-v2", adapter_name="base")
repo_default.set_adapters("base")
# repo_default.set_adapters(["base"], adapter_weights=[0.7])

# repo_pixel = StableDiffusionXLPipeline.from_pretrained("sd-community/sdxl-flash", vae=vae, controlnet=controlnet, torch_dtype=torch.float16, use_safetensors=True, add_watermarker=False)
# repo_pixel.load_lora_weights("artificialguybr/PixelArtRedmond", adapter_name="base")
# repo_pixel.load_lora_weights("nerijs/pixel-art-xl", adapter_name="base2")
# repo_pixel.set_adapters(["base", "base2"], adapter_weights=[1, 1])

repo_customs = {
    "Default": repo_default,
    "Realistic": None, # StableDiffusionXLPipeline.from_pretrained("SG161222/RealVisXL_V4.0", vae=vae, controlnet=controlnet, torch_dtype=torch.float16, use_safetensors=True, add_watermarker=False),
    "Anime": None, # StableDiffusionXLPipeline.from_pretrained("cagliostrolab/animagine-xl-3.1", vae=vae, controlnet=controlnet, torch_dtype=torch.float16, use_safetensors=True, add_watermarker=False),
    "Pixel": None, # repo_pixel,
}

print("-----------A")
print(repo_default.scheduler)
print("-----/-////B")
print(EulerAncestralDiscreteScheduler.from_config(repo_default.scheduler.config))
print("-----------Z")
# repo_default.scheduler = EulerAncestralDiscreteScheduler.from_config(repo_default.scheduler.config)


# Functions
def save_image(img, seed):
    name = f"{seed}-{uuid.uuid4()}.png"
    img.save(name)
    return name
    
def get_seed(seed):
    seed = seed.strip()
    if seed.isdigit():
        return int(seed)
    else:
        return random.randint(0, MAX_SEED)

@spaces.GPU(duration=60)
def generate(input=DEFAULT_INPUT, filter_input="", negative_input=DEFAULT_NEGATIVE_INPUT, model=DEFAULT_MODEL, height=DEFAULT_HEIGHT, width=DEFAULT_WIDTH, steps=1, guidance=0, number=1, seed=None):

    repo = repo_customs[model or "Default"]
    filter_input = filter_input or ""
    negative_input = negative_input or DEFAULT_NEGATIVE_INPUT
    steps_set = steps
    guidance_set = guidance
    seed = get_seed(seed)

    print(input, filter_input, negative_input, model, height, width, steps, guidance, number, seed)
    
    if model == "Realistic":   
        steps_set = 35
        guidance_set = 6
    elif model == "Anime":   
        steps_set = 35
        guidance_set = 6
    elif model == "Pixel":   
        steps_set = 15
        guidance_set = 1.5
    else:
        steps_set = 25
        guidance_set = 6

    if not steps or steps < 0:
        steps = steps_set
    if not guidance or guidance < 0:
        guidance = guidance_set
    
    print(steps, guidance)
    
    repo.to(DEVICE)
    
    parameters  = {
        "prompt": input,
        "negative_prompt": filter_input + negative_input,
        "height": height,
        "width": width,
        "num_inference_steps": steps,
        "guidance_scale": guidance,
        "num_images_per_prompt": number,
        "controlnet_conditioning_scale": 1,
        "cross_attention_kwargs": {"scale": 1},
        "generator": torch.Generator().manual_seed(seed),
        "use_resolution_binning": True,
        "output_type":"pil",
    }
    
    images = repo(**parameters).images
    image_paths = [save_image(img, seed) for img in images]

    print(image_paths)

    nsfw_prediction = repo_nsfw_classifier(Image.open(image_paths[0]))

    print(nsfw_prediction)

    return image_paths, {item['label']: round(item['score'], 3) for item in nsfw_prediction}

def cloud():
    print("[CLOUD] | Space maintained.")


# Initialize
with gr.Blocks(css=css) as main:
    with gr.Column():
        gr.Markdown("🪄 Generate high quality images on all styles between 10 to 20 seconds.")
        
    with gr.Column():
        input = gr.Textbox(lines=1, value=DEFAULT_INPUT, label="Input")
        filter_input = gr.Textbox(lines=1, value="", label="Input Filter")
        negative_input = gr.Textbox(lines=1, value=DEFAULT_NEGATIVE_INPUT, label="Input Negative")
        model = gr.Dropdown(label="Models", choices=repo_customs.keys(), value="Default")
        height = gr.Slider(minimum=8, maximum=2160, step=1, value=DEFAULT_HEIGHT, label="Height")
        width = gr.Slider(minimum=8, maximum=2160, step=1, value=DEFAULT_WIDTH, label="Width")
        steps = gr.Slider(minimum=-1, maximum=100, step=1, value=-1, label="Steps")
        guidance = gr.Slider(minimum=-1, maximum=100, step=0.001, value=-1, label = "Guidance")
        number = gr.Slider(minimum=1, maximum=4, step=1, value=1, label="Number")
        seed = gr.Textbox(lines=1, value="", label="Seed (Blank for random)")
        submit = gr.Button("▶")
        maintain = gr.Button("☁️")

    with gr.Column():
        images = gr.Gallery(columns=1, label="Image")
        nsfw_classifier = gr.Label()
            
    submit.click(generate, inputs=[input, filter_input, negative_input, model, height, width, steps, guidance, number, seed], outputs=[images, nsfw_classifier], queue=False)
    maintain.click(cloud, inputs=[], outputs=[], queue=False)

main.launch(show_api=True)