app.py

import argparse
import os
import time
from os import path
from safetensors.torch import load_file
import huggingface_hub
from huggingface_hub import hf_hub_download


cache_path = path.join(path.dirname(path.abspath(__file__)), "models")
os.environ["TRANSFORMERS_CACHE"] = cache_path
os.environ["HF_HUB_CACHE"] = cache_path
os.environ["HF_HOME"] = cache_path

import gradio as gr
import torch
from diffusers import FluxPipeline

torch.backends.cuda.matmul.allow_tf32 = True
huggingface_hub.login(os.getenv('HF_TOKEN'))
class timer:
    def __init__(self, method_name="timed process"):
        self.method = method_name
    def __enter__(self):
        self.start = time.time()
        print(f"{self.method} starts")
    def __exit__(self, exc_type, exc_val, exc_tb):
        end = time.time()
        print(f"{self.method} took {str(round(end - self.start, 2))}s")

if not path.exists(cache_path):
    os.makedirs(cache_path, exist_ok=True)

def load_and_fuse_lora_weights(pipe, lora_models):
    for repo, file_path, lora_scale in lora_models:
        lora_weights_path = hf_hub_download(repo_id=repo, filename=file_path)
        pipe.load_lora_weights(lora_weights_path)
        pipe.fuse_lora(lora_scale=lora_scale)

# List of LoRA models and their corresponding scales
lora_models = [
    ("mrcuddle/Character_Design_Helper", "CharacterDesign-FluxV2.safetensors", 0.125),
    ("mrcuddle/live2d-model-maker", "LIVE2D-FLUX.safetensors", 0.125)
]

pipe = FluxPipeline.from_pretrained("black-forest-labs/FLUX.1-dev", torch_dtype=torch.bfloat16)

# Load and fuse LoRA weights
load_and_fuse_lora_weights(pipe, lora_models)

pipe.to(device="cuda", dtype=torch.bfloat16)

with gr.Blocks(theme=gr.themes.Soft()) as demo:
    gr.Markdown(
        """
        <div style="text-align: center; max-width: 650px; margin: 0 auto;">
            <h1 style="font-size: 2.5rem; font-weight: 700; margin-bottom: 1rem; display: contents;">Hyper-FLUX-8steps-LoRA</h1>
            <p style="font-size: 1rem; margin-bottom: 1.5rem;">AutoML team from ByteDance</p>
        </div>
        """
    )

    with gr.Row():
        with gr.Column(scale=3):
            with gr.Group():
                prompt = gr.Textbox(
                    label="Your Image Description",
                    placeholder="E.g., A serene landscape with mountains and a lake at sunset",
                    lines=3
                )

                # Hidden textbox for the preset prompt
                preset_prompt = gr.Textbox(
                    label="Preset Prompt",
                    value="live2d,guijiaoxiansheng,separate hand,separate feet,separate head,multiple views,white background,CharacterDisgnFlux,magic particles, multiple references,color pallete reference,simple background,upper body,front,from side",
                    visible=False
                )

                with gr.Accordion("Advanced Settings", open=False):
                    with gr.Group():
                        with gr.Row():
                            height = gr.Slider(label="Height", minimum=256, maximum=1152, step=64, value=1024)
                            width = gr.Slider(label="Width", minimum=256, maximum=1152, step=64, value=1024)

                        with gr.Row():
                            steps = gr.Slider(label="Inference Steps", minimum=5, maximum=25, step=1, value=8)
                            scales = gr.Slider(label="Guidance Scale", minimum=0.0, maximum=10.0, step=1, value=3.5)

                        seed = gr.Number(label="Seed (for reproducibility)", value=-1, precision=0)

                generate_btn = gr.Button("Generate Image", variant="primary", scale=1)

        with gr.Column(scale=4):
            output = gr.Image(label="Your Generated Image")


    def process_image(height, width, steps, scales, prompt, seed, preset_prompt):
        global pipe
        with torch.inference_mode(), torch.autocast("cuda", dtype=torch.bfloat16), timer("inference"):
            # Concatenate the preset prompt with the user's input prompt
            full_prompt = f"{preset_prompt} {prompt}"
            return pipe(
                prompt=[full_prompt],
                generator=torch.Generator().manual_seed(int(seed)),
                num_inference_steps=int(steps),
                guidance_scale=float(scales),
                height=int(height),
                width=int(width),
                max_sequence_length=256
            ).images[0]

    generate_btn.click(
        process_image,
        inputs=[height, width, steps, scales, prompt, seed, preset_prompt],
        outputs=output
    )

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