app.py

import os
import numpy as np
import random
from pathlib import Path
from PIL import Image
import streamlit as st
from huggingface_hub import InferenceClient, AsyncInferenceClient
from gradio_client import Client, handle_file
import asyncio

MAX_SEED = np.iinfo(np.int32).max
HF_TOKEN = os.environ.get("HF_TOKEN")
HF_TOKEN_UPSCALER = os.environ.get("HF_TOKEN_UPSCALER")
client = AsyncInferenceClient()
llm_client = InferenceClient("mistralai/Mixtral-8x7B-Instruct-v0.1")
DATA_PATH = Path("./data")
DATA_PATH.mkdir(exist_ok=True)

def enable_lora(lora_add, basemodel):
    return lora_add if lora_add else basemodel

async def generate_image(combined_prompt, model, width, height, scales, steps, seed):
    try:
        if seed == -1:
            seed = random.randint(0, MAX_SEED)
        seed = int(seed)
        image = await client.text_to_image(
            prompt=combined_prompt, height=height, width=width, guidance_scale=scales,
            num_inference_steps=steps, model=model
        )
        return image, seed
    except Exception as e:
        return f"Error al generar imagen: {e}", None

def get_upscale_finegrain(prompt, img_path, upscale_factor):
    try:
        client = Client("finegrain/finegrain-image-enhancer", hf_token=HF_TOKEN_UPSCALER)
        result = client.predict(
            input_image=handle_file(img_path), prompt=prompt, negative_prompt="",
            seed=42, upscale_factor=upscale_factor, controlnet_scale=0.6,
            controlnet_decay=1, condition_scale=6, tile_width=112,
            tile_height=144, denoise_strength=0.35, num_inference_steps=18,
            solver="DDIM", api_name="/process"
        )
        return result[1] if isinstance(result, list) and len(result) > 1 else None
    except Exception as e:
        return None

async def gen(prompt, basemodel, width, height, scales, steps, seed, upscale_factor, process_upscale, lora_model, process_lora):
    model = enable_lora(lora_model, basemodel) if process_lora else basemodel
    improved_prompt = await improve_prompt(prompt)
    combined_prompt = f"{prompt} {improved_prompt}"
    
    if seed == -1:
        seed = random.randint(0, MAX_SEED)
    seed = int(seed)
    
    progress_bar = st.progress(0)
    progress_bar.progress(10) 
    image, seed = await generate_image(combined_prompt, model, width, height, scales, steps, seed)
    progress_bar.progress(50)  
    
    if isinstance(image, str) and image.startswith("Error"):
        progress_bar.empty()
        return [image, None, combined_prompt]

    image_path = DATA_PATH / f"image_{seed}.jpg"
    image.save(image_path, format="JPEG")
    
    prompt_file_path = DATA_PATH / f"prompt_{seed}.txt"
    with open(prompt_file_path, "w") as prompt_file:
        prompt_file.write(combined_prompt)

    if process_upscale:
        upscale_image_path = get_upscale_finegrain(combined_prompt, image_path, upscale_factor)
        if upscale_image_path:
            upscale_image = Image.open(upscale_image_path)
            upscale_image.save(DATA_PATH / f"upscale_image_{seed}.jpg", format="JPEG")
            progress_bar.progress(100)
            image_path.unlink()
            return [str(DATA_PATH / f"upscale_image_{seed}.jpg"), str(prompt_file_path)]
        else:
            progress_bar.empty()
            return [str(image_path), str(prompt_file_path)]
    else:
        progress_bar.progress(100) 
        return [str(image_path), str(prompt_file_path)]
        
async def improve_prompt(prompt):
    try:
        instruction = ("With this idea, describe in English a detailed txt2img prompt in a single paragraph of up to 200 characters maximum, developing atmosphere, characters, lighting, and cameras.")
        formatted_prompt = f"{prompt}: {instruction}"
        response = llm_client.text_generation(formatted_prompt, max_new_tokens=200)
        improved_text = response['generated_text'].strip() if 'generated_text' in response else response.strip()
        return improved_text
    except Exception as e:
        return f"Error mejorando el prompt: {e}"

def get_storage():
    files = list(DATA_PATH.glob("*.jpg"))
    usage = sum(file.stat().st_size for file in files)
    return files, f"Uso total: {usage/(1024.0 ** 3):.3f}GB"

def get_prompts():
    prompt_files = list(DATA_PATH.glob("*.txt"))
    return {file.stem.replace("prompt_", ""): file for file in prompt_files}

def run_gen(prompt, basemodel, width, height, scales, steps, seed, upscale_factor, process_upscale, lora_model, process_lora):
    loop = asyncio.new_event_loop()
    asyncio.set_event_loop(loop)
    return loop.run_until_complete(gen(prompt, basemodel, width, height, scales, steps, seed, upscale_factor, process_upscale, lora_model, process_lora))

st.set_page_config(layout="wide")
st.title("Generador de Imágenes FLUX y Escalador con IA")

prompt = st.sidebar.text_input("Descripción de la imagen")
basemodel = st.sidebar.selectbox("Modelo Base", ["black-forest-labs/FLUX.1-schnell", "black-forest-labs/FLUX.1-DEV"])
lora_model = st.sidebar.selectbox("LORA Realismo", ["Shakker-Labs/FLUX.1-dev-LoRA-add-details", "XLabs-AI/flux-RealismLora"])
format_option = st.sidebar.selectbox("Formato", ["9:16", "16:9"])
process_lora = st.sidebar.checkbox("Procesar LORA")
process_upscale = st.sidebar.checkbox("Procesar Escalador")

if format_option == "9:16":
    width = st.sidebar.slider("Ancho", 512, 720, 720, step=8)
    height = st.sidebar.slider("Alto", 912, 1280, 1280, step=8)
else:
    width = st.sidebar.slider("Ancho", 512, 1280, 1280, step=8)
    height = st.sidebar.slider("Alto", 512, 720, 720, step=8)

upscale_factor = st.sidebar.selectbox("Factor de Escala", [2, 4, 8], index=0)
scales = st.sidebar.slider("Escalado", 1, 20, 10)
steps = st.sidebar.slider("Pasos", 1, 100, 20)
seed = st.sidebar.number_input("Semilla", value=-1)

if st.sidebar.button("Mejorar prompt"):
    improved_prompt = asyncio.run(improve_prompt(prompt))
    st.session_state.improved_prompt = improved_prompt
    st.write(f"{improved_prompt}")

if st.sidebar.button("Generar Imagen"):
    with st.spinner("Generando imagen..."):
        image_paths, prompt_file = run_gen(st.session_state.get('improved_prompt', prompt), basemodel, width, height, scales, steps, seed, upscale_factor, process_upscale, lora_model, process_lora)
    
        if image_paths and isinstance(image_paths[0], str) and Path(image_paths[0]).exists():
            st.image(image_paths[0], caption="Imagen Generada")
            prompt_text = Path(prompt_file).read_text() if prompt_file else "No disponible"
            st.write(f"Prompt utilizado: {prompt_text}")

files, usage = get_storage()
st.text(usage)

cols = st.columns(6)
prompts = get_prompts()
for idx, file in enumerate(files):
    with cols[idx % 6]:
        image = Image.open(file)
        prompt_file = prompts.get(file.stem.replace("image_", ""), None)
        prompt_text = Path(prompt_file).read_text() if prompt_file else "No disponible"
        st.image(image, caption=f"Imagen {idx+1}")
        st.write(f"Prompt: {prompt_text}")