import gradio as gr import requests import io import random import os import time from PIL import Image from deep_translator import GoogleTranslator import json from gradio_client import Client # Project by Nymbo API_URL = "https://api-inference.huggingface.co/models/stabilityai/stable-diffusion-3.5-large" API_TOKEN = os.getenv("HF_READ_TOKEN") headers = {"Authorization": f"Bearer {API_TOKEN}"} timeout = 100 # Initialize the prompt enhancer client prompt_enhancer = Client("K00B404/mistral-nemo-prompt-enhancer") def enhance_prompt(prompt, enable_enhancement=True): """Enhance the given prompt using the Mistral Nemo prompt enhancer.""" if not enable_enhancement: print(f'\033[1mPrompt enhancement disabled, using original prompt:\033[0m {prompt}') return prompt try: system_message = "You are an expert at writing detailed, high-quality image generation prompts. Enhance the given prompt by adding relevant artistic details, style elements, and quality descriptors. Keep the original intent but make it more elaborate and specific." enhanced = prompt_enhancer.predict( message=prompt, system_message=system_message, max_tokens=512, temperature=0.7, top_p=0.95, api_name="/chat" ) print(f'\033[1mOriginal prompt:\033[0m {prompt}') print(f'\033[1mEnhanced prompt:\033[0m {enhanced}') return enhanced except Exception as e: print(f"Error enhancing prompt: {e}") return prompt # Fall back to original prompt if enhancement fails # Function to query the API and return the generated image def query(prompt, is_negative=False, steps=35, cfg_scale=7, sampler="DPM++ 2M Karras", seed=-1, strength=0.7, width=1024, height=1024, enable_enhancement=True): if prompt == "" or prompt is None: return None, None key = random.randint(0, 999) API_TOKEN = random.choice([os.getenv("HF_READ_TOKEN")]) headers = {"Authorization": f"Bearer {API_TOKEN}"} # Translate the prompt from Russian to English if necessary prompt = GoogleTranslator(source='ru', target='en').translate(prompt) print(f'\033[1mGeneration {key} translation:\033[0m {prompt}') # Enhance the prompt using the Mistral Nemo model if enabled enhanced_prompt = enhance_prompt(prompt, enable_enhancement) # Add some extra flair to the prompt final_prompt = f"{enhanced_prompt} | ultra detail, ultra elaboration, ultra quality, perfect." print(f'\033[1mGeneration {key} final prompt:\033[0m {final_prompt}') # Prepare the payload for the API call, including width and height payload = { "inputs": final_prompt, "is_negative": is_negative, "steps": steps, "cfg_scale": cfg_scale, "seed": seed if seed != -1 else random.randint(1, 1000000000), "strength": strength, "parameters": { "width": width, "height": height } } # Send the request to the API and handle the response response = requests.post(API_URL, headers=headers, json=payload, timeout=timeout) if response.status_code != 200: print(f"Error: Failed to get image. Response status: {response.status_code}") print(f"Response content: {response.text}") if response.status_code == 503: raise gr.Error(f"{response.status_code} : The model is being loaded") raise gr.Error(f"{response.status_code}") try: # Convert the response content into an image image_bytes = response.content image = Image.open(io.BytesIO(image_bytes)) print(f'\033[1mGeneration {key} completed!\033[0m ({final_prompt})') return image, enhanced_prompt except Exception as e: print(f"Error when trying to open the image: {e}") return None, None # CSS to style the app css = """ #app-container { max-width: 800px; margin-left: auto; margin-right: auto; } """ # Build the Gradio UI with Blocks with gr.Blocks(theme='Nymbo/Nymbo_Theme', css=css) as app: # Add a title to the app gr.HTML("

Stable Diffusion 3.5 Large with Prompt Enhancement

") # Container for all the UI elements with gr.Column(elem_id="app-container"): # Add a text input for the main prompt with gr.Row(): with gr.Column(elem_id="prompt-container"): with gr.Row(): text_prompt = gr.Textbox( label="Prompt", placeholder="Enter a prompt here - it will be automatically enhanced for better results", lines=2, elem_id="prompt-text-input" ) # Accordion for advanced settings with gr.Row(): with gr.Accordion("Advanced Settings", open=False): enable_enhancement = gr.Checkbox( label="Enable Prompt Enhancement", value=True ) negative_prompt = gr.Textbox( label="Negative Prompt", placeholder="What should not be in the image", value="(deformed, distorted, disfigured), poorly drawn, bad anatomy, wrong anatomy, extra limb, missing limb, floating limbs, (mutated hands and fingers), disconnected limbs, mutation, mutated, ugly, disgusting, blurry, amputation, misspellings, typos", lines=3, elem_id="negative-prompt-text-input" ) with gr.Row(): width = gr.Slider(label="Width", value=1024, minimum=64, maximum=1216, step=32) height = gr.Slider(label="Height", value=1024, minimum=64, maximum=1216, step=32) steps = gr.Slider(label="Sampling steps", value=35, minimum=1, maximum=100, step=1) cfg = gr.Slider(label="CFG Scale", value=7, minimum=1, maximum=20, step=1) strength = gr.Slider(label="Strength", value=0.7, minimum=0, maximum=1, step=0.001) seed = gr.Slider(label="Seed", value=-1, minimum=-1, maximum=1000000000, step=1) method = gr.Radio( label="Sampling method", value="DPM++ 2M Karras", choices=["DPM++ 2M Karras", "DPM++ SDE Karras", "Euler", "Euler a", "Heun", "DDIM"] ) # Add a button to trigger the image generation with gr.Row(): text_button = gr.Button("Generate Enhanced Image", variant='primary', elem_id="gen-button") # Image output area to display the generated image with gr.Row(): image_output = gr.Image(type="pil", label="Image Output", elem_id="gallery") # Text output area to display the enhanced prompt with gr.Row(): prompt_output = gr.Textbox(label="Enhanced Prompt", elem_id="prompt-output") # Bind the button to the query function with all inputs text_button.click( query, inputs=[text_prompt, negative_prompt, steps, cfg, method, seed, strength, width, height, enable_enhancement], outputs=[image_output, prompt_output] ) # Launch the Gradio app app.launch(show_api=True, share=False)