import gradio as gr import os import sys from pathlib import Path import random import string import time from queue import Queue from threading import Thread import emoji def pipe(val): return val ru2en = gr.Interface.load("huggingface/Helsinki-NLP/opus-mt-ru-en") text_gen=gr.Interface.load("spaces/phenomenon1981/MagicPrompt-Stable-Diffusion") def get_prompts(prompt_text): return text_gen("photo, " + prompt_text + ", high details, harmony, ideal proportions") proc1=gr.Interface.load("models/dreamlike-art/dreamlike-photoreal-2.0") def restart_script_periodically(): while True: time.sleep(600) # 10 minutes try: os.execl(sys.executable, sys.executable, *sys.argv) except: pass restart_thread = Thread(target=restart_script_periodically, daemon=True) restart_thread.start() queue = Queue() queue_threshold = 800 def add_random_noise(prompt, noise_level=0.00): if noise_level == 0: noise_level = 0.00 # Get the percentage of characters to add as noise percentage_noise = noise_level * 5 # Get the number of characters to add as noise num_noise_chars = int(len(prompt) * (percentage_noise/100)) # Get the indices of the characters to add noise to noise_indices = random.sample(range(len(prompt)), num_noise_chars) # Add noise to the selected characters prompt_list = list(prompt) # Add numbers, special characters, and all emojis to the list of characters used to add noise noise_chars = string.ascii_letters + string.punctuation + ' ' + string.digits + emoji.emojize(":all:") for index in noise_indices: prompt_list[index] = random.choice(noise_chars) return "".join(prompt_list) def send_it(inputs, noise_level, proc1=proc1): prompt_with_noise = add_random_noise(inputs, noise_level) output = proc1(prompt_with_noise) return output with gr.Blocks(analytics_enabled=False, css='style.css') as demo: with gr.Column(elem_id="col-container"): with gr.Row(elem_id="shorts"): input_text_ru = gr.Textbox( label="Short Ru", placeholder="девушка", ) Translate = gr.Button("Ru ▶ En", elem_id="translate").style(full_width=False) input_text = gr.Textbox( label="Short En", placeholder="girl", ) with gr.Row(elem_id="longs"): with gr.Column(scale=1000): prompt = gr.Textbox( label="Prompt", placeholder="proto, girl, high details", ) with gr.Column(scale=1, elem_id="longs-fillers"): Literally = gr.Button("◀ Literally") Explain = gr.Button("◀ Explain") with gr.Row(elem_id="params"): with gr.Column(): noise_level = gr.Slider(minimum=0.0, value=1.0, maximum=3, step=0.1, label="Noise Level") with gr.Row(elem_id="paints"): with gr.Column(): run1 = gr.Button("🔻 Paint") output1=gr.Image() with gr.Column(): run2 = gr.Button("🔻 Paint") output2=gr.Image() with gr.Column(): run3 = gr.Button("🔻 Paint") output3=gr.Image() # with gr.Column(): # run4 = gr.Button("🔻 Paint") # output4=gr.Image() gr.Markdown("Honor artists: Yayoi Kusama, Pablo Picasso, Leonardo da Vinci, Banksy, Rembrandt, Frida Kahlo, Vincent van Gogh, Henri Matisse, Salvador Dali, Claude Monet, Andy Warhol, Georgia O'Keeffe, Jackson Pollock, Marcel Duchamp, Edward Hopper, Willem de Kooning, Mark Rothko, David Hockney") gr.Markdown("Popuar Styles: Anime, Abstract, Minimalist, Cyberpunk, Steampunk, Organic, Geometric, Sci-Fi, Futuristic, Vaporwave, Gothic") gr.Markdown("Known Moods: joyful, light-hearted, exciting, calming, soothing, playful, fun, bright, colourful, dynamic, energetic, passionate, romantic, vibrant, vivid") gr.Markdown("Typical techniques: oil on canvas, chinese painting, graffiti, watercolour, graphite, cinematic, film noir, fluorescent, moody lighting, silhouette, ultraviolet, x-ray, olaroid, double exposure, fisheye lens, bokeh") Literally.click(pipe, inputs=[input_text], outputs=[prompt], queue=False) Translate.click(ru2en, inputs=[input_text_ru], outputs=[input_text], queue=False) Explain.click(text_gen, inputs=[input_text], outputs=[prompt], queue=False) run1.click(send_it, inputs=[prompt, noise_level], outputs=[output1]) run2.click(send_it, inputs=[prompt, noise_level], outputs=[output2]) run3.click(send_it, inputs=[prompt, noise_level], outputs=[output3]) # run4.click(send_it, inputs=[prompt, noise_level], outputs=[output4]) demo.launch(enable_queue=True, inline=True, show_api=False) block.queue(concurrency_count=2)