import gradio as gr import requests import io import random import os from PIL import Image # List of available models list_models = [ "SDXL 1.0", "SD 1.5", "OpenJourney", "Anything V4.0", "Disney Pixar Cartoon", "Pixel Art XL", "Dalle 3 XL", "Midjourney V4 XL", "Open Diffusion V1", "SSD 1B", "Segmind Vega", "Animagine XL-2.0", "Animagine XL-3.0", "OpenDalle", "OpenDalle V1.1", "PlaygroundV2 1024px aesthetic", ] # Function to generate images from text def generate_txt2img(current_model, prompt, is_negative=False, image_style="None style", steps=50, cfg_scale=7, seed=None): if current_model == "SD 1.5": API_URL = "https://api-inference.huggingface.co/models/runwayml/stable-diffusion-v1-5" elif current_model == "SDXL 1.0": API_URL = "https://api-inference.huggingface.co/models/stabilityai/stable-diffusion-xl-base-1.0" elif current_model == "OpenJourney": API_URL = "https://api-inference.huggingface.co/models/prompthero/openjourney" elif current_model == "Anything V4.0": API_URL = "https://api-inference.huggingface.co/models/xyn-ai/anything-v4.0" elif current_model == "Disney Pixar Cartoon": API_URL = "https://api-inference.huggingface.co/models/stablediffusionapi/disney-pixar-cartoon" elif current_model == "Pixel Art XL": API_URL = "https://api-inference.huggingface.co/models/nerijs/pixel-art-xl" elif current_model == "Dalle 3 XL": API_URL = "https://api-inference.huggingface.co/models/openskyml/dalle-3-xl" elif current_model == "Midjourney V4 XL": API_URL = "https://api-inference.huggingface.co/models/openskyml/midjourney-v4-xl" elif current_model == "Open Diffusion V1": API_URL = "https://api-inference.huggingface.co/models/openskyml/open-diffusion-v1" elif current_model == "SSD 1B": API_URL = "https://api-inference.huggingface.co/models/segmind/SSD-1B" elif current_model == "Segmind Vega": API_URL = "https://api-inference.huggingface.co/models/segmind/Segmind-Vega" elif current_model == "Animagine XL-2.0": API_URL = "https://api-inference.huggingface.co/models/Linaqruf/animagine-xl-2.0" elif current_model == "Animagine XL-3.0": API_URL = "https://api-inference.huggingface.co/models/cagliostrolab/animagine-xl-3.0" elif current_model == "OpenDalle": API_URL = "https://api-inference.huggingface.co/models/dataautogpt3/OpenDalle" elif current_model == "OpenDalle V1.1": API_URL = "https://api-inference.huggingface.co/models/dataautogpt3/OpenDalleV1.1" elif current_model == "PlaygroundV2 1024px aesthetic": API_URL = "https://api-inference.huggingface.co/models/playgroundai/playground-v2-1024px-aesthetic" API_TOKEN = os.environ.get("HF_READ_TOKEN") headers = {"Authorization": f"Bearer {API_TOKEN}"} if image_style == "None style": payload = { "inputs": prompt + ", 8k", "is_negative": is_negative, "steps": steps, "cfg_scale": cfg_scale, "seed": seed if seed is not None else random.randint(-1, 2147483647) } elif image_style == "Cinematic": payload = { "inputs": prompt + ", realistic, detailed, textured, skin, hair, eyes, by Alex Huguet, Mike Hill, Ian Spriggs, JaeCheol Park, Marek Denko", "is_negative": is_negative + ", abstract, cartoon, stylized", "steps": steps, "cfg_scale": cfg_scale, "seed": seed if seed is not None else random.randint(-1, 2147483647) } elif image_style == "Digital Art": payload = { "inputs": prompt + ", faded , vintage , nostalgic , by Jose Villa , Elizabeth Messina , Ryan Brenizer , Jonas Peterson , Jasmine Star", "is_negative": is_negative + ", sharp , modern , bright", "steps": steps, "cfg_scale": cfg_scale, "seed": seed if seed is not None else random.randint(-1, 2147483647) } elif image_style == "Portrait": payload = { "inputs": prompt + ", soft light, sharp, exposure blend, medium shot, bokeh, (hdr:1.4), high contrast, (cinematic, teal and orange:0.85), (muted colors, dim colors, soothing tones:1.3), low saturation, (hyperdetailed:1.2), (noir:0.4), (natural skin texture, hyperrealism, soft light, sharp:1.2)", "is_negative": is_negative, "steps": steps, "cfg_scale": cfg_scale, "seed": seed if seed is not None else random.randint(-1, 2147483647) } image_bytes = requests.post(API_URL, headers=headers, json=payload).content image = Image.open(io.BytesIO(image_bytes)) return image # Function to read CSS from file def read_css_from_file(filename): with open(filename, "r") as file: return file.read() # Read CSS from file css = read_css_from_file("style.css") PTI_SD_DESCRIPTION = '''
Stable Diffusion Models Image Generation
Generate images directly from text prompts (no parameter tuning required)
''' # Prompt examples prompt_examples = [ "A blue jay standing on a large basket of rainbow macarons.", "A dog looking curiously in the mirror, seeing a cat.", "A robot couple fine dining with Eiffel Tower in the background.", "A chrome-plated duck with a golden beak arguing with an angry turtle in a forest.", "A transparent sculpture of a duck made out of glass. The sculpture is in front of a painting of a landscape.", "A cute corgi lives in a house made out of sushi.", "A single beam of light enter the room from the ceiling. The beam of light is illuminating an easel. On the easel there is a Rembrandt painting of a raccoon.", "A photo of a Corgi dog riding a bike in Times Square. It is wearing sunglasses and a beach hat." ] # Creating Gradio interface with gr.Blocks(css=css) as demo: gr.Markdown(PTI_SD_DESCRIPTION) with gr.Row(): with gr.Column(): current_model = gr.Dropdown(label="Select Model", choices=list_models, value=list_models[1]) text_prompt = gr.Textbox(label="Input Prompt", placeholder="Example: woman in the street ", lines=2) text_prompt.examples = prompt_examples with gr.Column(): negative_prompt = gr.Textbox(label="Negative Prompt (optional)", placeholder="Example: blurry, unfocused", lines=2) image_style = gr.Dropdown(label="Select Style", choices=["None style", "Cinematic", "Digital Art", "Portrait"], value="None style") generate_button = gr.Button("Generate Image", variant='primary') with gr.Row(): image_output = gr.Image(type="pil", label="Image Output") generate_button.click(generate_txt2img, inputs=[current_model, text_prompt, negative_prompt, image_style], outputs=image_output) # Launch the app demo.launch()