#!/usr/bin/env python import gradio as gr from PIL import Image import math from visual_markov_patterns_generator import create_generator def create_image_grid(images, rows=None, cols=None): """Images grid generation""" n_images = len(images) if rows is None and cols is None: cols = min(4, n_images) rows = math.ceil(n_images / cols) elif rows is None: rows = math.ceil(n_images / cols) elif cols is None: cols = math.ceil(n_images / rows) w, h = images[0].size grid = Image.new("RGB", size=(cols * w, rows * h)) for i, img in enumerate(images): if i >= rows * cols: break row = i // cols col = i % cols grid.paste(img, box=(col * w, row * h)) return grid def generate_image_interface(temperature, seed, use_seed, batch_size): """Interface function for image generation""" try: images = [] for i in range(batch_size): actual_seed = (seed + i) if use_seed else None image = generator.generate_image(temperature=temperature, seed=actual_seed) scale_factor = 4 new_size = (image.width * scale_factor, image.height * scale_factor) image = image.resize(new_size, Image.BICUBIC) images.append(image) if batch_size == 1: result_image = images[0] else: cols = min(4, batch_size) rows = math.ceil(batch_size / cols) result_image = create_image_grid(images, rows, cols) info = f""" **Generation Parameters:** - Temperature: {temperature} - Starting Seed: {seed if use_seed else 'random'} - Batch Size: {batch_size} - Image size: {images[0].size} """ return result_image, info except Exception as e: error_info = f"Generation error: {str(e)}" empty_img = generator._create_random_image() scale_factor = 4 empty_img = empty_img.resize( (empty_img.width * scale_factor, empty_img.height * scale_factor), Image.BICUBIC, ) return empty_img, error_info def get_model_info(): """Get model information""" info = """ ## How to Use: 1. **Temperature (0.1-3.0):** Controls generation creativity - Low values (0.1-0.5): More conservative generation - Medium values (0.5-1.5): Balanced generation - High values (1.5-3.0): More creative generation 2. **Seed:** For reproducible results (optional) 3. **Batch Size:** Controls the number of generated images 3. Click "Generate Image" to create a new image ## Pattern Types: - haar - gradient - texture - color - frequency """ return info with gr.Blocks(title="Visual Markov Pattern Generator", theme=gr.themes.Soft()) as demo: gr.Markdown("# Visual Markov Mosaic Generator") gr.Markdown( "Image generator based on multi-dimensional Markov chains and neural network" ) with gr.Row(): with gr.Column(scale=1): gr.Markdown("## Generation Parameters") temperature = gr.Slider( minimum=0.1, maximum=3.0, value=1.0, step=0.1, label="Temperature", info="Controls generation creativity", ) with gr.Row(): use_seed = gr.Checkbox( label="Use seed", value=False, info="For reproducible results", ) seed = gr.Number(value=42, precision=0, label="Seed", interactive=True) batch_size = gr.Slider( minimum=1, maximum=20, value=1, step=1, label="Batch Size", info="Number of images to generate", ) generate_btn = gr.Button("Generate Images", variant="primary", size="lg") gr.Markdown("## Information") info_output = gr.Markdown(value=get_model_info()) with gr.Column(scale=2): gr.Markdown("## Result") image_output = gr.Image( label="Generated Images", type="pil", show_label=True, container=True ) generation_info = gr.Markdown(label="Generation Information") generate_btn.click( fn=generate_image_interface, inputs=[temperature, seed, use_seed, batch_size], outputs=[image_output, generation_info], ) if __name__ == "__main__": print("Initializing generator...") generator = create_generator() demo.launch(share=False, server_name="0.0.0.0", server_port=7860)