One Button Prompt is a Stable Diffusion prompt generation tool designed for both beginners and advanced users seeking creative inspiration. It automates the creation of detailed, randomized prompts for text-to-image generation, supporting various Stable Diffusion workflows and models.

How It Works

The tool generates complete prompts from scratch using a controlled random approach, combining elements like subjects (objects, animals, humanoids, landscapes, concepts), artists, and techniques. Users can influence the output through various modes, subject filters, and by providing custom artists or subjects, allowing for controlled randomness and exploration of creative possibilities.

Quick Start & Requirements

• Installation: Available as a script for Automatic1111/SD.next (via Extensions -> Available -> Load From, or install from URL: https://github.com/AIrjen/OneButtonPrompt) and as a custom node for ComfyUI. A fork with built-in support exists for RuinedFooocus.
• Prerequisites: Stable Diffusion WebUI (Automatic1111/SD.next) or ComfyUI.
• Usage: Select the "One Button Prompt" script in TXT2IMG or IMG2IMG and press "Generate" with empty prompt fields.
• Documentation: User guide available here.

Highlighted Details

• Supports TXT2IMG, IMG2IMG, ControlNET, inpainting, and latent couple.
• Features prompt generation modes, workflow assist for prompt variants, and an "Overwrite subject" option for focused generation.
• Includes automated generation, classification, upscaling, and LoRA/custom artist list integration.
• Offers a "One Button Run and Upscale" feature for a complete generation-to-upscaling workflow.

Maintenance & Community

The project acknowledges inspiration from Stylepile and Dynamic Prompts, and thanks the SD artists list and openart.ai. Ongoing development focuses on prompt generation refinements and documentation. Suggestions are welcomed.

Licensing & Compatibility

The repository does not explicitly state a license in the provided README. Compatibility for commercial use or closed-source linking is not specified.

Limitations & Caveats

The README notes that not every generated image will be "awesome" due to randomness, with an estimated 1 in 5 being "good." Recent UI changes with the introduction of Presets might be confusing for existing users, requiring a change to "Custom..." to access detailed settings. Subtypes have been integrated into the subject field.