Gradio Workflow Rebuilds AUTOMATIC1111 Features as a Single Workflow Canvas

Introducing Workflow1111

Gradio has announced Workflow1111, a project that reconstructs most of the features found in AUTOMATIC1111's stable-diffusion-webui within a single Gradio workflow canvas. This initiative aims to provide a structured, node-based environment for complex AI media generation tasks.

Core Functionality and Structure

Workflow1111 is built as a graph of eleven distinct media pipelines, utilizing 73 individual nodes. These pipelines cover a range of functionalities, including state-of-the-art models for text-to-image generation, hi-resolution image fixing, image-to-image transformations, prompt-matrix grids, VLM interrogation, detection-to-inpaint masks, ControlNet-style annotators, background removal, PNG Info storage, and image-to-video conversion.

The system uses four operator kinds: Python functions (fn), models called via InferenceClient (model), other Gradio Spaces (space), and rows from a Hugging Face Hub dataset (dataset). Each node on the canvas encapsulates one of these operators.

Access and Usage

Users can access and run these pipelines by signing in with a Hugging Face account or providing an access token. All model calls made through Workflow1111 utilize the user's own quota, providing a direct link between usage and resource allocation. The project is available for direct use or duplication for custom modifications.

Key Pipelines Detailed

The core text-to-image pipeline includes controls for negative prompts, steps, CFG, seed, dimensions, and model selection. A prompt-builder function node processes the prompt, and a post-process function node embeds generation parameters into PNG metadata. The hi-resolution fix pipeline uses a FLUX.1-Kontext model node for upscaling and denoising. This same Kontext node also handles image-to-image tasks, allowing users to upload an image and describe desired changes.