skymanbp/autoshop

AI-assisted RAW photo developer: GPT vision advisor proposes an EditRecipe, a deterministic Rust engine renders it — Lightroom-compatible XMP sidecars, 24 RAW formats + baked images, measured (not guessed) Lightroom mask geometry, local GUI/web UI, AI denoise & segmentation sidecars

What it solves

Autoshop automates the tedious process of developing RAW photographs. Instead of manually adjusting sliders for exposure, white balance, and contrast, it uses AI to analyze a photo and decide the optimal adjustments, while ensuring the final image is rendered by a deterministic engine to avoid AI-generated hallucinations in the actual pixels.

How it works

The system separates the "judgment" from the "application":

  1. Analysis: A vision model (like GPT) looks at the photo and generates an EditRecipe (a JSON file of slider values).
  2. Verification: A second AI (like Claude) verifies the proposed recipe.
  3. Rendering: A deterministic Rust-based engine applies these slider values to the original RAW file to produce a 16-bit TIFF master or an XMP sidecar file that can be opened in Adobe Lightroom.

It also includes specialized AI modules for denoising (using SCUNet) and experimental generative features for retouching and reimagining images.

Who it’s for

Photographers who want to automate the initial grading of their RAW files while maintaining professional-grade, non-destructive control and compatibility with industry-standard tools like Lightroom.

Highlights

  • Non-destructive workflow: The AI suggests slider values; it does not modify pixels directly in the main path.
  • Lightroom Integration: Exports XMP sidecars, allowing AI-suggested edits to be fine-tuned manually in Lightroom.
  • Visual Feedback Loop: The AI can render its proposal, judge the result, and perform one guided revision.
  • Look Matching: Can "reverse-fit" a target image to solve for the specific EditRecipe needed to reproduce that look.
  • AI Denoising: Integrated GPU-accelerated SCUNet for high-ISO and astrophotography frames.
  • Comprehensive Interface: Available via a native desktop GUI, a web UI, and a command-line interface.

Related

  • Project
    Gigaforshovel/photo-enhancer-boost

    An AI-powered photo enhancement workspace for Windows that provides non-destructive editing, masking, and compositing tools.

  • Project
    barefootford/buttercut

    An AI video editing agent that generates XML cuts for professional software like Final Cut Pro, Premiere, and DaVinci Resolve to automate the initial assembly of footage.

  • Project
    aoguai/LiYing

    An automated ID photo processing tool that uses AI to handle face detection, background replacement, and layout arrangement for professional-grade ID photos.

  • Project
    Vincentwei1021/video-shotcraft

    An AI agent skill and toolkit for crafting cinematic product videos using Remotion, featuring a library of shot recipes and production-ready templates.

  • Dispatch
    Training AI to Paint with Code using Reinforcement Learning

    Surya Narreddi and team developed a system that trains a language model to generate editable p5.brush JavaScript code to create paintings, using a reinforcement learning loop based on aesthetic judgment.