whatisgithub

What is maquette?

nhannht/maquette — explained in plain English

Analysis updated 2026-07-25

0JavaScriptAudience · generalComplexity · 2/5LicenseSetup · moderate

In one sentence

Maquette is a macOS app that turns a single photo into a 3D model using AI. It loops between a coder AI and a judge AI, letting you refine the result, and exports a model ready for games or AR.

Mindmap

mindmap
  root((repo))
    What it does
      Photo to 3D model
      AI loop with judge
      Human edits at gates
    Tech stack
      JavaScript
      Apple Silicon
      macOS 14
    Use cases
      Game assets
      Web 3D models
      AR on iPhone
    Privacy
      Photos stay local
      BYO API key
      Keychain storage
    Audience
      Creators
      Indie devs
    Setup
      Homebrew install
      CLI tool included

Code map

Detail Auto

An interactive map of this repo's files and how they connect — its source is parsed live in your browser. Click Visualize to build it.

filefunction / class

What do people build with it?

USE CASE 1

Turn a product photo into a 3D asset for a game or website.

USE CASE 2

Create 3D models from single images for augmented reality on iPhone.

USE CASE 3

Iteratively refine AI-generated 3D geometry using custom instructions.

USE CASE 4

Run the photo-to-3D pipeline automatically via the command line tool.

What is it built with?

JavaScriptmacOSApple Silicon

How does it compare?

nhannht/maquette00kaku/gallery-slider-block0xkinno/vellum
Stars00
LanguageJavaScriptJavaScriptJavaScript
Last pushed2021-05-19
MaintenanceDormant
Setup difficultymoderateeasyhard
Complexity2/52/54/5
Audiencegeneralgeneraldeveloper

Figures from each repo's GitHub metadata at analysis time.

How do you get it running?

Difficulty · moderate Time to first run · 5min

Requires an Apple Silicon Mac with macOS 14 or newer and an API key from an OpenAI-compatible AI provider.

Use freely for any purpose, including commercial use, as long as you keep the copyright notice.

So what is it?

Maquette is a macOS app that takes a photograph and produces a 3D model you can use in games, on the web, or in augmented reality on an iPhone. You drop a photo in, and two AI models work together in a loop. One model writes code that builds 3D geometry from scratch. The app renders that geometry from four angles, then a second model looks at the renders and compares them to your original photo. It assigns a score and tells the first model what to fix. The loop repeats, and the best scoring version is the one that gets exported as the final model. You can step in at two points during the process. The first gate lets you edit the plain language description of what the app thinks it sees in your photo. You can add details that are not visible, like saying a box should open to reveal earbuds inside. The second gate lets you modify the instruction the judge model gives to the coder model before the next cycle. You can also let the whole thing run on autopilot. Every cycle is saved, so you can pick any version to export or promote as the new starting point for further refinement. The app brings your own API key, meaning you choose which AI provider to use. Any service that matches the OpenAI interface works, including OpenAI, Anthropic, OpenRouter, and local tools like Ollama that require no key at all. Your key is stored in the macOS Keychain, never in plain files or logs. The 3D rendering happens entirely offline, so your photo never leaves your computer. The only network traffic is talking to the AI endpoint you selected. The creator notes the entire validation process cost under ten cents. Maquette runs on Apple Silicon Macs with macOS 14 or newer. You can install it through Homebrew or download a signed disk image from the GitHub releases page. A command line tool is also included for running the full process headless. The project is honest about its limits: it works best with single objects on plain backgrounds, and scores vary between runs. The code and the underlying idea are open source under the MIT license.

Copy-paste prompts

Prompt 1
How do I configure Maquette to use a local Ollama instance instead of a paid API key for generating 3D models?
Prompt 2
Using Maquette's human gates, what plain English prompt should I write to tell the AI that a box should open to reveal earbuds inside?
Prompt 3
Write a shell script to run the Maquette CLI headless on a folder of product photos and export the best 3D models automatically.
Prompt 4
If the Maquette judge AI gives a low score on my 3D model, what kind of correction instruction should I provide to get a better next cycle?

Frequently asked questions

What is maquette?

Maquette is a macOS app that turns a single photo into a 3D model using AI. It loops between a coder AI and a judge AI, letting you refine the result, and exports a model ready for games or AR.

What language is maquette written in?

Mainly JavaScript. The stack also includes JavaScript, macOS, Apple Silicon.

What license does maquette use?

Use freely for any purpose, including commercial use, as long as you keep the copyright notice.

How hard is maquette to set up?

Setup difficulty is rated moderate, with roughly 5min to a first successful run.

Who is maquette for?

Mainly general.

Open on GitHub → Ask about another repo

This repo across BitVibe Labs

Verify against the repo before relying on details.