Turn snapshots into 3D meshes without shipping them to the cloud

What it does

Modly is a desktop application for Windows and Linux that turns photographs into 3D meshes using open-source AI models. All inference happens locally on your GPU, so nothing gets uploaded to the cloud. The project is built by Lightning Pixel and distributed under the MIT license.

The interesting bit

Instead of baking models into the app, Modly treats each AI model as a plug-in extension pulled from its own GitHub repository. Each extension is just a manifest.json and a generator.py, which keeps the architecture modular and lets new models arrive without waiting for a full app release.

Key highlights

• Runs entirely on your local GPU; no cloud upload or remote queue required.
• Supports five official extensions covering Hunyuan3D 2 Mini (and its Turbo and Fast variants), TripoSG, and Trellis2 GGUF.
• New models install directly from GitHub via the in-app Models page.
• MIT licensed, though the license explicitly demands visible attribution in your UI or documentation if you fork the project.

Caveats

• macOS support is listed as “coming soon,” so Apple silicon users are left waiting for now.
• The README is light on technical specifics: exact GPU requirements, output mesh formats, and hardware baselines are not stated.

Verdict

Worth a look if you want to experiment with image-to-3D conversion without sending assets to a remote API. Skip it if you are on macOS or need detailed hardware and export-format documentation.