Home / Extensions / Pixal3D

Pixal3D

Model
@DrHepa· Jul 2026
About

Pixal3D image-to-3D model extension for Modly. It converts a single input image into a textured GLB mesh.

Repository
https://github.com/DrHepa/modly-pixal3d-extension
View on GitHub
Install in Modly
  1. 1
    Open “Models” in the app
    Go to the Models page and click Install from GitHub.
  2. 2
    Paste the repository URL
    Use the Copy button above, then confirm the installation.
  3. 3
    Download the model
    Once installed, download the model or one of its variants.
Documentation

From the project README

Turn an image into a textured 3D asset, combine multiple views, or reconstruct objects in a scene with WorldSculpt—all inside Modly workflows.

Requires Modly 0.4.3 or newer. Model weights are downloaded through Modly Models; they are not bundled here.

1. Open Models/Extensions → Install from GitHub and enter: https://github.com/DrHepa/modly-pixal3d-extension 2. Let setup finish. If dependencies need recovery, use the extension's Repair action. 3. In Models, download the shared weight groups for your chosen node (below). 4. Connect your inputs in a workflow and run the node.

Normal setup automatically attempts the special scene-preparation and WorldSculpt environments on supported hosts. No separate manual setup is needed for a fresh install.

Version 0.7.3 — Windows RTX 5090 / SM120 (experimental): normal Install/Reinstall/Repair keeps the checksum-pinned r3 CPython 3.11 / CUDA 12.8 native archive unchanged. It adds a separate MV Python overlay, a private WorldSculpt dependency overlay, and extension-owned CPython 3.12 + SAM3/DA3 for scene preparation and automatic MV camera calibration. No PowerShell, administrator rights, Developer Mode, manual Python installation, model copies, or implicit model downloads are needed. The primary stack stays torch 2.7.1+cu128 / torchvision 0.22.1+cu128 / NATTEN 0.21.6. The four stable wheelhouse lanes are unchanged.

Read the full README on GitHub →
← Browse all extensions