Skip to content

GNM-Maya 1.2.0

Choose a tag to compare

View all tags
@enriquevelmai enriquevelmai released this 16 Jul 18:06
v1.2.0
9e5ae20

GNM-Maya goes cross-platform: Windows, Linux and macOS from one codebase — the full artist toolset wrapping Google GNM parametric human heads inside Autodesk Maya.

What's new since 1.1.0

  • Linux & macOS support. The module's own Python runtime is now built per platform:
    • Windows — unchanged: a portable embeddable CPython is downloaded once (self-contained, relocatable, no system Python needed).
    • Linux/macOS — a venv is created inside the module on first run from your python3 (3.9+), with Maya's own mayapy as an automatic fallback. All packages stay under the module's runtime/ folder — nothing touches your system.
  • The drag-and-drop installer, first-run setup, shape-gallery renderer, photo-fitting installer and dev rebuild script are all platform-aware; the installer knows a posix venv isn't relocatable and rebuilds it in place instead of copying a broken one.
  • Docs updated with per-OS requirements and notes.

Install (Maya 2022–2027)

  1. Download Source code (zip) below and extract it anywhere.
  2. Drag drag_and_drop_install.py (top of the extracted folder) into a running Maya viewport.
  3. Accept the first-run setup (Python runtime + GNM model ~40 MB; shape images render locally ~5 min).

The installer copies the module into ~/Documents/maya/modules, so you can delete the zip and extracted folder afterwards. Everything runs offline after setup.

Per-platform requirements

  • Windows — nothing extra; the runtime is fully self-contained.
  • Linux — python3 (3.9+) with the venv module (Debian/Ubuntu: sudo apt install python3-venv); mayapy is the fallback.
  • macOS — python3 (3.9+) via Xcode CLT or Homebrew; mayapy is the fallback.

Notes

  • The Windows path is fully validated (headless test suite); Linux/macOS follow the same code with platform branches unit-tested — if you hit an issue on those platforms, every failure now surfaces in an error dialog, and reports are welcome.
  • Photo fitting (optional ~290 MB MediaPipe/OpenCV) was validated on Windows; the same install is attempted on Linux/macOS.
  • GNM is Apache-2.0 (Google); this tool is MIT and not affiliated with Google. See NOTICE.md.
Assets 2
Loading