genvid-nuke 0.1.0rc1

Genvid platform integration for Foundry Nuke 14+

genvid-nuke

Genvid platform integration for Foundry Nuke 14+.

This package lets compositors authenticate with the Genvid backend from inside Nuke, browse project comps, render to the platform, and surface live notifications about server-side comp generation.

Requirements

• Foundry Nuke 14.0 or newer (bundled Python is 3.9+).
• macOS or Windows. Linux is best-effort.
• A Genvid account with access to at least one project.

Layout

This is a standard pip-installable Python package using a src/ layout:

genvid_nuke/
├── pyproject.toml
└── src/genvid_nuke/
    ├── __init__.py
    ├── __about__.py
    ├── config.py
    └── tests/

Nuke discovers plugins by executing init.py and menu.py files placed on its NUKE_PATH (typically ~/.nuke/). Those bootstrap files are added in later tasks of the Phase 2 plan; they are intentionally thin shims that import this genvid_nuke package via standard Python import once it is installed into Nuke's Python environment.

Why pip-installable + thin loader? Nuke's bundled Python is a normal CPython interpreter, so a properly installed package on sys.path works exactly the same as any other library. Studios get clean upgrades via pip install --upgrade genvid-nuke, and we avoid the historical pitfall of shipping a single 3000-line menu.py.

Install

Two distribution artifacts are published, both built from the same source tree via dist/build.sh:

Wheel (recommended) — install into Nuke's bundled Python

"<Nuke install>/python" -m pip install genvid_nuke-1.0.0-py3-none-any.whl

This is the cleanest path for individual workstations: the package lands on Nuke's sys.path and import genvid_nuke works directly. Upgrades are pip install --upgrade.

.zip — drop onto a shared NUKE_PATH

For studios that distribute plugins by extracting to a shared NUKE_PATH location:

unzip genvid_nuke-1.0.0.zip -d ~/.nuke/genvid_nuke/

Then add the extracted directory itself (e.g. ~/.nuke/genvid_nuke/) to NUKE_PATH — in ~/.nuke/init.py or your studio launch script — and restart Nuke. The .zip ships a top-level init.py shim alongside a genvid_nuke/ package directory; on Nuke startup, the shim adds the extracted folder to sys.path and then imports genvid_nuke.init to wire the on-script-load sync. Once that runs, Nuke's bundled Python can import genvid_nuke directly.

Development

cd genvid_nuke
python -m pip install -e ".[dev]"

nuke itself is never a dependency — it is provided by the Nuke host. The test suite mocks sys.modules["nuke"] so the package can be exercised in CI without a real Nuke license.

Sign-in (preview)

Authentication uses the Supabase OAuth Password grant against the Genvid project, with refresh tokens cached via the OS keychain (keyring). Implementation lands in later Phase 2 tasks.

Usage (preview)

After install and sign-in, the Nuke menu gains a Genvid entry that lets users:

• Pick a project / comp.
• Push the current comp to the Genvid backend (POST /comps).
• See status updates as the server processes the comp.

Detailed usage docs live alongside the implementation in later tasks.

Configuration

All runtime configuration is environment-variable driven (see src/genvid_nuke/config.py). Defaults target Genvid production; studios behind a corporate proxy or running against a staging environment can override:

Variable	Default
GENVID_API_BASE_URL	https://api.genvidtech.com
GENVID_SUPABASE_AUTH_BASE_URL	https://<supabase-project>.supabase.co/auth/v1 (placeholder — filled at install time)
GENVID_SUPABASE_CLIENT_ID	genvid-nuke-plugin
GENVID_POLL_INTERVAL_SECONDS	5.0
GENVID_POLL_FAST_INTERVAL_SECONDS	2.0
GENVID_POLL_FAST_WINDOW_SECONDS	30.0

Testing

cd genvid_nuke
python -m pip install ".[dev]"
pytest -v -m "not e2e" --cov=src/genvid_nuke

The e2e marker is reserved for tests that require a real Nuke license; those are skipped in CI.