Add Blender addons

[nicolalandro]

It is possible to add addons to the blender image? Some addons could be slow to install and made the system crash. (Tested addons: DA3, MPBF2, potential interestd: blender-mcp)

I try to put the zip of the addons inside the addons folder and I create a install.py:

import bpy
import os

os.environ["PIP_ROOT_USER_ACTION"] = "ignore" # also if I add that I obtain a warning about root pip

# upgrade pip to latest version
__import__("ensurepip")._bootstrap()
__import__("pip._internal")._internal.main(
    ["install", "-U", "pip", "setuptools", "wheel"]
)

for f in os.listdir("/code/addons"):
    bpy.ops.extensions.package_install_files(
        filepath=f"/code/addons/{f}",
        repo="user_default",
        enable_on_install=True,
        overwrite=True,
    )

After I create a Docekrfile:

FROM lscr.io/linuxserver/blender:latest

WORKDIR /code
COPY addons ./addons
COPY install.py .
ENV PIP_ROOT_USER_ACTION=ignore
RUN blender -b -y --python install.py

And a docker-compose.yml:

---
services:
    blender:
        # image: lscr.io/linuxserver/blender:latest
        build: .
        container_name: blender
        environment:
            - PUID=1000
            - PGID=1000
            - TZ=Etc/UTC
        volumes:
            - ./config:/config
        ports:
            - 3001:3001
        shm_size: "1gb"
        restart: unless-stopped
        deploy:
            resources:
                reservations:
                    devices:
                        - driver: nvidia
                          device_ids: ["0"]
                          capabilities: [gpu]

[thelamer]

If you do the legwork on what core deps are required we can get them in the image.

Just in general no I am not testing plugins, but what I would do is not work like this I would close blender right click on the desktop open xterm and install a mini dev env in the container (with a new terminal xterm is awful for work) and try to install plugins, you have sudo and the correct userspace for everything down to the python venv in this mode. Once you know that process (capture history) you can replicate it in bash.

This is the majority of my workflow for making application containers, spin up a base and try to get it running in a terminal.