GLB and compressed models

Instead of the OBJ files the card can load a .glb file: a single binary file, faster to load. mtlfile is not needed then.

path: /local/floor3d/
objfile: home.glb

From OBJ to GLB

obj2gltf converts the export of Sweet Home 3D. It needs Node.js:

npm install -g obj2gltf

In the folder of the export (the .obj, the .mtl and the pictures):

obj2gltf --checkTransparency -i home.obj -o home.glb

It takes from a few seconds to a few minutes. Copy home.glb into the folder of the model in Home Assistant.

Compressed models

A .glb can be compressed with glTF Transform, often several times smaller:

npx @gltf-transform/cli meshopt home.glb home-meshopt.glb
npx @gltf-transform/cli draco home.glb home-draco.glb

The names of the objects stay the same, so the entities keep working.

Caution

Use only these two commands: gltf-transform optimize also merges objects, and the entities would no longer find them.

• meshopt: the decoder is part of the card, about 25 KB, loaded only by these models. Doors, covers and fans move as in the uncompressed model.
• Draco: usually the smallest file. The decoder (about 250 KB, the glTF build that three.js includes, by Google under the Apache License 2.0) comes with the card, next to its other files, and is loaded only by these models; it works without Internet access. draco_decoder_path sets another folder with draco_wasm_wrapper.js and draco_decoder.wasm, for example /local/draco/.

The card reads the file first and loads a decoder only when the model needs it: an uncompressed model loads as before. The idea comes from Steven-D-Morgan/hass-3d-floorplan.