Page MenuHome

Collada Import Ignores Vertex Normals
Open, NormalPublic


Blender seems to be ignoring vertex normals when importing COLLADA the way I'm defining them.

I'm the developer of a voxel editor called VoxelShop. A simple file containing only a cube would look as following.


<?xml version="1.0" encoding="UTF-8"?>
<COLLADA xmlns="" version="1.4.1">
      <geometry id="Plane-tex-mesh-0" name="Plane-tex">
            <polylist count="12" material="lambert0-material">
               <input offset="0" semantic="VERTEX" source="..." />
               <input offset="1" semantic="NORMAL" source="..." />
               <input offset="2" semantic="TEXCOORD" set="0" source="..." />

This imports correctly into Open3Mod and validates against the COLLADA xml schema. However,

Blender ignores the vertex normals when importing.

Excluding the normals altogether from the COLLADA file results in the same normals being generated. This leads me to believe that they are simply ignored.

I tried looking at the blender code here, but couldn't really find where the relevant part of the xml is processed.

Any insight would be appreciated!



Event Timeline

Bastien Montagne (mont29) triaged this task as Normal priority.

@Gaia Clary (gaiaclary) sounds more like a TODO than a bug to me, since support for custom vertex normals in Blender is fairly recent?

Vertex Normal import is not yet added. I will look into this shortly.

Vertex Normal import is not yet added. I will look into this shortly.

Thank you. Still very much looking forward to this!

Was this ever looked at? It seems like a pretty big issue to me.

arghh, yes, i have this still on my list. I had to pause my work on the Blender Collada module for a while, but i am eager to get back on this as soon as possible, to make sure the Collada module will be ready for Blender 2.8

I also believe the support for importing custom normals is not a big thing to do. But - if you like to do so - a patch (based on the collada branch) is always welcome and might speed up the implementation a lot :)