Blender

3427 readers

1 users here now

A community for users of the awesome, open source, free, animation, modeling, procedural generating, sculpting, texturing, compositing, and rendering software; Blender.

Rules:

Be nice
Constructive Criticism only
If a render is photo realistic, please provide a wireframe or clay render

founded 2 years ago

Blender .obj-export precision (feddit.org)

submitted 2 weeks ago by Luccus@feddit.org to c/blender@lemmy.world

3 comments fedilink hide all child comments

I am currently creating some meshes that are then exported as OBJs. I am really only interested in the vertex positions and normals, as everything else will be done with shaders. But I keep coming across… well, not really a issue.

Some exported vertex positions deviate from their in-editor position by literally a micrometer. Again, I know that this is basically irrelevant. Perhaps, if I were to compress the files, I might have a few kilobytes more data than actually necessary.

…but I keep seeing this and it annoys me for no practical reason. Does anyone know if there is a "fix" for this behavior?

top 3 comments

sorted by: hot top controversial new old

[–] zener_diode@feddit.org 9 points 2 weeks ago

This looks a lot like floating point errors. There's probably some math involved in exporting that causes these slight inaccuracies.

It probably isn't too hard to write a short python script to go over the exported file, and round those values.

[–] itsathursday@lemmy.world 5 points 2 weeks ago* (last edited 2 weeks ago)

The precision is default afaik, and necessary for accuracy when dealing with objects far from origin. The noise you see though is due to floating point precision which is a thing with numbers on computers. This is common to all 3d software. https://docs.oracle.com/cd/E19957-01/806-3568/ncg_goldberg.html

If you know your tolerances are much less and disk space is a concern you could post process this data in the file as text, but I wouldn’t bother.

[–] blazebra@programming.dev 3 points 2 weeks ago

For 32-bit float 6 decimal digital are precise in representation, beyond that expect to have rounding errors.

I don’t see if blender ever has 64-bit support but there were rejected requests.