In the code: Handedness makes no sense for view/model matrices until someone proves me wrong about vek HOT 3 CLOSED

This fix is now published in 0.9.7 :)

from vek.

Hi,

Very interesting question! :)

Well, my reasoning was that, the only time handedness matters is when coordinates are mapped from your space (LH or RH) to your graphics API's NDC (Normalized Device Coordinates).
Because look_at does not define such a mapping (it's supposedly a "your-space"-to-"your-space" mapping), handedness would not matter.
To me, it would be like, for instance, basic translation/rotation/scale: They are defined relatively to X, Y or Z, regardless of handedness.

... But the thing I failed to notice was that look_at, though such a translation/rotation combination, is not actually defined relatively to absolute axii. Rather, look_at is usually called this way :

look_at(position, position + forward, up);

... and that forward vector implies a dependency on handedness, because it changes with it. Oops.

So I was wrong, and should've followed GLM's example. :) What I'm going to do is make look_at_lh and look_at_rh public, and deprecate look_at. If your code "worked" when flipping the forward vector (assuming the given example), then that's definitely where your problem is coming from.

Anyways, thanks for reporting this. I very much appreciate it. :)
My next commit will close this issue, but feel free to leave other comments or reopen if you still have problems.

from vek.

Thank you for your explanation! Yes, flipping my forward vector did in fact fix the issue, so it's exactly as you described. Could you release a new version so I can start using this right away? Thank you again :)

from vek.