[gamepad-controls] don't copy camera rotation, copy camera rig rotation instead #396
Conversation
|
I tested with the Quest joysticks only. |
|
You can reproduce the first issue if you have rotation="0 90 0" on camera rig for example. |
|
Oh I see that previously the code changed the camera rotation, not the camera rig rotation. The behavior changed in https://github.com/n5ro/aframe-extras/pull/373/files#diff-87251e16fae2ca8e4158c45c3374b3446d1f0a9546f2cfac67a43f416163b623 |
…ra rig rotation instead
vincentfretin
force-pushed
the
gamepad-controls-rotation
branch
from
349a297 to
b7cab0c
Compare
|
I rebased since I merged another fix. You can test with the build of this PR: I'll merge those changes in a few days if I don't have any feedback. |
| // First copy camera rig pitch/yaw, it may have been changed from | ||
| // another component. | ||
| yaw.rotation.y = this.el.object3D.rotation.y; | ||
| pitch.rotation.x = this.el.object3D.rotation.x; |
There was a problem hiding this comment.
Note that this is coherent with what we do in init().
This closes #394
It doesn't make sense to me to copy the rotation from camera to the camera rig. That's why we have the issue that the camera jumps when first touching the joystick (first issue), and second issue it jumps again if some other component modified the camera rig rotation because it wasn't copied before doing the calculation.
My guess is that code was written/thought when we didn't use camera rig, if movement-controls and camera with look-controls was on the same entity, the code made sense. That's not the case today.
I may have overlooked something, so if someone can review that and test it, I'll appreciate it.
Changes proposed: