Mirrors/overte-JulianGro
3
0
Fork 0
mirror of https://github.com/JulianGro/overte.git synced 2025-04-10 04:53:04 +02:00
Code Issues Projects Releases Packages Wiki Activity Actions

Bug fix for rotation culling in AvatarData::toByteArray()

Browse source 
The commit 5a0de0f103 inadvertently introduced a regression in how the AvatarMixer and clients encode AvatarData packets.
When encoding packets that with the cullSmallChanges flag set to true, we only include rotations that have changed significantly from the previous packet.
A bug was introduced that incorrectly preformed the comparison to detect changed rotations, this needs to handle the case when
two quaternions are almost equal but have negative signs.  Previously, this was preformed with an absolute value after the 4-component dot product.
In commit 5a0de0f103, this absolute value was removed, causing the culling to happen much more frequently.

This PR re-introduces that absolute value.
This commit is contained in:
Anthony Thibault 2018-10-03 12:54:49 -07:00
parent 4805d3564a
commit c0ae95f853
1 changed files with 1 additions and 1 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

2
libraries/avatars/src/AvatarData.cpp
View file

@ -594,7 +594,7 @@ QByteArray AvatarData::toByteArray(AvatarDataDetail dataDetail, quint64 lastSent
// The dot product for larger rotations is a lower number.
// So if the dot() is less than the value, then the rotation is a larger angle of rotation
if (sendAll || last.rotationIsDefaultPose || (!cullSmallChanges && last.rotation != data.rotation)
|| (cullSmallChanges && glm::dot(last.rotation, data.rotation) < minRotationDOT) ) {
|| (cullSmallChanges && fabsf(glm::dot(last.rotation, data.rotation)) < minRotationDOT) ) {
validity |= (1 << validityBit);
#ifdef WANT_DEBUG
rotationSentCount++;