mirror of
https://github.com/overte-org/overte.git
synced 2025-04-20 12:04:18 +02:00
Merge branch 'master' of https://github.com/highfidelity/hifi into temp1
This commit is contained in:
commit
76ac8d7eec
1 changed files with 19 additions and 62 deletions
|
@ -3112,17 +3112,25 @@ void Application::renderRearViewMirror(const QRect& region, bool billboard) {
|
|||
_myAvatar->getOrientation() * glm::vec3(0.0f, 0.0f, -1.0f) * MIRROR_REARVIEW_BODY_DISTANCE * _myAvatar->getScale());
|
||||
|
||||
} else { // HEAD zoom level
|
||||
_mirrorCamera.setFieldOfView(MIRROR_FIELD_OF_VIEW); // degrees
|
||||
if (_myAvatar->getSkeletonModel().isActive() && _myAvatar->getHead()->getFaceModel().isActive()) {
|
||||
// as a hack until we have a better way of dealing with coordinate precision issues, reposition the
|
||||
// face/body so that the average eye position lies at the origin
|
||||
eyeRelativeCamera = true;
|
||||
_mirrorCamera.setPosition(_myAvatar->getOrientation() * glm::vec3(0.0f, 0.0f, -1.0f) * MIRROR_REARVIEW_DISTANCE * _myAvatar->getScale());
|
||||
// FIXME note that the positioing of the camera relative to the avatar can suffer limited
|
||||
// precision as the user's position moves further away from the origin. Thus at
|
||||
// /1e7,1e7,1e7 (well outside the buildable volume) the mirror camera veers and sways
|
||||
// wildly as you rotate your avatar because the floating point values are becoming
|
||||
// larger, squeezing out the available digits of precision you have available at the
|
||||
// human scale for camera positioning.
|
||||
|
||||
} else {
|
||||
_mirrorCamera.setPosition(_myAvatar->getHead()->getEyePosition() +
|
||||
_myAvatar->getOrientation() * glm::vec3(0.0f, 0.0f, -1.0f) * MIRROR_REARVIEW_DISTANCE * _myAvatar->getScale());
|
||||
}
|
||||
// Previously there was a hack to correct this using the mechanism of repositioning
|
||||
// the avatar at the origin of the world for the purposes of rendering the mirror,
|
||||
// but it resulted in failing to render the avatar's head model in the mirror view
|
||||
// when in first person mode. Presumably this was because of some missed culling logic
|
||||
// that was not accounted for in the hack.
|
||||
|
||||
// This was removed in commit 71e59cfa88c6563749594e25494102fe01db38e9 but could be further
|
||||
// investigated in order to adapt the technique while fixing the head rendering issue,
|
||||
// but the complexity of the hack suggests that a better approach
|
||||
_mirrorCamera.setFieldOfView(MIRROR_FIELD_OF_VIEW); // degrees
|
||||
_mirrorCamera.setPosition(_myAvatar->getHead()->getEyePosition() +
|
||||
_myAvatar->getOrientation() * glm::vec3(0.0f, 0.0f, -1.0f) * MIRROR_REARVIEW_DISTANCE * _myAvatar->getScale());
|
||||
}
|
||||
_mirrorCamera.setAspectRatio((float)region.width() / region.height());
|
||||
|
||||
|
@ -3149,58 +3157,7 @@ void Application::renderRearViewMirror(const QRect& region, bool billboard) {
|
|||
|
||||
// render rear mirror view
|
||||
glPushMatrix();
|
||||
if (eyeRelativeCamera) {
|
||||
// save absolute translations
|
||||
glm::vec3 absoluteSkeletonTranslation = _myAvatar->getSkeletonModel().getTranslation();
|
||||
glm::vec3 absoluteFaceTranslation = _myAvatar->getHead()->getFaceModel().getTranslation();
|
||||
|
||||
// get the neck position so we can translate the face relative to it
|
||||
glm::vec3 neckPosition;
|
||||
_myAvatar->getSkeletonModel().setTranslation(glm::vec3());
|
||||
_myAvatar->getSkeletonModel().getNeckPosition(neckPosition);
|
||||
|
||||
// get the eye position relative to the body
|
||||
glm::vec3 eyePosition = _myAvatar->getHead()->getEyePosition();
|
||||
float eyeHeight = eyePosition.y - _myAvatar->getPosition().y;
|
||||
|
||||
// set the translation of the face relative to the neck position
|
||||
_myAvatar->getHead()->getFaceModel().setTranslation(neckPosition - glm::vec3(0, eyeHeight, 0));
|
||||
|
||||
// translate the neck relative to the face
|
||||
_myAvatar->getSkeletonModel().setTranslation(_myAvatar->getHead()->getFaceModel().getTranslation() -
|
||||
neckPosition);
|
||||
|
||||
// update the attachments to match
|
||||
QVector<glm::vec3> absoluteAttachmentTranslations;
|
||||
glm::vec3 delta = _myAvatar->getSkeletonModel().getTranslation() - absoluteSkeletonTranslation;
|
||||
foreach (Model* attachment, _myAvatar->getAttachmentModels()) {
|
||||
absoluteAttachmentTranslations.append(attachment->getTranslation());
|
||||
attachment->setTranslation(attachment->getTranslation() + delta);
|
||||
}
|
||||
|
||||
// and lo, even the shadow matrices
|
||||
glm::mat4 savedShadowMatrices[CASCADED_SHADOW_MATRIX_COUNT];
|
||||
for (int i = 0; i < CASCADED_SHADOW_MATRIX_COUNT; i++) {
|
||||
savedShadowMatrices[i] = _shadowMatrices[i];
|
||||
_shadowMatrices[i] = glm::transpose(glm::transpose(_shadowMatrices[i]) * glm::translate(-delta));
|
||||
}
|
||||
|
||||
displaySide(_mirrorCamera, true);
|
||||
|
||||
// restore absolute translations
|
||||
_myAvatar->getSkeletonModel().setTranslation(absoluteSkeletonTranslation);
|
||||
_myAvatar->getHead()->getFaceModel().setTranslation(absoluteFaceTranslation);
|
||||
for (int i = 0; i < absoluteAttachmentTranslations.size(); i++) {
|
||||
_myAvatar->getAttachmentModels().at(i)->setTranslation(absoluteAttachmentTranslations.at(i));
|
||||
}
|
||||
|
||||
// restore the shadow matrices
|
||||
for (int i = 0; i < CASCADED_SHADOW_MATRIX_COUNT; i++) {
|
||||
_shadowMatrices[i] = savedShadowMatrices[i];
|
||||
}
|
||||
} else {
|
||||
displaySide(_mirrorCamera, true);
|
||||
}
|
||||
displaySide(_mirrorCamera, true);
|
||||
glPopMatrix();
|
||||
|
||||
if (!billboard) {
|
||||
|
|
Loading…
Reference in a new issue