Proper HMD scaling.

This commit is contained in:
Howard Stearns 2015-08-20 12:34:50 -07:00
parent 8d256a8e9c
commit f26849c7e1

View file

@ -846,7 +846,7 @@ void MyAvatar::sendKillAvatar() {
void MyAvatar::updateLookAtTargetAvatar() {
//
// Look at the avatar whose eyes are closest to the ray in direction of my avatar's head
//
// And set the correctedLookAt for all (nearby) avatars that are looking at me.
_lookAtTargetAvatar.reset();
_targetAvatarPosition = glm::vec3(0.0f);
@ -870,14 +870,39 @@ void MyAvatar::updateLookAtTargetAvatar() {
smallestAngleTo = angleTo;
}
if (Application::getInstance()->isLookingAtMyAvatar(avatar)) {
// Alter their gaze to look directly at my camera; this looks more natural than looking at my avatar's face.
// Offset their gaze according to whether they're looking at one of my eyes or my mouth.
glm::vec3 gazeOffset = avatar->getHead()->getLookAtPosition() - getHead()->getEyePosition();
const float HUMAN_EYE_SEPARATION = 0.065f;
float myEyeSeparation = glm::length(getHead()->getLeftEyePosition() - getHead()->getRightEyePosition());
gazeOffset = gazeOffset * HUMAN_EYE_SEPARATION / myEyeSeparation;
avatar->getHead()->setCorrectedLookAtPosition(Application::getInstance()->getViewFrustum()->getPosition()
+ gazeOffset);
// The camera isn't at the point midway between the avatar eyes. (Even without an HMD, the head can be offset a bit.)
// First find out where (in world space) the person is looking relative to that bridge-of-the-avatar point.
// (We will be adding that offset to the camera position, after making some other adjustments.)
glm::vec3 lookAtPosition = avatar->getHead()->getLookAtPosition();
glm::vec3 gazeOffset = lookAtPosition - getHead()->getEyePosition();
// Scale by proportional differences between avatar and human.
glm::mat4 leftEye = Application::getInstance()->getEyeOffset(Eye::Left); // Pose?
glm::mat4 rightEye = Application::getInstance()->getEyeOffset(Eye::Right);
glm::vec3 leftEyePosition = glm::vec3(leftEye[3]);
glm::vec3 rightEyePosition = glm::vec3(rightEye[3]);
float humanEyeSeparationInModelSpace = glm::length(leftEyePosition - rightEyePosition);
float avatarEyeSeparation = glm::length(getHead()->getLeftEyePosition() - getHead()->getRightEyePosition());
gazeOffset = gazeOffset * humanEyeSeparationInModelSpace / avatarEyeSeparation;
// If the camera is also not oriented with the head, adjust by getting the offset in head-space...
/* Not needed (i.e., code is a no-op), but I'm leaving the example code here in case something like this is needed someday.
glm::quat avatarHeadOrientation = getHead()->getOrientation();
glm::vec3 gazeOffsetLocalToHead = glm::inverse(avatarHeadOrientation) * gazeOffset;
// ... and treat that as though it were in camera space, bringing it back to world space.
// But camera is fudged to make the picture feel like the avatar's orientation.
glm::quat humanOrientation = Application::getInstance()->getViewFrustum()->getOrientation(); // or just avatar getOrienation() ?
gazeOffset = humanOrientation * gazeOffsetLocalToHead;
*/
// And now we can finally add that offset to the camera.
glm::vec3 corrected = Application::getInstance()->getViewFrustum()->getPosition() + gazeOffset;
avatar->getHead()->setCorrectedLookAtPosition(corrected);
} else {
avatar->getHead()->clearCorrectedLookAtPosition();
}
@ -1114,6 +1139,7 @@ void MyAvatar::renderBody(RenderArgs* renderArgs, ViewFrustum* renderFrustum, fl
getHead()->render(renderArgs, 1.0f, renderFrustum);
}
// This is drawing the lookat vectors from our avatar to wherever we're looking.
if (qApp->isHMDMode()) {
glm::vec3 cameraPosition = Application::getInstance()->getCamera()->getPosition();