approved these changes Feb 28, 2019
Request discussion in the next telecon/FTF
and removed agenda
Request discussion in the next telecon/FTFlabels
Mar 4, 2019Closed
This value instructs the UA to prevent the values left in the depth buffer after a frame is rendered for compositing.
toji deleted the ignore-depth-values branch
March 7, 2019 23:44klausw added a commit to klausw/webxr that referenced this pull request
Mar 26, 2019For screen-based input, it seems overly specific to require that the ray origin must be at the near plane. If the system knows the actual position of the screen in 3D space, i.e. for a zSpace-style display or for smartphone AR, it would be preferable to use the actual mapping of the touch point into virtual world space for this. For example, if you use touch controls to add bubbles or other particle effects to the scene, it would look most natural to have those actually originate at the touched point. As an additional complication, immersive-web#548 changed the core spec to explicitly allow reversed-Z where `depthFar` is less than `depthNear`. If the actual screen plane isn't available, it would be more consistent to use the closer of the two instead of always using `depthNear`. As far as I can tell the spec itself doesn't currently require any particular depth for `XRTargetRayMode` `screen`. See also https://github.com/immersive-web/webxr/issues/489#issuecomment-455291966 - I think it's problematic to assume that the screen plane corresponds to the near plane.
Merged
toji pushed a commit that referenced this pull request
Mar 26, 2019For screen-based input, it seems overly specific to require that the ray origin must be at the near plane. If the system knows the actual position of the screen in 3D space, i.e. for a zSpace-style display or for smartphone AR, it would be preferable to use the actual mapping of the touch point into virtual world space for this. For example, if you use touch controls to add bubbles or other particle effects to the scene, it would look most natural to have those actually originate at the touched point. As an additional complication, #548 changed the core spec to explicitly allow reversed-Z where `depthFar` is less than `depthNear`. If the actual screen plane isn't available, it would be more consistent to use the closer of the two instead of always using `depthNear`. As far as I can tell the spec itself doesn't currently require any particular depth for `XRTargetRayMode` `screen`. See also https://github.com/immersive-web/webxr/issues/489#issuecomment-455291966 - I think it's problematic to assume that the screen plane corresponds to the near plane.