Open robman opened 9 years ago
We should bring measureType
back if:
I'll close this. We can revisit this when we are aware of multiple independent implementations that are able to support measureType
.
This is critical for accurate reprojection and should really be included in v1.
Based on discussion with @huningxin this should probably be able to be requested in the constraints too (and be introspected/throw errors) as some devices let you configure if you want to use optical or principal ray.
Copying the quote from the comment here to define measure type: https://github.com/w3c/mediacapture-depth/pull/77#issuecomment-97066083
the measureType defines if the length is measured along the optical axis (depth down the X3 axis in the pinhole camera model) or as a ray (along the effective hypotenuse). Again this tells us what algorithm to use.
robman@, I understand that the use case for this is inverse linear relationship in disparity mode of RealSense R200 camera:
float depth_in_meters = scale / image[pixel_index];
For the reference, in z16 mode it behaves similar to Kinect2, with near = 0, far = 65535mm (scale == 1mm) and
float depth_in_meters = scale * image[pixel_index];
.
Are there other cases you know of? Thanks.
@robman I agree this is critical for accurate reprojection.
However, to my knowledge, the depth value is commonly OpticalAxis
measurement type, like RealSense and Kinect. Do you know any camera or SDK provide the OpticalRay
depth value?
If OpticalAxis
is the default one, can we specify the measurement of depth value in this spec is OpticalAxis
instead of adding measurementType
constraint? WDYT?
The discussion in #77 did highlight that this IS necessary - yet it got dropped 8( See: https://github.com/w3c/mediacapture-depth/pull/77#issuecomment-97066083 https://github.com/w3c/mediacapture-depth/pull/77#issuecomment-97083537