Interesting comment by Alan Yates on frensel lenses and other issues

I guess most of you already read it, but for those who didn’t:

i would really like if Pimax could comment frankly and extensively on their view on these matters, especially regarding to these bits:
“It is technically difficult to make frensel lenses with low stray light by injection molding, you also have to be careful about spatial frequencies and a bunch of other important details. The Vive panels are brighter than other HMDs so we actually did quite well to keep the stray light to the levels you experience in the Vive lens. This is one of the many knobs HMD makers can use to deal with the various trade-offs in the optical system; turn down the brightness, soften the contrast, adjust the lens MTF, reduce the FOV, shrink the eye-box.”
“you need to measure the radial distortion of the lens, for each wavelength of the display light emitters and compute the distortion polynomial coefficients for each channel. Plus if you want it to be really correct you need to deal with the pose between the lens and the panel not being perfectly parallel, and the lens center not being on the central pixel of the display - which matters a lot at the edges, you actually need sub-pixel centering to do it properly.”
and especially:
“They were the only practical lens technology for hitting the overall set of optimisations we wanted, especially minimising eye-position dependent distortion with a single element.”
EDITED to add one more quotation:
“The calibration machine is pretty simple, a calibrated camera with the right lens, but the software and the maths to do it properly is non-trivial. Sure you can just fiddle with the coefficients until it looks right - I know some HMD manufacturers do this and some that use the same calibration for every HMD. This may be OK if they can hold tolerances or their tracking or lens MTF is so crap it doesn’t matter anyway.”
@deletedpimaxrep1
I think Alan’s comment may be interesting for Pimax optical engineers too.
do Pimax acknowledges and takes care of all the issues stated here? are those related to the problems of the v5 lenses? Are the new lenses for M1 able to minimize the eye position dependent distortion for a 200° FOV? if yes, how? i know you already wrote that all previous problems are fixed with the new lenses but can you give us as much details as possible?

9 Likes

I’ve been yammering on about this here:

http://community.openmr.ai/t/why-are-these-idiots-making-these-mistakes