Relation between every refresh rate, FOV and Steam suggestion for SS in a Pimax 5k+ series 203

Here is the relation between every refresh rate, FOV and SteamVR suggestion for supersampling/downsampling in a Pimax 5k+ series 203 (9 months use, with need to change stock cable due to snow pixels 4 months ago).

My config:
I7 7700k 4.5ghz
RTX 2080ti KFA2 SG edition OC 2000mhz
32gb ddr4

I´m using Pitools 1.0.1.253, firmware .263, with 1.0 image quality in pitools.

  • My analysis starts from 120hz, that has only normal and small FOV, but in pitools .253, the FOV seems to doesn´t change, giving me a FOV between normal and small in comparison to other refresh rates.

  • My analysis continues for 90hz, 72hz and 64hz. N (normal) is the image without parallel projection, and PP (parallel projection) is the image with parallel projection on. After N and PP, I described the SteamVR recommendation for supersampling/downsampling for each FOV in a decreasing way (large - normal - small- potato modes). Below this, I described the resolution in result of SS that SteamVR recommends for each FOV.

120hz

N: 98 (unique)
PP: 66 (unique)

gap: 32 (48%)

unique n: 2432 x 2636
unique pp: 2664 x 2704


90hz

N: 70 - 100 - 120 - 140 (large - normal - small - potato)
PP: 38 - 74 - 104 - 130 (large - normal - small - potato)

gap: 32 (84%) - 26 (35%) - 16 (15%) - 10 (7,6%) (gap between N and PP for large - normal - small - potato)

l: 3568 x 2228
lpp: 4284 x 2052
n: 3200 x 2664
npp: 3312 x 2864
s: 2888 x 2916
spp: 2840 x 3396
p: 2684 x 3152
ppp: 2532 x 3796


72hz:

N: 88 - 124 - 150 - 150
PP: 48 - 94 - 128 - 150

gap: 40 (83%) - 30 (31%) - 22 (17%) - 0

l: 4000 x 2496
lpp: 4812 x 2308
n: 3564 x 2964
npp: 3732 x 3288
s: 3228 x 3260
spp: 3152 x 3768
p: 2780 x 3260
ppp: 2720 x 4080


64hz

N: 100 - 140 - 150 - 150
P: 54 - 104 - 146 - 150

gap: 46 (85%) - 36 (34%) - 4 (2,7%) - 0

l: 4268 x 2664
lpp: 5104 x 2448
n: 3788 x 3152
npp: 3928 x 3396
s: 3228 x 3260
spp: 3364 x 4024
p: 2780 x 3260
ppp: 2720 x 4080

My conclusions:

  • The impact of Parallel projection is much more agressive in large FOV (170º).
  • In real life, large FOV with parallel projection on is almost impossible, since the resolution is low even in 64hz.
  • Parallel Projection seems to have much less negative performance impact by reducing FOV than by reducing refresh rate.
  • Perhaps the golden pattern to use PP on in higher refresh rates is using it with small FOV, with high resolution even with 90hz.
  • For those who wants to use at least normal FOV in anyway, 64hz with PP on in normal FOV is equivalent to 90hz PP on in small FOV in comparison of SteamVR SS recomendation.
  • SteamVR seems to block autoSS recommendation above 150%.
  • Taking this score sheet, we can see that a “diagonal move” between higher FOV without PP and smaller FOV with PP on for the same refresh rate seems to be closer in SteamVR SS recommendation.

Take your conclusions and describe it here.

Thanks, and sorry for my prolixity.

11 Likes

I have previously described why the resolution rendered goes up exponentially with FOV.
https://community.openmr.ai/t/new-5k-3-game-specific-issues-black-box-popping-parallel-projection-games-weird-shadows/13993/24

 
However, I was a bit lazy at the time and didn’t actually do the math. Also, I made a mistake and actually measured 130 degrees (Pimax Small FOV) instead of 150.

Starting with the Laws of Sines (lowercase = length, uppercase = angle):
law-of-sines

In our case:

  • C = 90 degrees (or pi/2)
  • b = same for all FOV
  • A = FOV / 2


 
Simplifying the equations yields:
The ratio of pixels rendered between two FOVs = (a2)/(a1) = tan(A2)*cot(A1)

Plugging in the values from my previous (estimated) example:

  • A2 = 130/2 = 65 degrees (Pimax normal FOV)
  • A1 = 90/2 = 45 degrees (Oculus Rift / HTC Vive FOV)
  • The ratio is 2.145 (close to my estimate of 2.15).

 
Large FOV to Normal FOV

  • A2 = 170/2 = 85 degrees
  • A1 = 150/2 = 75 degrees
  • The ratio is 3.063

As the FOV approaches 180 degrees the number of pixels increases exponentially, which explains why the ratio of 130 to 90 degrees (a 40 degree difference) is so much smaller than the ratio of 170 to 150 degrees (a 20 degree difference).

4 Likes

This topic was automatically closed 60 days after the last reply. New replies are no longer allowed.