VR amazing but I mean... It's bad too.
- Thread starter rdizz
- Start date
Many thanks for this !
Seem very reasonable to me. Personally I had 2nd thoughts about lowering SS just so I could increase HMD Quality and switched from 0.75/1.75 to 1.0/1.25. But if it's working for you then go with it. Overall, how does it seem compared to when you first posted here?
Last edited:
I think a lot depends on the headset you have, I have the Pimax 4k which gives a clearer picture than any of the others I've tried so far but it does put some pressure on the bridge of my nose. I have a 3d printer so I'm going to attempt to modify the straps to be a bit more supportive.
![[VR] Merkir](https://forums-cdn.zaonce.net/data/avatars/m/141/141091.jpg?1553527154){kind=avatar}
Same here. I see many people with SS < 1 and HMD > 1 which intuitively seems wrong, ie. reducing image resolution (quality) first, then up-sampling it again. I also use SS=1 and HMD=1.25/1.5, which shouldn't reduce image quality in either step. I've tried both methods and interestingly they give very similar results, but I still think down-sampling first gives a very slightly more blurred result. Theoretically this is what I'd expect, but perhaps there's a bit of a placebo effect in play here.
Try it, SS in-game mostly affects text and menus, in VR, but it hits the performance a ton. SteamVR SS does a better job, and gives you a clearer picture overall. HMD manages the image quality - so you leave that at 1.0, upscale through SteamVR. What you lose from the menus and text by lowering in game SS is more than made up for by SteamVR up scaling
Super Sampling isn't Super Sampling isn't Super Sampling... It's like compression/encryption/etc, the magic is in the code, and Steam have much more invested in VR than FD does. When the HMD slider appeared in Elite, you didn't need to use SteamVR for SS anymore, but Steam has released a bunch of updates since then and it's now a better option.
Ingame SS @.65
Ingame HMD @ 1.0
SteamVR SS @ 2.0
The above gives me a clearer picture, and better FPS than if I left SteamVR out of the equation and managed the settings in game. You'll find plenty of people who agree, after trying it, here
Last edited:
Interesting. I have the rift btw. Many times I've experimented with SS=0.65 or 0.75 with a correspondingly higher HMD of say, 1.75. As I mentioned, I see very little difference in the final result compared to SS=1.0/HMD=1.25. With the rift, the in-game HMD is the same as the Oculus Debug Tool's "Pixel Per Display" (which is what I used to use), so in effect there's no third setting in the rift, unlike the vive with it's extra SteamVR SS apparently.
But you've brought up a very important point - FPS. I may do some benchmarking. I'll test SS=0.75/HMD=2.0 vs SS=1.0/HMD=1.5, each of which should give the same final resolution, to see how the FPS stacks up.
Last edited:
Worth looking at, for sure. I was monitoring load/temps/voltage and FPS... But to be honest, for me moving the actual super sampling outside of Elite was like a penny dropping. I hadn't even changed any graphics settings, at the time, apart from moving the sampling to SteamVR and went from having asynchronous reprojection kick in at times, to turning it off altogether and maintaining 90FPS with no reprojection even enabled through SteamVR. It was that big a jump.
I am assuming that, similarly, Oculus put more time into the relevant bits of code than FD does, and expect a similar result if you move the workload over to your Oculus software. I'd like to hear how you went
Also, sorry I jumped into the thread half way and noticed the comment i replied to and associated it with Steam
Last edited:
Interesting preliminary benchmarking results. This is for the rift. It's far from being a scientific testing method because I'm simply not prepared to spend the time for more accurate profiling. I think these basic results are quite adequate for the current purpose, which is to determine whether down-sampling using SS gives inferior, the same or superior results compared to keeping SS=1.0.
Methodology: I turned off ASW, flew my ship near to the surface of an uninhabited moon, oriented my ship and brought it to a complete stop, ensuring no NPCs or any other movement/motion/dynamic effects. Everything in the viewport appeared static. I checked that the FPS was quite constant, varying mainly by the very slight positional tracking motion of my head. I repeated the same sequence of benchmarks three times to determine if there was any drift in results due to positional/tracking changes. There were no discernible FPS drifts. Here are the results:
SS=1.00 HMD=1.5 (ie. resultant multiplier = 1.5) gave 54 FPS
SS=0.75 HMD=2.0 (ie. resultant multiplier = 1.5) gave 54 FPS. No discernible change in image quality (subjective assessment ofc)
To check that the down-sampling and up-sampling changes were actually working, I then tried:
SS=0.65 HMD=2.0 (ie, resultant multiplier = 1.3) gave 74 FPS. There was a slight downgrade in image quality, as expected. That is, a lower multiplier (resolution) gives worse image quality, but better FPS.
Then, switching SS and HMD:
SS=1.50 HMD=1.0 (ie. resultant multiplier = 1.5) gave 54 FPS. With a discernible drop in image quality. This was the interesting result. The final multipier (resolution) was the same as the first two tests, but gave worse image quality.
Conclusion:
There appears to be little or no difference between the two methods as long as SS <= 1.0. The main test results (the first two, above) make sense with regard to forum feedback where many people believe SS < 1.0 works well, but probably just as many people believe SS=1.0 works well. My view now is that they give practically identical results as long as the final multipliers (resolutions) are the same.
It would be great to see others post their results.
Methodology: I turned off ASW, flew my ship near to the surface of an uninhabited moon, oriented my ship and brought it to a complete stop, ensuring no NPCs or any other movement/motion/dynamic effects. Everything in the viewport appeared static. I checked that the FPS was quite constant, varying mainly by the very slight positional tracking motion of my head. I repeated the same sequence of benchmarks three times to determine if there was any drift in results due to positional/tracking changes. There were no discernible FPS drifts. Here are the results:
SS=1.00 HMD=1.5 (ie. resultant multiplier = 1.5) gave 54 FPS
SS=0.75 HMD=2.0 (ie. resultant multiplier = 1.5) gave 54 FPS. No discernible change in image quality (subjective assessment ofc)
To check that the down-sampling and up-sampling changes were actually working, I then tried:
SS=0.65 HMD=2.0 (ie, resultant multiplier = 1.3) gave 74 FPS. There was a slight downgrade in image quality, as expected. That is, a lower multiplier (resolution) gives worse image quality, but better FPS.
Then, switching SS and HMD:
SS=1.50 HMD=1.0 (ie. resultant multiplier = 1.5) gave 54 FPS. With a discernible drop in image quality. This was the interesting result. The final multipier (resolution) was the same as the first two tests, but gave worse image quality.
Conclusion:
There appears to be little or no difference between the two methods as long as SS <= 1.0. The main test results (the first two, above) make sense with regard to forum feedback where many people believe SS < 1.0 works well, but probably just as many people believe SS=1.0 works well. My view now is that they give practically identical results as long as the final multipliers (resolutions) are the same.
It would be great to see others post their results.
Last edited:
Nice work! Very satisfying conclusion. The only thing I would rephrase is ..
"There appears to be little or no difference between the two methods as long as
In other words, I suspect the image quality of your final test looked worse, not because SS was greater than 1.0 but because HMD Quality wasn't.
Good pickup, and because it's impossible to tell which without further testing, I'll work out a suitable edit to my post.
Edit: After thinking about it, I'll leave it as is. It describes the (limited) results, with the acknowledgement that further testing may refine the conclusion. Note that I can't really say HMD >= 1.0 because the final test results say otherwise.
Last edited:
Very interesting and thanks for testing that. To make it perfect and complete we need the numbers for changing the SS Settings via the Oculus Debug Tool instead of the ingame controls
My comment must only affect Vive/SteamVR users. Here's another thread (only a page and a half) whereby another Vive user added 20~30 frames per second by using SteamVR instead of ED. Few comments from other Vive users who have the same success.
Ah well, as I said I didn't realize it was an Oculus thread. Sorry it doesn't help you guys, all the same
Ah well, as I said I didn't realize it was an Oculus thread. Sorry it doesn't help you guys, all the same
Last edited: