Anyone tryed ED Odyssey with the 8K X? I am curious how the performance is and also if the Aliasing problem improved over horizons. I am also using mainly my 8K X for Elite, great Immersion.
o7 Commanders!
Be careful, for most 3090 models, the DRAM on the back of the board does not have sufficient cooling.
Even if you donāt overclock the graphics memory, it can overheat, because some heat is transferred from the GPU on the other side of the boards. You should try to keep the DRAM temps below 95Ā°C (90Ā° would even be better).
1 Like
I played it (on my flat monitor) during the beta. The Aliasing problem was unchanged.
1 Like
Performance suffers rather a lot, even when not around any of the new stuff.
It is possible to run around (ā¦but not interact) in VR, by hopping into third person camera, but expect having to kick down your render resolution a lot if doing that, at least in anything wider than small mode (things stretch out quite beautifully in large mode, though) - things actually tick along fairly well when using the virtual cinema screen, but in full VR, your resolution and FOV settings really bear down on the work force under the GPU hood.
Like before, when frame rates get really low, things will jump, and spatter, and I donāt know what all to call it.
I think there is something bad about the ancient head tracking code in Elite (pretty sure there is rubberbanding, on top of the lag), which is exacerbated by the work load imparted by 8kX grade resolution requests, as well as Pimaxās own tracking issuesā¦
There is actually aliasing on more stuff in the game in general ā I am presuming many things that may not have been in the deferred rendering pipeline before, may be now.
The absolutely ridiculous aliasing that occurred on planets seen from space has been addressed though (they looked pretty much like TV static during the āalphaā preview period).
2 Likes
Did you see the note that was attached to the Hotfix 1 change notes where FD suggest:
Navigate to this file location: %LOCALAPPDATA%\Frontier Developments\Elite Dangerous\Options\Graphics
-
Make a backup of every file in the graphics folder, and place it somewhere safe for future reference.
-
Delete the original files in the Graphics folder.
-
Restart the game.
This was said to have improved performance for some users.
2 Likes
@neal_white_iii Thanks for your reply. I somehow thought that this will be the case,
unfortunately. I hope that FDEV will solve this in the future. The aliasing problem is a sad thing for ED VR, without it the immersion would be much better.
@jojon Thanks for your reply. Ok then I guess I will not buy odyssey until the performance is fixed. I am right now happy that I can use my 8K X with horizons on every aspect of the game without any stuttering and a good supersampling for best possible sharpness.
With Odyssey this is sadly the case, from what I am hearing from you guys and from redditā¦
LetĀ“s hope until console launch they will fixed it, otherwise horizons maybe will have also performance issue due to horizons will then get also the new planet tech and graphic updates.
Or how is your opinion on the situation that horizons will get in Q4 the new tech and so on? Will it make the performance of horizons worse?
@Troz Thanks, good to know for the future!
The problem is that they use a deferred renderer, which means that standard ātrueā AA techniques cannot be used. FDev already has a solution: Super sampling. Iāve played around with it and (unfortunately) to really fix the aliasing you need at least 3x super sampling (11520 x 6480) for a 4K flat screen. It looks great, but my framerate was in the low single-digits.
That means the ED aliasing problem will be solved for your 8KX once nVidia releases their RTX 90090 Ti (in 2045). 
2 Likes
@neal_white_iii That is interesting, good to know, thanks! I am already happy that I can use 1.5 super sampling, but yeah your alright, with 3.0 you would need a few next generation GPUs. Maybe FDEV will implement DLS in the future, but I think the hope for that is not so big.
Then I have to live with the aliasing problem - but I can overlock it 
1 Like
FYI: Just in case your GPU can handle it, Elite D maxes out at 2x super sampling, but you can add that to SteamVR or nVidia āDynamic Super Resolutionā (for flat monitors). I was able to reach 4x (on my 4K monitor with my RTX 2080) that way. It looked GREAT, but it was a slideshow. For VR, itās best to just increase PiTool Quality and SteamVR super sampling.
1 Like
@neal_white_iii Interesting, thanks for the tipp 
I am already at the edge of possible increasing of the supersampling. I using the rocommended settings from the spreadshit. .With 1.5 I had to downgrade some settings but it was woth it form me for the extra sharpness. With 1.25 I can crank up everything on ultra (good for SRV stuff).
But because I am mostly huntings bountys in space I am using 1.5 super sampling
LetĀ“s hope that odyssey gets soon a performance fix, so that I can enjoy the new planet tech in my SRV 
Also the stars lighting in odyssey is more realistic than in horizons. More realism is better for me in form of higher Immersion.
1 Like
@neal_white_iii do you know if there is an advantage to prioritising nVidiaās DSR or Eliteās supersampling? In other words, if my target were, say, 4x, would it be better to have Elite at 2x and nVidia at 2x, or Elite off and nVidia at 4x?
Unfortunately, this is more complicated than you might expect.
If you are in VR, you should first use PiTool Quality and then SteamVR super sampling (and not touch nVidia super sampling at all).
If you are using a flat screen, to reach a target of 4x youāll need to use both Elite (at 2x) and nVidia (at 4x). I know that seems wrong, but nVidia uses a different nomenclature: Number of pixels, not dimension multiplier. That is, 4x == 2x SS, because 2x X and 2x Y == 4x pixels.
As for quality, PiTool is better than SteamVR; nVidia is better than Elite (for flat screen). In both cases, this is because PiTool and nVidia do extra filtering for a higher quality result.
In particular, nVidia is a little slower than Elite, but the result is better, plus you can tweak the nVidia filtering (āDSR - Smoothnessā). I found that setting smoothness to a lower value (like 25%, instead of the 33% default) yielded a nicer result imo. A huge advantage of using Elite SS is that it can be adjusted in-game, so you can easily decrease it, when you are going to be on a planet surface or landing in a station.
In practice, I set PiTool Quality to the largest Quality value I can use in a low-framerate area (on a planet surface) and then increase Elite super sampling (if Iām just going to be in space).
If you are planning to do this, you should use some sort of FPS display to determine the maximum values you can use, without ruining your framerate. FPSvr for VR and FRAPS or Steam (not SteamVR, use Steam / Settings / In-Game / āin-game FPS counterā). I tend to use Steam instead of FRAPS, but sometimes the Steam display draws incorrectly (too big and in the wrong places). FRAPS only works with Direct-X 11.
TLDR; It doesnāt matter all that much which settings you use, but keep in mind that nVidia and SteamVR measure super sampling in pixels, while PiTools and Elite use dimension multipliers (4x pixels means 2x X and Y dimensions). Elite ss has the advantage of being adjustable in-game. Use some sort of on-screen FPS display to tweak your performance.
2 Likes
I worry, Iām afraid.
People probing around with profiling tools are seeing a complete dearth of optimisations, but even so I wonder how much can be gained, and this is not helped by a post from David Braben today, where he wrote that they had been working with a 30fps target frame rate, somehow being under the impression this is acceptable for anything more fast paced than a theme part management simulator.
ā¦and to be frank, I am on the fence on whether I wouldnāt prefer to keep the current Horizons planets; The new ones are not necessarily an improvement in every way ā or in ANY way, it feels like at timesā¦
1 Like
Does nothing, Iām afraid, but in that vein, one can of course in some ways reduce oneās graphical quality to the point frame rate becomes tolerable, but at which point one end up having to guess what is going on on screen, because there is certainly no way of resolving it. :7
Weāll see what FDev do over the coming weeks.
ā¦
One thing I quite like, is lighting being starker, and as such IMHO generally much more realistic, and giving a better sense of room.
However: It seems this may be a serendipitous positive side effect of oversights in bitmap typing and tone mapping, which result in way, way too much contrast, applied on the frame in full, as well as HDR mapping somehow working opposite to how it should (look toward something bright, and right now the whole screen will often brighten up, instead of normalising).
ā¦so whilst I do enjoy the now black space, stars looking less like cotton balls, and objects in space being brightly lit on the star-facing side and black on their shadow sides, these things seem to come as part in parcel with many things looking really bad, crunching to black and blowing out to white.
I rather expect the things I do like, to consequently revert to their previous state, when the above mentioned issues are addressed, fixing the bad balancing.
You were running the 8KX with the 1080TI all of this time? Nice 
Do you know which type of filtering PiTool does? All my tests led to the observation that there is no difference in the rendered resolution from the app (game) when you set an equivalent combinations of PiTool RQ and SteamVR SS.
1 Like
No, I donāt. I assume itās some sort of smoothing, which is what nVidia Dynamic Super Resolution does.
Iāve found that setting SteamVR super sampling to 100% and only setting PiTool Quality seems to yield a sharper image, at least to my eye. I think that thereās an extra resize involved, when SteamVR is not 100%. The advantage of using SteamVR ss is that it has a lot more steps, so you can tweak it to get the optimum framerate.
Well, if you set SteamVR to 100% and change PiTool RQ then you get the different render target resolution in the end and most likely different image quality as well.
I however understood you comment in a way that setting SteamVR SS and PiTool RQ gives a better result when PiTool RQ is set higher (which is a recurring theme on this forum from the first day), even if the render target resolution is the same.
For example: Setting SS=100% and RQ=1.0 is equivalent to SS=25% and RQ=2.0, both settings yield the same render target res and therefore produce exactly same image at the Pimax driver level.
Now, while I am sure of that (as I have tested it in many different ways), I can still accept that there is something happening to the image in Pimax compositor which makes one image look better than the other and this is what I believe you were trying to point out.
I would be quite interested in knowing what it is, and why it does not happen every time (regardless the RQ settings), because apparently it can process the exactly same image in two different ways (where one gives apparently better result) and tying it to the particular RQ setting does not make much sense (to me at least ).
2 Likes
That makes sense. Setting SS=25% isnāt possible through the SteamVR UI, which only allows even SS % values, so I havenāt tested that. Using RQ=2.0 might even provide more filtering at the PiTool / driver level. Because I only have a 2080, Iām generally running with RQ at 1.0 or 1.25 (with SS at 100%).
My main observation is that arbitrary SS % values seem to slightly degrade the resulting image. I have done a fair bit of testing, but it was by no means exhaustive. Iām simply sharing my observations.
Tweaking VR settings can be time consuming, so holding SteamVR at 100% reduces the testing permutations and results in a good image.
This topic was automatically closed 60 days after the last reply. New replies are no longer allowed.