Many users are getting higher image quality by setting the Pixels Per Display Pixel Override manually using the SDK tool.
But computers are supposed to automate things.
Why cant the oculus driver have a setting which enables automatic Pixel density that ramps up or down based on the load (to maintain 90fps) and to maximise image quality if there is some spare overhead..
With some user changeable maximum limits, this could set the rift another level higher in quality. Automatically.
A similar principle exists in the FlightSim world where a software tool (FSXFiber Accelerator) tweaks the game settings "on the fly" to maximise framerate vs image quality. The user can define priorities of quality vs framerate. While the rift will insist on 90fps constant, image Pixel Density can be raised and lowered depending on the available overhead.
On higher spec machines, the density can be raised to improve the picture automatically. Perhaps provide this as an "advanced" options for the savvy VR enthusiast.
In fact this is exactly what Oculus tell developers to do in the sdk docs. Developers should be exposing a setting for texel per pixel ratio and should ramp up or down the render texture viewport size to dynamically scale performance as needed.
You can't have the oculus driver ramp this value up and down while a game is running, because it requires you to modify how the graphics engine is rendering (render texture resolutions, viewport dimensions, etc),, it has to be done by the game developer. The reason the debug tool can change the ratio value is because it makes the function that tells the developer which resolution to use ignore one of it's parameters (replace the developer specified ratio with the user specified ratio). After that call, the game is not going to ask again to see changing values.
In Elite Dangerous, I can tweak the PixelDensity via the SDK tool and see the difference on the fly. Each time you enter or leave a hangar, the software must call / rebuild the viewport and picks up the Pixel Density change.
Is there a setting I can put in the netserver.cfg which will set the pixel density rather than having to use the SDKtool ?
I made the parts of this post that aren't the 100% on-topic, to be easy skippable. I won't delete it, cause it might help someone better understand what I meant to say and why I think what I think, so I hope "TL;DR" people are happy 😉 sorry for needlesly long text and a few repetitions, but it's 5AM when I type this.)
How many CPUs are fast enough to handle CV1 (including overclocked)? How many GPUs are fast enough (including overclocked)?
How much time it would take to do test on each configuration in appropriate (written by Oculus team) benchmark? How much effort it would take to use the results to create performance boundaries which could be then used for automatic settings? Dynamic res, draw distance, LoD etc. All could be set automatically.
Compared to the pluses that it would bring, it's nothing. Really. Instead of relying on just (very badly!) created "paper" list, Oculus should go this way.
[DIGRESSION] ----------------------- It's not a high cost for a company with budget as big as the one Oculus has. Reality turned out to be rather disappointing: https://forums.oculus.com/community/discussion/comment/396770/ PCs were always suffering (to a different degree depending on which "era" we talk about) because people who are laymans, cannot enjoy what their PCs can offer to the fullest. Well, that plus changing in marketing/PR tactics which now rely on consusing the layman even more rather than make it easy to choose the reasonable specs so gamers who are tech-laymans could assemble a gaming PC with good capabilities vs. cost ratio, instead of being put on a mine field, where most of them end up blowing up their wallets for very overpriced machines or 10% cheaper crap, which is 80% slower/less capable (Alienware PCs, Steamachines etc.) -----------------------
Back to the idea of creating a proper performance tables and automation of the settings. I could understand reasonable arguments about it being difficult when we talked about diversed sound card market. Or graphics card market. Or CPU market. But all that changes when we talk about VR. There's no Cyrix or VIA in the game anymore. A LOT of AMD CPUs and quite a lot of Intel CPUs are out of the race too. It is not as difficult now to create a benchmark, a result table, an algorythm and a ready solution for developers. Then all that would be left would be a test performed on end user's machine, just to check if results match, if the machine has any issues like antivirus software or anything else that might cause problems (from hard drive access to something running in the background, eating up the CPU or GPU etc.)
[End of on-topic post, the rest is just slightly related or something I already wrote in the past]
Of course, there will always be people who will say that this is hard. That's what they say when Ubisoft spends $200M for marketing of AC:Unity but PC port is outsourced to some third-world country located dev studio, which has nowhere near budget/abilities to create a proper port. Same people now cry about how bad it will be to have two PS4 versions. So earlier the problem was that there was SO MUCH different PC configurations, and now they scream and cry about just two, based on same GPU/CPU architectures. Yeah, sounds like a great idea to consider what they say most important. They are devs, after all, right? And everything the devs say is the best for everyone. That's why we should go back to DirectX 10 and resign from Vulkan, because some devs said it makes creating games too difficult and causes the game creation process to be more expensive. Or that we shouldn't think about 4K gaming, because the assets would have to be much bigger. I can go on with the examples, but let's stop here.
I fail to understand what makes big corporations fail so miserably when they try to work on PC gaming market. Steam machines were announced = I was sure it (solving the settings for laymans) will be done. It seemed obvious. Then Oculus promised to focus on PC VR and again I thought this is obvious. Apparently not. Not for corporations which don't really treat PC gamers seriously. And neither Valve or Oculus does, in my opinion (which isn't based on nothing). To some, what Oculus does is enough to believe they care. But I see lack of big "core games" and laughably low amounts of money pumped into core gaming in Oculus, non-existent in Valve, when relatively "tiny" Sony could do better despite having smaller budget for the VR effort than Oculus had, the Oculus who announced that it understands the importance of being the company that has the birth of whole VR on it's shoulders, and would do everything what makes VR better as fast as possible. And yet, we see Oculus not caring about PC VR at all. So I see no chance of ever seeing such solution (automated settings changed on the fly) or actually, feature. This would help a lot to make PC VR more attractive. But there was a lot of things that could help PC VR and were possible but Oculus choose not to go for them: - lack of Xone controller to lower the price - talks with AMD/Intel/Nvidia to stop ridiculous and harmful price craze - lower price of CV1. More units produced for the launch OR much better display.
(1070/1080 prices for VR are bad. VERY BAD. And 1080 could easily cost half of what it does if all the corporations didn't see PC gaming as a tiny market with no potential. IF they think just a tiny group of people buy fast cards, they assume majority of those people will pay whatever they are asked for, cause there are no alternatives. No cheap 144Hz monitors, no fast CPUs, no very fast GPUs have any real competition now (Polaris might very soon change the part about GPUs). That's why 1080 costs 200% of what it should, that's why Intel, having basically no competition in fast CPUs suitable for gaming, asks so much and limits OC capabilities, that's why Vive costs so much, and that's why CV1+Touch will cost a lot too. That's also why Oculus doesn't mind SLOWING DOWN PC VR progress by gimping the games to mobile and Xone/Scorpio level of performance. All the big corporations don't believe in PC VR. They treat it like a side project, something not important. That's why work on PC SDK was so slow, that's why mobile is the focus, that's why games aren't priority for Oculus anymore, that's why we got a laughable 1080p PenTile screen, and that's why there's the darn Xone controller included in the package. That's also why owners of both HMDs (Vive+CV1) have issues. Corporations should see the huge potential here, work together, cause 20% share of a market is better than 100% share of a market that's 50x smaller.
Valve proved with the Steam Machines how much they don't understand what are the things that make people game on PCs and like it. Oculus is no better. PC VR is just ignored. Treated without any faith, it's the same problem as one that was so obvious when a certain interview with MS employee became famous, after the MS person accidentally practically admitted that for MS PC gaming means games like Minecraft, casual games and maybe some of those big games like WoW, DOTA 2 etc. those which make enough profit to show up on Microsoft's radar.
Microsoft doesn't care about PC VR gaming, Valve doesn't, Facebook doesn't. Nvidia and AMD (at least until now) doesn't either. I think realizing that might help you guys better predict what Oculus can/will do and why some good ideas aren't happening.
Not an Oculus hater, but not a fan anymore.
Still lots of respect for the team-Carmack, Abrash.
Oculus is driven by big corporation principles now. That brings painful effects already, more to come in the future. This is not the Oculus I once cheered for.
Hehe, I reckon all games should have dynamic settings (as an option) Nvidia GeForce Experience kinda does this, but not on the fly. It just changes game settings based on a pre-calculated reference table.
Perhaps its fun for many to tweak manually (and that option should always be available), but some in-game automated feature would also be a great option, especially for those games which have varying scenery and other complexities which result in a lowest common denominator "fixed" setting rather than something dynamic that maximizes the quality based on the "current" demands.
The http://www.thefsps.com people are doing exactly this as an addon to the Flight Simulator and it works very well.
In open areas, things automatically crank up, to improve the asthetics. Approach an airport and things crank down just enough as required to maintain a user defined framerate. The user sets some performance boundaries and the rest happens automatically maximizing the user experience. And most of the time, the user is completely unaware that the scenery options are being altered. (Again, better than picking a low fixed setting for the heavy areas that gets applied everywhere).
Elite Dangerous is one VR enabled app that could greatly benefit from this approach as you can crank things up when in the stars, but have to wind down in the space stations and planets.
We need some fancy name for this that marketing can latch onto as a "feature". HyperDynamic UltraVision or something like that 😉 Then the idea might get some traction.