• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

What do gamers actually think about Ray-Tracing?

1080p DLSS Balanced would be 720 right? or is it lower

A bit lower, around 1114x626.
Attached is a list of resolutions from Alan Wake 2 (since it's say the internal resolution, is easier to have the exact numbers, probably will be the same all around).

1440p and 1620p I have them as DLDSR while 1871p and 4k just regular DSR.

However, with stuff like DLSS Tweaks you can play around and change the ration, ultimately changing those resolutions.

 
Arguing for "selective use" of nanite is pointless, you either go all in or not all - perhaps better to just stick cu regular tessellation then. I really don't care for such optimization which is basically staying with the old, withhold progress. Like I've said, to me it's either full or nothing.
Ok, so your argument pretty much boils down for devs to not do their job (optimising game and using what works when it should be used) and just be lazy. What I don't understand is what do YOU as a gamer hope to get out of that? Because you won't get cheaper (and better) games and you won't get more FPS with better visuals. So what will you get, a "Nanite" sticker to feel better? :)

TI isn't optimizing like that, he's getting firmly stuck in the past. :)
Absolutely not true, which you'd know if you actually watched their vids. They use UE5, they talk about various modern methods used to make games both look good and work properly on wide rage of hardware and not just top one. If you call that stuck in the past, I don't know what to tell you. :)
I haven't seen any issues with the leaves, do you have a screenshot or a short clip?
You said yourself, you use Supersampling with DLSS from 1080p - that's higher base resolution than it is when set to Quality in 1440p. Hence, no wonder you have less issues with noise and loss of details. The higher the resolution the less artefacts with RT/PT. That said, not installing whole game again to take a screenshot.
Metro Exodus EE runs faster and looks better than the original release. If there is room for similar optimization, then it should be done. The idea is to code taking advantage fully of the current hardware and new opportunities.
Current hardware is not designed for PT, never was - that came later, being pushed by NVIDIA. Don't confuse base RT with PT. For full, proper PT we need differently designed hardware. For crutches you might need soon 5000 series to run all the fancy new AI, just hope your wallet is thick enough too.
The engine/game (Indiana Jones, I mean), isn't quite perfect. The devs said with DF that the water for instance, has part of the code for PT reflections, but with everything going on in the jungle, performance isn't there yet ( ergo, they're not enabled), so it needs further polish that they hope will come (relative) soon. On the same train of thought, perhaps other stuff like latency can be improved, too.
But why wait? You just said earlier, they should've just enabled it all and move on, there's no need for optimization, just turn it on! :)
 
Last edited:
Nothing complex in the vanilla game, but I've been doing some modding including testing vegetation, etc.
Fair enough, but that explains why it might work badly - engine's been designed mostly for indoor environment, as that's what the main game based on it is.
Some games have started to implement Ray Reconstruction but it is early days yet, same for stuff like RESTIR but there is a lot of working going on behind the scenes which will eventually make it into games - some guy has a fork of Quake 2 RTX which reduces noise considerably. Increasing ray count all helps even a relatively small increase can make a big difference with the right algorithms in place.

EDIT: The biggest cheat with Quake 2 RTX is that only the sun light does multiple bounces with full colour transfer, etc. all other lights are running much more constrained.
It all sounds to me like "It kinda works, but it's more of a preview of what we could see in the future, just much better." - the questions I would have are: how far ahead is said future and how much for the xx80+ card to run it properly then. :) Don't get me wrong, I have finished whole Q2 RTX and enjoyed it for what it was, but I was able to see all kinds of issues with that implementation. Similar thing to CP2077 with PT - it was really bad till RR came, with RR very playable but still many shortcuts and issues visible, as RR is far from perfect. More development needed, along with faster hardware. We're not there yet. And that is the best of the best with PT in games, I believe, everything else is just worse still.
 
This scene in CP2077 made me wonder what I was missing out on having RT turned off. I'm running optimised settings using Screen Space Reflections at Medium (rocking 117 fps below)

JOvWWwK.jpeg


So I turned on all the RT + set RT lighting to Ultra (resulting in 47 FPS), and was rather 'surprised' by the result:

7PVMqTM.jpeg


So then I thought, okay, maybe realistic lighting means you don't get the lit up plant reflections at this angle (didn't even register the lighting from the pillars). Lets check path tracing:

wzfIyts.jpeg


Errr... once again surprised seeing that Path Tracing resembles Screen Space Reflections Medium (but mostly worse quality) rather than RT Ultra. Getting 24 fps here :cry:

Full screen images (FPS counter should confirm I'm not BSing):

This has cured my RT FOMO for the foreseeable future lol
 
Fair enough, but that explains why it might work badly - engine's been designed mostly for indoor environment, as that's what the main game based on it is.

It all sounds to me like "It kinda works, but it's more of a preview of what we could see in the future, just much better." - the questions I would have are: how far ahead is said future and how much for the xx80+ card to run it properly then. :) Don't get me wrong, I have finished whole Q2 RTX and enjoyed it for what it was, but I was able to see all kinds of issues with that implementation. Similar thing to CP2077 with PT - it was really bad till RR came, with RR very playable but still many shortcuts and issues visible, as RR is far from perfect. More development needed, along with faster hardware. We're not there yet. And that is the best of the best with PT in games, I believe, everything else is just worse still.

Quake 2 RTX is basically a preview of the technology - even with the issues it still does better ray tracing (path tracing) than any game out as even CP2077 and the new Indiana Jones game, etc. are still using a lot of non-ray tracing shortcuts and making use of low resolution off-screen rendering, etc. but Quake 2 RTX was developed against the tech then and the GPUs then - there is a lot of improvements possible when you have the performance of 4000 series GPUs and progress in areas like de-noising since.
 
This scene in CP2077 made me wonder what I was missing out on having RT turned off. I'm running optimised settings using Screen Space Reflections at Medium (rocking 117 fps below)

JOvWWwK.jpeg


So I turned on all the RT + set RT lighting to Ultra (resulting in 47 FPS), and was rather 'surprised' by the result:

7PVMqTM.jpeg


So then I thought, okay, maybe realistic lighting means you don't get the lit up plant reflections at this angle (didn't even register the lighting from the pillars). Lets check path tracing:

wzfIyts.jpeg


Errr... once again surprised seeing that Path Tracing resembles Screen Space Reflections Medium (but mostly worse quality) rather than RT Ultra. Getting 24 fps here :cry:

Full screen images (FPS counter should confirm I'm not BSing):

This has cured my RT FOMO for the foreseeable future lol

PT/RT should draw realistic reflections, meaning that depending on the angle and other variables, you could get realistically less detailed (but correct) reflections - in your case, assuming there's not some sort of a bug.

Other than that, it depends, I'd say you see the difference easily.

 
Last edited:
Ok, so your argument pretty much boils down for devs to not do their job (optimising game and using what works when it should be used) and just be lazy. What I don't understand is what do YOU as a gamer hope to get out of that? Because you won't get cheaper (and better) games and you won't get more FPS with better visuals. So what will you get, a "Nanite" sticker to feel better? :)


Absolutely not true, which you'd know if you actually watched their vids. They use UE5, they talk about various modern methods used to make games both look good and work properly on wide rage of hardware and not just top one. If you call that stuck in the past, I don't know what to tell you. :)

You said yourself, you use Supersampling with DLSS from 1080p - that's higher base resolution than it is when set to Quality in 1440p. Hence, no wonder you have less issues with noise and loss of details. The higher the resolution the less artefacts with RT/PT. That said, not installing whole game again to take a screenshot.

Current hardware is not designed for PT, never was - that came later, being pushed by NVIDIA. Don't confuse base RT with PT. For full, proper PT we need differently designed hardware. For crutches you might need soon 5000 series to run all the fancy new AI, just hope your wallet is thick enough too.

But why wait? You just said earlier, they should've just enabled it all and move on, there's no need for optimization, just turn it on! :)

There's one thing to optimize the code on which the Nanite tech works, to get it faster (better algorithm), and another to not use Nanite at all or use it sparely. It will be akin to using just reflections or shadows RT instead of PT or, at least, a collection of multiple RT effects if not PT. That's fine if TI has the time to optimize that to a fault (realistically, not always possible), but also it should provide full quality assets alongside Nanite for those that can run it. If those assets are not available / made due to his "optimization"... not good.

I don't know what you mean by white noise, it doesn't strike me directly even at 1080p quality DLSS (which is 720p) or Supersampling as you say, from higher resolution but similar input/internal res to around 720p. There is more "blockiness" in the vegetation indeed, perhaps is that. Disabling Sun shadows I think it helps. An optimization pass here done by TI would be that - disabling sun shadows for the jungle or not use RT/PT (if it's actually that) at all. Assuming you can't get it better looking the normal way (the vegetation I mean), then allow the player to increase the rays, from the Sun in this case, and see if it can compensate.

I assume that, at least in part, some issue with the vegetation comes from the subsurface scattering and why it getts better with Sun shadows off (or it could be just placebo). Can't remember seeing such issues in CB77, but I could be mistaken.
 
Quake 2 RTX is basically a preview of the technology - even with the issues it still does better ray tracing (path tracing) than any game out as even CP2077 and the new Indiana Jones game, etc. are still using a lot of non-ray tracing shortcuts and making use of low resolution off-screen rendering, etc. but Quake 2 RTX was developed against the tech then and the GPUs then - there is a lot of improvements possible when you have the performance of 4000 series GPUs and progress in areas like de-noising since.
I see last patch released with fixes and improvements is from April 2023. It's more like it started as a preview but got polished over time with multiple fixes, some of them include better denoising, better CPU performance, better AA, upscaling, new Vulkan extensions etc. Just out of curiosity I've run it now on my 4090 and performance and noise are considerably better than I've seen whilst playing it soon after 4090 release (about 120FPS with all details to the max, no upscaling, QWHD res). Quite a bit of improvement for just a preview. It doesn't change the fact that it's an ancient game, putting very little strain on the modern CPUs. Unlike modern games. PT/RT or not, GPU still needs to calculate whole geometry etc. - there's not even a comparison of performance required for Q2 vs CP2077, for example. Keeping on mind that CP2077 is a flagship PR game of NVIDIA, with plenty of monies and expertise poured into it, I'd say this is the best they can do on 4000 series, top of the line. Hence, all the shortcuts they had to do is telling about the performance they were able to squeeze out of the current best GPUs.
 
There's one thing to optimize the code on which the Nanite tech works, to get it faster (better algorithm), and another to not use Nanite at all or use it sparely.
Or, use it selectively when it matters and makes a difference. Like you know, standard practice in programming since the dawn of computers? You seem to assume somehow modern technology excuses ignoring basic ideas about programming, it really doesn't. One always use what makes sense when it makes sense - be it programming, or manufacturing. That's how you achieve the goal, like for example proper level of protection without increasing a weight too much in a car or an airplane, etc. It's not different from producing a game, same rules apply. The difference is that in case of cars and airplanes we have legal requirements and plenty of organisations overseeing things. In gaming we have just end customer deciding if they want to spend monies on whatever or demand more.

It will be akin to using just reflections or shadows RT instead of PT or, at least, a collection of multiple RT effects if not PT.
And then you get I.J. game which has PT yet you can actually pick and choose what you want from that. So, how is that actually a PT if you can turn off bits and bobs of it? How is that different? In traditional PT you can in theory just regulate number of bounces and samples per pixel, not turn on/off whole effects at will. And yet devs did that - so it can be done and it's not going to the past. Even PT is tweakable, so is Nanite and other bits of UE - as a dev, their job is to massage it into proper shape, not just flip a switch and forget about it. Your argument simply makes no sense to me at all.

I don't know what you mean by white noise
Simply lights artefacts between moving leaves - doesn't exist without DLSS, exists with DLSS. If you look online at articles, you'll find quite a few where reviewers noticed the same in multiple places and advise to just use TAA in that game, as it produces often better results than DLSS.

Assuming you can't get it better looking the normal way (the vegetation I mean), then allow the player to increase the rays, from the Sun in this case, and see if it can compensate.
In this case it's the fault of DLSS - it works well with raster, it fails with RT/PT because it only upscales a final image after rendering and the less input pixels you get to use in RT/PT the more artefacts you get. So it can only upscale all these artefacts and introduce more. It needs a different AI algo beforehand to improve the image before it gets upscaled, it feels.
I assume that, at least in part, some issue with the vegetation comes from the subsurface scattering and why it getts better with Sun shadows off (or it could be just placebo). Can't remember seeing such issues in CB77, but I could be mistaken.
I've not seen this particular issue in CP2077.
 
I see last patch released with fixes and improvements is from April 2023. It's more like it started as a preview but got polished over time with multiple fixes, some of them include better denoising, better CPU performance, better AA, upscaling, new Vulkan extensions etc. Just out of curiosity I've run it now on my 4090 and performance and noise are considerably better than I've seen whilst playing it soon after 4090 release (about 120FPS with all details to the max, no upscaling, QWHD res). Quite a bit of improvement for just a preview. It doesn't change the fact that it's an ancient game, putting very little strain on the modern CPUs. Unlike modern games. PT/RT or not, GPU still needs to calculate whole geometry etc. - there's not even a comparison of performance required for Q2 vs CP2077, for example. Keeping on mind that CP2077 is a flagship PR game of NVIDIA, with plenty of monies and expertise poured into it, I'd say this is the best they can do on 4000 series, top of the line. Hence, all the shortcuts they had to do is telling about the performance they were able to squeeze out of the current best GPUs.

The path tracing system used in Quake 2 RTX doesn't massively benefit from it being an ancient game - much of the performance is eaten into by overheads which are the same almost regardless of scene complexity until you are dealing with things like advanced caustics, etc. which no games are doing currently.

This is just stuff I'm experimenting with in the engine so lots of issues and things you can pick holes with - I'm very limited by my artistic ability and knowledge, and the renderer has a lot of limits in its material system, etc. due to not being required by the original game and still has quite a lot of engine limits due to the underlying game engine still being 25 years old but it is far more capable than the original game levels demonstrate:

oUEpVGH.jpeg


K1itxtm.jpeg


And this is basically a first viable preview of the technology from several years back now - things have moved on quite a bit since in terms of what is viable now.

EDIT: In the right conditions the glasses will actually have crude caustics rendered but I don't have a setup (lighting angles and nearby surfaces, etc.) in my test map to show it properly and don't have time at the moment to set something up.
 
Last edited:
Or, use it selectively when it matters and makes a difference. Like you know, standard practice in programming since the dawn of computers? You seem to assume somehow modern technology excuses ignoring basic ideas about programming, it really doesn't. One always use what makes sense when it makes sense - be it programming, or manufacturing. That's how you achieve the goal, like for example proper level of protection without increasing a weight too much in a car or an airplane, etc. It's not different from producing a game, same rules apply. The difference is that in case of cars and airplanes we have legal requirements and plenty of organisations overseeing things. In gaming we have just end customer deciding if they want to spend monies on whatever or demand more.


And then you get I.J. game which has PT yet you can actually pick and choose what you want from that. So, how is that actually a PT if you can turn off bits and bobs of it? How is that different? In traditional PT you can in theory just regulate number of bounces and samples per pixel, not turn on/off whole effects at will. And yet devs did that - so it can be done and it's not going to the past. Even PT is tweakable, so is Nanite and other bits of UE - as a dev, their job is to massage it into proper shape, not just flip a switch and forget about it. Your argument simply makes no sense to me at all.


Simply lights artefacts between moving leaves - doesn't exist without DLSS, exists with DLSS. If you look online at articles, you'll find quite a few where reviewers noticed the same in multiple places and advise to just use TAA in that game, as it produces often better results than DLSS.


In this case it's the fault of DLSS - it works well with raster, it fails with RT/PT because it only upscales a final image after rendering and the less input pixels you get to use in RT/PT the more artefacts you get. So it can only upscale all these artefacts and introduce more. It needs a different AI algo beforehand to improve the image before it gets upscaled, it feels.

I've not seen this particular issue in CP2077.
IJ is a bit much to be called PT anyway since it misses reflections. It just one more game with ray tracing effects to various degrees. In general, GI makes the most difference.

It's funny to me, the need to optimize for something less than current gen can do, especially for games coming out in 2-4 years or more. Look below at their initial demo running on a PS5. At 5:58 you have a statue, directly from z brush, with more than 33 million triangles. Then in the next room you have almost 500 hundred of those statues for about 16 billion triangles from them alone. Runs great on the original PS5.

What you had and have to optimize is where nanite cannot be applied yet.

But, if TI wants to go old style, good luck!

 
IJ is a bit much to be called PT anyway since it misses reflections. It just one more game with ray tracing effects to various degrees. In general, GI makes the most difference.

I agree, yet they call that patch PT, not my invention. :)

It's funny to me, the need to optimize for something less than current gen can do, especially for games coming out in 2-4 years or more.

Current gen includes 4060 class cards and xx60 class are the majority. Ignoring that fact, we see more and more high refresh rate monitors out there, with even over 400Hz now - high FPS in games is more desired now than in the past apparently, since there's a market for such expensive monitors. Ergo, optimization is for all GPUs, not just top 1% for 60fps only.

Look below at their initial demo running on a PS5. At 5:58 you have a statue, directly from z brush, with more than 33 million triangles. Then in the next room you have almost 500 hundred of those statues for about 16 billion triangles from them alone. Runs great on the original PS5.

That's not Nanite itself, that's also geometry instancing - the real calculation happens for only one statue. 3D Mark had a benchmark for that for a while now. Now, put 500 different meshes, so instancing doesn't work anymore, and you will see how FPS drops on its face. Then check how many triangles are really needed for good enough details of this statue, where it still makes real difference, cut down the mesh to that. And suddenly you can run a lot of different meshes that don't kill FPS so much anymore, with proper quality, that use less memory and space on the drive too. It's basic optimisation really, to not waste space and performance for nothing. You don't have to turn off Nanite, you just have to use it with head and not flip the switch on and forget about it.

That aside, highly optimised demo showing one scene is really nothing like actual games are. Might as well say 3D Mark tech demos are a good indicative of how games behave, which was never true. :)

What you had and have to optimize is where nanite cannot be applied yet.

But, if TI wants to go old style, good luck!

What Devs need to do is to follow basic guidelines of UE and ML Devs - so far, as witnessed in many games and demos on ue5 engine, they often do not do even that. Your arguments make really little sense when even said ML Devs ask developers in their post to finally please follow performance optimisation guidelines and not just be lazy about it. That also, by the way, increases image quality and not just performance - bad practices introduce noise and attracts where it didn't have to be.
 
The path tracing system used in Quake 2 RTX doesn't massively benefit from it being an ancient game - much of the performance is eaten into by overheads which are the same almost regardless of scene complexity until you are dealing with things like advanced caustics, etc. which no games are doing currently.

This is just stuff I'm experimenting with in the engine so lots of issues and things you can pick holes with - I'm very limited by my artistic ability and knowledge, and the renderer has a lot of limits in its material system, etc. due to not being required by the original game and still has quite a lot of engine limits due to the underlying game engine still being 25 years old but it is far more capable than the original game levels demonstrate:

oUEpVGH.jpeg


K1itxtm.jpeg


And this is basically a first viable preview of the technology from several years back now - things have moved on quite a bit since in terms of what is viable now.

EDIT: In the right conditions the glasses will actually have crude caustics rendered but I don't have a setup (lighting angles and nearby surfaces, etc.) in my test map to show it properly and don't have time at the moment to set something up.
Out of curiosity, if you have any experience with that, how do you compare it to PT used in Portal RTX, which was another Nvidia playground?
 
Out of curiosity, if you have any experience with that, how do you compare it to PT used in Portal RTX, which was another Nvidia playground?

Haven't played Portal RTX, think I've got it on my account though so might download it after work and have a play.
 
PT/RT should draw realistic reflections, meaning that depending on the angle and other variables, you could get realistically less detailed (but correct) reflections - in your case, assuming there's not some sort of a bug.

Other than that, it depends, I'd say you see the difference easily.


The screenshot you've provided doesn't look any more impressive than some sort of ReShade.

I've now seen some dramatic Cyberpunk graphic overhauls on Youtube, and there's no mention of any of them being dependent on RT / PT.

I'm wondering what could've been achieved if the last eight years had been spent building on raster and general lighting fakery, rather than surrendering to RT. Imagine that Screen Space Reflections reflected 50% more than what's immediately bottom or below the visual field displayed on screen. Imagine AI being trained to conjure up a likely reflective image (rather than there being 0 reflections at all with RT off), as in CP2077 when you look at one of its low res tinted windows. Laughably I saw a realistic reflection of the background behind me with RT/PT on, while the interior that was coming though was of some dude at a slot machine that looked about 320x200, ruining it.
 
Last edited:
The screenshot you've provided doesn't look any more impressive than some sort of ReShade.

I've now seen some dramatic Cyberpunk graphic overhauls on Youtube, and there's no mention of any of them being dependent on RT / PT.

I'm wondering what could've been achieved if the last eight years had been spent building on raster and general lighting fakery, rather than surrendering to RT. Imagine that Screen Space Reflections reflected 50% more than what's immediately bottom or below the visual field displayed on screen. Imagine AI being trained to conjure up a likely reflective image (rather than there being 0 reflections at all with RT off), as in CP2077 when you look at one of its low res tinted windows. Laughably I saw a realistic reflection of the background behind me with RT/PT on, while the interior that was coming though was of some dude at a slot machine that looked about 320x200, ruining it.
The screen shot was a simple way to show that there are difference, more or less obvious, which makes the lighting look realistic compared to a "game-looking" one with raster.

Reshade with RT done in screen space (or any other mode that does it in screen space), is limited by... screen space, just like SSR are. Some look better due to color grading, basically to why the movies look better than your standard clip done with your phone, camera, etc.

I don't see why you couldn't render a 50% larger image, outside of player's FOV. The player gets the crop version for what he sees and you use the rest for more room to play in screen space. The AI is already here to a point, DLSS + frame gen, which will help somewhat with the extra burden of performance needed (since you render 50% more). But... you still run into the same problems of screen space. Everything that is behind you or outside of the view of the image rendered (what you see 50%), won't be captured and won't add info to current frame. Only what's ahead.

If the AI would know the entire scene, then he'll render directly the entire scene, not just what's outside your view. We're probably heading that way, but still not really there just yet.

Reflections on windows are tradeoff. I know Tinek will call them lazy, but you can't (realistically talking) have the entire city's buildings complete, with interiors for large, open cities. You could do some random generation of stuff, but that's more to engine general ability / gameplay elements, not related necessarily with graphics advancements.

Going back to the first paragraph and my example, I think those large billboards are large area lights, which are (apparently) rather impossible to do correctly in raster to have shadows that correspond with their shape.
 
Last edited:
I agree, yet they call that patch PT, not my invention. :)



Current gen includes 4060 class cards and xx60 class are the majority. Ignoring that fact, we see more and more high refresh rate monitors out there, with even over 400Hz now - high FPS in games is more desired now than in the past apparently, since there's a market for such expensive monitors. Ergo, optimization is for all GPUs, not just top 1% for 60fps only.



That's not Nanite itself, that's also geometry instancing - the real calculation happens for only one statue. 3D Mark had a benchmark for that for a while now. Now, put 500 different meshes, so instancing doesn't work anymore, and you will see how FPS drops on its face. Then check how many triangles are really needed for good enough details of this statue, where it still makes real difference, cut down the mesh to that. And suddenly you can run a lot of different meshes that don't kill FPS so much anymore, with proper quality, that use less memory and space on the drive too. It's basic optimisation really, to not waste space and performance for nothing. You don't have to turn off Nanite, you just have to use it with head and not flip the switch on and forget about it.

That aside, highly optimised demo showing one scene is really nothing like actual games are. Might as well say 3D Mark tech demos are a good indicative of how games behave, which was never true. :)



What Devs need to do is to follow basic guidelines of UE and ML Devs - so far, as witnessed in many games and demos on ue5 engine, they often do not do even that. Your arguments make really little sense when even said ML Devs ask developers in their post to finally please follow performance optimisation guidelines and not just be lazy about it. That also, by the way, increases image quality and not just performance - bad practices introduce noise and attracts where it didn't have to be.

High refresh rate monitors doesn't mean we need to get stuck in older times, with graphics scaled down to accommodate de xxx fps - plus is not just down to graphics card alone. Those displays are for competitive MP gaming.

Stalker 2 has it's problems (you posted about them), 4060 gets only around 50fps in 1080p and yet... it sold well. That's because a good game will sell. Crysis also sold pretty good. Plus, using DLSS it will get you good enough performance + the saving grace over TAA :))

Of course you can use a best practices method (that goes for RT/PT as well, to not end up with reflections like in Hogwarts), but don't get your game limited by the low end of today. Start with some good general RT (again, Metro Exodus EE) and build upon that.
 
Last edited:
High refresh rate monitors doesn't mean we need to get stuck in older times, with graphics scaled down to accommodate de xxx fps - plus is not just down to graphics card alone. Those displays are for competitive MP gaming.

Competitive gaming (on a pro level where gear matters) is a tiny fraction of one percent of the market. No, these monitors aren't for competitive market, they exist because people like image to be clear in motion and for full clarity you need 1000Hz - we aren't there yet but we're getting there. RT/PT is currently the opposite - increasing realism but causing whole image to be blurry and unclear. These 2 things can't be combined currently but Nvidia and AMD seem to have plans to get there eventually. Eventually they will have to decouple FPS from actual image on the screen (e.g. by using reprojecton where irrelevant of the FPS generated by the GPU camera movement is still fluid and quick).

Stalker 2 has it's problems (you posted about them), 4060 gets only around 50fps in 1080p and yet... it sold well.

Stalker sold because of the online component - where people go to low details to have fun together and where graphics do not really matter much. How many modern AAA single player games restricting to high system requirements sold well? CP2077 is highly scalable and doesn't require PT nor RT, neither did Harry Potter. Anything actually sold well?

[/QUOTE]That's because a good game will sell.[/QUOTE]

Problem is almost none of the modern AAA games are actually good. They put all the money into the looks and there's nothing good in the actual gameplay. Almost all of them are financial flops too.

Crysis also sold good.

Really? You bring up ancient game from the time where getting fast enough computer costed less than middle range GPU itself cost how? It actually stands against your argument and not for it.

Plus, using DLSS it will get you good enough performance + the saving grace over TAA :))

What saving over TAA? TAA is everywhere now, it's part of the denoising and GI and plenty of other effects - DLSS changes nothing here, you can't turn off TAA. It's used as temporal accumulation in most if not all modern AAA games. One of the main reasons they look so blurry and have visual artifacts like ghosting etc.

Of course you can use a best practices method (that goes for RT/PT as well, to not end up with reflections like in Hogwarts), but don't get your game limited by the low end of today. Start with some good general RT (again, Metro Exodus EE) and build upon that.

Here we agree. However, it's not about limitations, it's about scalability - they can do fancy things that will work only on 6000 series GPUs in the future but they also need to give options to scale it down to current mainstream without killing visuals completely.
 
Haven't played Portal RTX, think I've got it on my account though so might download it after work and have a play.
I just tested it again - intially. I played it on my older 3080 12GB. It was laggy, slow, noisy and just not very impressive. Now on 4090 it works better, kiiinda ok. They added RR, improved denoising, etc. Without RR it's still a total mess with noise in reflections (plenty of glass they use in levels there). Without DLSS it's unplayable of course in 1440p. With just DLSS Quality it still isn't (I'd have to get down to Balanced to cross 60FPS in many places). FG works ok at least. :) With RR quality way better, but still noise and lag in GI and reflections is visible, just much smaller than without it - huge difference, yet not perfect. All in all, I would say in my res 4090 still feels like a bare min. for such games, to me.
 
Competitive gaming (on a pro level where gear matters) is a tiny fraction of one percent of the market. No, these monitors aren't for competitive market, they exist because people like image to be clear in motion and for full clarity you need 1000Hz - we aren't there yet but we're getting there. RT/PT is currently the opposite - increasing realism but causing whole image to be blurry and unclear. These 2 things can't be combined currently but Nvidia and AMD seem to have plans to get there eventually. Eventually they will have to decouple FPS from actual image on the screen (e.g. by using reprojecton where irrelevant of the FPS generated by the GPU camera movement is still fluid and quick).

Stalker sold because of the online component - where people go to low details to have fun together and where graphics do not really matter much. How many modern AAA single player games restricting to high system requirements sold well? CP2077 is highly scalable and doesn't require PT nor RT, neither did Harry Potter. Anything actually sold well?


That's because a good game will sell.

Problem is almost none of the modern AAA games are actually good. They put all the money into the looks and there's nothing good in the actual gameplay. Almost all of them are financial flops too.



Really? You bring up ancient game from the time where getting fast enough computer costed less than middle range GPU itself cost how? It actually stands against your argument and not for it.



What saving over TAA? TAA is everywhere now, it's part of the denoising and GI and plenty of other effects - DLSS changes nothing here, you can't turn off TAA. It's used as temporal accumulation in most if not all modern AAA games. One of the main reasons they look so blurry and have visual artifacts like ghosting etc.



Here we agree. However, it's not about limitations, it's about scalability - they can do fancy things that will work only on 6000 series GPUs in the future but they also need to give options to scale it down to current mainstream without killing visuals completely.

Again, if frame rate was king, people would not be playing 30fps on consoles, or with some downgraded looks for 60fps, where possible. It would HF2 graphics with at least 120fps. Or move entirely to PC gaming. But that doesn't happen. Besides, not only the GPU is the limit, but rather the rest of the system.

CB77 was running so poorly on PS4 that it was taken off the digital store :))

Black Myth: Wukong, developed by Game Science, has achieved remarkable commercial success since its release on August 20, 2024.

Sales Figures:

  • Launch Day: The game reached over 2.2 million concurrent players on Steam, becoming the highest single-player game by concurrent players of all time on the platform.

    Wikipedia

  • First Three Days: Game Science announced that Black Myth: Wukong sold 10 million units across all platforms within the first three days.

    Game World Observer

  • First Two Weeks: The game sold 18 million units within two weeks, making it one of the fastest starts in the global gaming industry.

    The Tech Game

  • First Month: By the end of its first month, sales surpassed 20 million units, solidifying its status as one of the fastest-selling games of all time.


  • Revenue Estimates:
    • As of early September 2024, Black Myth: Wukong generated over $850 million in gross revenue on Steam alone.

      Game World Observer

    • Estimates suggest that the game made approximately $2.05 billion in gross revenue across all platforms, with the developer's net revenue around $605 million.

      Steam Revenue Calculator
  • Development Budget:
    • The game was developed over six years with a budget of $70 million.

The above also was called a PT game. I'm not sure that it is, but at least is RT. Oh, and is on UE5.

Stalker has no online component, is just SP. It sold based on the series history, even though it always required some good PC to run it. Even today's hardware can't overcome the software issues of the past.

S.T.A.L.K.E.R. 2: Heart of Chornobyl, developed by GSC Game World, has achieved significant sales milestones since its release on November 20, 2024.

Sales Figures:

  • First 48 Hours: The game sold over one million copies across PC and Xbox platforms within the first two days of release.

    Gematsu

  • First Four Days: By November 24, 2024, sales on Steam alone surpassed 1.4 million copies, with gross revenue estimated at approximately $66 million.

    Game World Observer

  • Profitability: As of December 18, 2024, S.T.A.L.K.E.R. 2 has turned a profit, indicating robust sales beyond the initial launch period.

    Game Rant
Additional Insights:

  • Concurrent Players: At launch, the game attracted over 100,000 concurrent players on Steam, highlighting its immediate popularity.

    Game Rant

  • Xbox Game Pass Impact: Despite being available on Xbox Game Pass from day one, the game achieved impressive sales figures, suggesting that its inclusion in the subscription service did not significantly cannibalize direct purchases.

    Pure Xbox
These achievements are particularly noteworthy given the challenging development circumstances, including the ongoing conflict in Ukraine, where GSC Game World is based. The team's dedication has resonated with players worldwide, contributing to the game's commercial success.

The development of graphics isn't the issue, some fail because they're seen as a business, ran by business people too much.

Outside of the ghosting issues, DLSS (to me), solves the muddiness under some form or another.

Crysis is a good example, because even though hardware was relatively cheaper then, people would still buy as cheap as possible and often times I've heard the "is just a graphics demo, there's no real game there" mantra.

And good luck getting to 60fps without dropping the settings.

 
Last edited:
Back
Top Bottom