Soldato
Huge difference.
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.
I think it's just bugs like they couldn't run Ray tracing in ratchet and clank but assuming there is no bugs they look the same on both
So we had fake pixels, then fake frames and now fake light rays - none of these came from game engine but were instead added by AI algorithms. You can say it's better but it's as true as AI generated images - they are what AI thinks they should be and not what would really be there if it was actually calculated properly. But then again, current GPUs are still way way too slow for true calculations and such shortcuts are necessary for real time graphics in games.So now you can literally say dlss 3.5 looks better than native because it's adding additional ray tracing detail into the frames that aren't there with native - for example see the reflection on the floor and walls with dlss 3.5 vs dlss 3.0/native
So we had fake pixels, then fake frames and now fake light rays - none of these came from game engine but were instead added by AI algorithms. You can say it's better but it's as true as AI generated images - they are what AI thinks they should be and not what would really be there if it was actually calculated properly. But then again, current GPUs are still way way too slow for true calculations and such shortcuts are necessary for real time graphics in games.
What I got from the video presentation is that it's fixing what the normal denoiser does by using AI to make a more informed decision vs the more "hard-coded" approach. I mean everything in a game is generated, so why not have it be intelligent and dynamic vs static.So we had fake pixels, then fake frames and now take light rays - none of these came from game engine but were instead added by AI algorithms. You can say it's better but it's as true as AI generated images - they are what AI thinks they should be and not what would really be there if it was actually calculated properly. But then again, current GPUs are still way way too slow for true calculations and such shortcuts are necessary for real time graphics in games.
Looks like a speed of light issue..... does light move at walking speed in Cyberpunk.
The so called reflected image is rotated significantly from the source tower. I also dislike just how just how glassy the reflections are.
AI isn't a simulation. It's closer to hallucinations. That said, it matters but only in photorealistic image, which current games even with best possible RT implementations aren't even close to (for various reasons, mostly textures and other assets).Does it matter if it looks better? Every part of every game ever made is fake - it's a simulation.
What do you mean by static? Game engine with RT renders things dynamically not statically. AI has nothing to do with dynamism. And yes, it masks (and not fixes) the problem with current GPUs just being still magnitudes too slow for what Nvidia is pushing gaming market to implement. Which is why you need denoising in the first place, as noise is the effect of very small number of rays actually being calculated. All else is approximation of approximation really, including AI layer on top of that.What I got from the video presentation is that it's fixing what the normal denoiser does by using AI to make a more informed decision vs the more "hard-coded" approach. I mean everything in a game is generated, so why not have it be intelligent and dynamic vs static.
And horrible textures in many places, so close to 0 details when moving up close. To me textures matter the most, on proper assets and then light added as fluff. Textures cost no FPS (but certain vendor love to skimp on vRAM), reflections cost a lot. We got it backwards, but then again, shiny sells better. Same as loud music sells better than properly mastered.Extra glassy RT, the new bloom, chuck it everywhere. In cyberpunk all the showcase vids look the same.
pavement - glass
wall - glass,
car - glass
glass - x2 glass
concrete - frosted glass
RT is dynamic, but the denoising routines are relatively static and only have the current data plus whatever instructions have been given to it. The AI model is trained on a ton of varied data and uses that to make a decision on a case-by-case basis, which makes it more dynamic.What do you mean by static? Game engine with RT renders things dynamically not statically. AI has nothing to do with dynamism. And yes, it masks (and not fixes) the problem with current GPUs just being still magnitudes too slow for what Nvidia is pushing gaming market to implement. Which is why you need denoising in the first place, as noise is the effect of very small number of rays actually being calculated. All else is approximation of approximation really, including AI layer on top of that.
That said, I didn't say it's bad, it's instead necessary. But one should realise it's all just approximations and hallucinations and not what actually would be there if it all was properly calculated. We will never get actually photorealistic images this way, before GPUs become magnitudes faster than they are now. None of which influences how fun games are (or aren't) to play, though.