Saturday, September 23rd 2023
NVIDIA Foresees Greater Reliance on AI Rendering Techniques in Games
Digital Foundry organized an AI Visuals Roundtable earlier this week, with Alex Battaglia (Video Producer) acting as host—the main topic being a recently visually upgraded version of Cyberpunk 2077. Guests included: Bryan Catanzaro (NVIDIA's Vice President Applied Deep Learning Research), Jakub Knapik (VP Art and Global Art Director at CD Projekt RED), Jacob Freeman (GeForce Marketing Evangelist) and Pedro Valadas (PCMR Founder). Team Green, naturally, advocates its newly released DLSS 3.5 and Ray Reconstruction technologies—brute force rendering techniques are considered passé, and no longer an ideal solution for upcoming GPU generations. Valadas was curious enough to ask whether NVIDIA had any plans to target reasonable performance levels—minus DLSS—at native resolutions.
Catanzaro focused on the benefits introduced in Update 2.0 and Phantom Liberty: "I think that DLSS 3.5 actually makes Cyberpunk 2077 even more beautiful than native rendering...that's my belief. The reason being that the AI is able to make smarter decisions about how to render a scene...and I think that's going to continue to develop." He half jokingly states that rasterization is a "bag of fakeness." A combination of DLSS and path tracing is preferred over old hat methods—thus attaining the most realistic visual results. He summarizes with his own slogan: "Native rendering is fake frames." Catanzaro predicts that real-time graphics industries will become increasingly reliant on AI-processed image reconstruction and rendering technologies in the future. He hopes that AAA game environments will become cheaper to make in due time—with a shift to neural rendering. DLSS 10 could interface with game engine technology at a higher level, with a possible outcome being the creation of more "immersive and beautiful" experiences.
Digital Foundry's video description states: "Many thanks to all participants in this roundtable chat: Bryan Catanzaro, Vice President Applied Deep Learning Research at Nvidia, Jakub Knapik VP Art and Global Art Director at CD Projekt RED, GeForce evangelist Jacob Freeman and Pedro Valadas of the PCMR sub-Reddit."
Cyberpunk 2077 2.0 is here—the first showcase for DLSS 3.5 ray reconstruction, integrated into a full DLSS package including super resolution and frame generation, all combining to produce a state-of-the-art visual experience. In this roundtable discussion, we discuss how ray reconstruction works, how it was developed and its applications outside of path-traced games. We also talk about the evolution of the original DLSS, the success of DLSS in the modding community and the future of machine learning in PC graphics.
Sources:
Eurogamer, Tom's Hardware
Catanzaro focused on the benefits introduced in Update 2.0 and Phantom Liberty: "I think that DLSS 3.5 actually makes Cyberpunk 2077 even more beautiful than native rendering...that's my belief. The reason being that the AI is able to make smarter decisions about how to render a scene...and I think that's going to continue to develop." He half jokingly states that rasterization is a "bag of fakeness." A combination of DLSS and path tracing is preferred over old hat methods—thus attaining the most realistic visual results. He summarizes with his own slogan: "Native rendering is fake frames." Catanzaro predicts that real-time graphics industries will become increasingly reliant on AI-processed image reconstruction and rendering technologies in the future. He hopes that AAA game environments will become cheaper to make in due time—with a shift to neural rendering. DLSS 10 could interface with game engine technology at a higher level, with a possible outcome being the creation of more "immersive and beautiful" experiences.
Cyberpunk 2077 2.0 is here—the first showcase for DLSS 3.5 ray reconstruction, integrated into a full DLSS package including super resolution and frame generation, all combining to produce a state-of-the-art visual experience. In this roundtable discussion, we discuss how ray reconstruction works, how it was developed and its applications outside of path-traced games. We also talk about the evolution of the original DLSS, the success of DLSS in the modding community and the future of machine learning in PC graphics.
47 Comments on NVIDIA Foresees Greater Reliance on AI Rendering Techniques in Games
Nvidia pushed them forward and made them benchmarks.
Benchmarks that they themselves sit atop of and control.
Talk about pushing a certain POV(Huang's)
Personally I think Gamers nexus showed it's real worth.
And as for AI based rendering, eww the smears or effects it adds or deleted isn't for me.
Surely nothing bad could come of that /s
(AMD and Intel do too, its why they invest so much in AI hardware and tools)
Why isn't DLSS used to improve upon native for example? Or why a 1800€+ can't do RT? Stuff like that is more suitable for mid to low end GPUs.They are basically part of the marketing division of nVidia.
And people gulp this crap with a smile on their faces.
RT is incredibly taxing. Rendering a game with full RT at 1440p@60FPS so it can be upscalled to 4K is beyond anything a graphics card can pull in the next two decades. The missing performance is raster, as simple as that.
For the same reason we don’t complain about cube maps or screen space reflections and their ridiculous bugs and limitations, we shouldn’t complain for any tech that provides better image.
And no, there’s no tech that delivers 100% with no bugs or glitches.
thus sayeth Jacob Freeman, GeForce Evangelist
:roll:
Yes, fudge's.
Because that's what they are, path tracing requires fudges to make it viable, but those fudges don't equal better than native IMHO , different yes, closer to real yes.
But streaky effects, effects going missing and chevron patches turning up now and again, and that's If you bought a 4090 lower end cards , this is a joke on.
And Nvidia move the goal posts past your hardware quicker than the console cycle.
Most sites are towing the line and sticking to Nvidia's notes, GN are not and give a fair, not hyped assessment, and test fairer, though not like they were told to.
They also show, in motion why it's a pass from me, and that's on Nvidia's hardware Showcase sponsor title.
No wonder PC gaming is dying.A fair assessment that concludes ray reconstruction is a good feature... and considering this is the first release, it will only get better.
News at 11.That still exists. But NVidia's job here is to sell AI because that's NVidia's key advantage over AMD. If something is "simply" a graphics effect, AMD offers similar, or possibly even greater performance (thanks to Infinity Cache). NVidia can't win on that front.
I generally liked the video, I like Alex's coverages of topics, but many segments in this video, it really just felt like an hour long nvidia ad, and how they are descending upon us mere mortals, granting us technologies of the future, shit man, they even had an evangelist, a fucking evangelist, why are they here? What do they bring to the discussion besides brainwashing?
They even started saying normal frames are fake lol.
DF tends to be pretty level headed but they have done Nvidia sponsored stuff in the past and they all run Nvidia stuff on their main rigs which is just a bit....yeah unbalanced.
And the extremely lame softball questions that Alex asked here are just cringeworthy imo.
and in the end I want to state again, I want to be in a time period where all this nonsense is just behind us, no more DLSS/FSR/XESS and just one hardware agnostic solution that just works for all so we can once again properly focus on hardware.
And RT at levels that we dont even need a denoiser anymore.
Its just unprecedented this focus on software that determine how a game looks per hardware vendor and I cant wait for it to be behind us.
On top of that, each AI technology has drawbacks which often include reductions in visual quality in certain areas. In the past technologies like FXAA have been called out for a reduction in image quality despite being performant so it stands to reason that something like RT + AI would be called out for it's performance and visual quality negatives just to enable Ray Tracing who's benefit at this moment is questionable for many individuals.
Ray tracing will certainly get better and is the path forward for games that want a hyper-realistic style but let's not pretend that as of right now it's just only positives.
If they drop all the shader units & anything that's built for rasterization. Then just add all the raytracing parts alone by selfs (tensors cores, RT cores,opitical flow accelators, BHV trassvesal parts. The only thing you'll likely need more of is Render out puts (R.O.Ps) to do full raytracing without rasterization.
Much as we love this forum, we're a very small segment... if you went by TPU's numbers, Radeon would have like a 70% market share. Don't let it get to you, mate.