Digital Foundry Article Technical Discussion [2020]

Status
Not open for further replies.
That ties in with our discussions about UE5 being everywhere, and the issues with competing with in-house Engines. CDPR are using their in house engine, REDengine 4, and I just don't think they can create an engine with all the bells and whistles of the major players simply because of the size of those teams, funded by the entire industry backing a few players.

It's the software equivalent of a new console platform. No-one can realistic create a new console now unless they are huge. No-one can really create a competitive engine unless they are huge. CDPR isn't small, but how much can 1000 people creating a huge open world game achieve versus 2,000 or 3,000 working on an engine and tools year after year?
 
Last edited:
Quite impressive, considering I was one of the people going "RT in consoles in 2020?? NEVER!"
So, wig = snatched -> confirmed. Nice. :cool:

--------------------

I'm missing the talk about RT in lighting/shadows, though. Current reflection "hacks" can be so convincing that we got to the point of pausing videos and speculating if RT reflections are used or not. So, yes, I like RT reflections, I think it's cool and in some cases they are clearly noticeable, no need to pause and squint, but I hope this doesn't come at the price of a less realistic/interactive lighting model, for instance.
 
I'm missing the talk about RT in lighting/shadows, though. Current reflection "hacks" can be so convincing that we got to the point of pausing videos and speculating if RT reflections are used or not.
Ish. The moment reflections break, it's really off-putting, a real Matrix moment.

I totally agree that the emphasis should be directed towards lighting, but I do also think we want to reduce/eliminate high-impact artefacts. Shadow crawl is one such high impact, whereas missing shadows isn't so jarring and you can adapt to viewing a world with missing shadows without being beaten over the head with the problems. Shimmer is another problem. Much as we hate blurry textures, they are far less distracting than PS2-level texture aliasing.

I guess it's really temporal artefacts (that sounds very Doctor Who!) that are the problem. We will acclimate to consistent graphical problems, but those which generate high deltas frame to frame act upon the sensitive motion detection parts of the brain. Hmm, well I guess it's not that easy, as there are also artistic faults, like my blue cat/orange rock problems with games. I guess any delta that's too large, an object not lit consistently with others, or a load of pixels changing more than they should, drags our attention away from the game immersion.
 
Ish. The moment reflections break, it's really off-putting, a real Matrix moment.

I totally agree that the emphasis should be directed towards lighting, but I do also think we want to reduce/eliminate high-impact artefacts. Shadow crawl is one such high impact, whereas missing shadows isn't so jarring and you can adapt to viewing a world with missing shadows without being beaten over the head with the problems. Shimmer is another problem. Much as we hate blurry textures, they are far less distracting than PS2-level texture aliasing.

I guess it's really temporal artefacts (that sounds very Doctor Who!) that are the problem. We will acclimate to consistent graphical problems, but those which generate high deltas frame to frame act upon the sensitive motion detection parts of the brain. Hmm, well I guess it's not that easy, as there are also artistic faults, like my blue cat/orange rock problems with games. I guess any delta that's too large, an object not lit consistently with others, or a load of pixels changing more than they should, drags our attention away from the game immersion.
Yeah, that's the point. We want it ALL but we still have to deal with sacrifices. :rolleyes:
 
On the insanity of the tech that is in nextgen consoles, has anybody rewatched any of the Cyberpunk 2077 footage and had a real "meh" moment? Because I remember thinking those graphics would even be a stretch even for nextgen consoles but from what we've seen recently, it is not going to a great-looking game - not unless CDPR have a ton of tech that they've been holding back on showing. I don't play games for pretty graphics but I am a little anxious about Cyberpunk.
Graphically it's good. But it's not like how we would imagine the next gen experience. Gameplay wise however, I feel it's pushing boundaries further with respect to these FPS RPGs. Reminds me of a much better system shock or deus ex
 
That ties in with our discussions about UE5 being everywhere, and the issues with competing with in-house Engines. CDPR are using their in house engine, REDengine 4, and I just don't think they can create an engine with all the bells and whistles of the major players simply because of the size of those teams, funded by the entire industry backing a few players.

I don't think this is a fair comparison. Cyber Punk is targeting current gen consoles so if it looks worse than an Unreal game on current gen then fair enough.
 
As mentioned above, Minecraft RTX was a point of comparison of 30-60fps at 1080p compared to 2080Ti, which had a locked 60fps. Ergo, we might infer that SX simply has up to half the raw RT performance for really heavy scenes, which would then put it closer to a 2060-level of RT power.

Obviously, developers are still grappling with implementations for future (hybrid) use, but those are performance optimizations that should carry through to every platform.
I don't know if early code on potentially unfinished hardware should be used to draw conclusions as to the analogous PC equivalence. What then happens if come release, they get the performance to locked 1080p/60? Does it suddenly become as good at RT as 2080Ti? Bear in mind this is on a different hardware RT implementation without much known into its characteristics/driver status. These are early codes ported by a member of the team in however many weeks as against a code that has been worked on constantly by the Minecraft RT team in collaboration with Nvidia's engineers/software devs. I think it would be wise to wait until we have the final product before we begin to draw any conclusion as to the exact capability of the system.
I am sure we will see a PC AMD gpu analog to the XSX. Hell they might end up locking it at 1080p/30 for all we know.
 
I don't know if early code on potentially unfinished hardware should be used to draw conclusions as to the analogous PC equivalence. What then happens if come release, they get the performance to locked 1080p/60? Does it suddenly become as good at RT as 2080Ti? Bear in mind this is on a different hardware RT implementation without much known into its characteristics/driver status. These are early codes ported by a member of the team in however many weeks as against a code that has been worked on constantly by the Minecraft RT team in collaboration with Nvidia's engineers/software devs. I think it would be wise to wait until we have the final product before we begin to draw any conclusion as to the exact capability of the system.
I am sure we will see a PC AMD gpu analog to the XSX. Hell they might end up locking it at 1080p/30 for all we know.
Sure, it's WIP. The other problem with the comparison is that we don't know just how much faster it would run on 2080Ti if it weren't capped at 60fps while I assume any sort of performance optimizations would carry over between platforms for the most part because it's DXR though we won't know unless they say.
 
Sure, it's WIP. The other problem with the comparison is that we don't know just how much faster it would run on 2080Ti if it weren't capped at 60fps while I assume any sort of performance optimizations would carry over between platforms for the most part because it's DXR though we won't know unless they say.
True but there is a DF video where Alex interviews several Nvidia devs and they go into detail about the work they have been doing, are doing, and will still do in the future to improve performance and features. While you would expect these to be translated into the XSX version because of DXR, my other point is that those optimizations might run particularly well because its running on RTX hardware. AMD's stab at RTRT might require a different set of optimizations to be more performance. It certainly will have its own set of peculiarities different from RTX RTRT implementation.
 
Richard mentioned you need a 2070-2080 to future proof but I think he's lowballing it by heaps, at least if you wanna turn on Raytracing. We saw quite a few PS5 games running RT in native 4k/30fps already and some with graphics vastly superior to the likes of BF5, Metro, Control or Shadow of the Tomb Raider. Yet those cards struggled heaps in those games at native 4k with RT On, only the 2080 TI can run them in a much playable fps. I think he underestimated the console optimization, efficiency a lot here, you'd definitely need a 2080 ti to future proof or just wait for the Big Navi and 3000 RTX series.

Console games are certainly better optimized than on PC but dont underestimate the performance implications of tacking on RT to existing engines.
 
Richard mentioned you need a 2070-2080 to future proof but I think he's lowballing it by heaps, at least if you wanna turn on Raytracing. We saw quite a few PS5 games running RT in native 4k/30fps already and some with graphics vastly superior to the likes of BF5, Metro, Control or Shadow of the Tomb Raider. Yet those cards struggled heaps in those games at native 4k with RT On, only the 2080 TI can run them in a much playable fps.
As shown in DF's RT video, the PS5 games shown achieve great concessions in making their RT work, they use mostly RT reflections, but with less than 1/4 resolution and use it selectively. This is in a very stark contrast to RT in the above mentioned games, where it was applied in a vastly more general and sweeping manner.
 
EawbH0dU8AA0CY9


EawbLfyUcAAHADR


Eaww7KAXYAE4gDF


Found these on Twitter, hinting that at least some dynamic elements are seemingly missing from the reflections. What do you think?
 
EawbH0dU8AA0CY9


EawbLfyUcAAHADR


Eaww7KAXYAE4gDF


Found these on Twitter, hinting that at least some dynamic elements are seemingly missing from the reflections. What do you think?
Yea lots of little compromises. Later in the trailer where you see in car gameplay; i was unable to see my own cars reflection in the trunk of the red Mazda. It’s clear it’s reflecting properly, but my car isn’t there.

doesn't mean there isn’t RT. But they are doing their best to incorporate elements of it without blasting their budget.
 
You have to be careful where you look and what you look at though. They have experimented with all sorts of technology including some basic ray tracing with the SPUs on PS3 I think.

For Unreal Engine vs The Witcher, don’t underestimate the potential for inefficiency when you have to support everything and the kitchen sink versus a much more focused target. And smart developers can create an engine with tool support that can really make the pipeline of a specific company working on a specific type of game sing, witness Decima being so well setup that it could both create great graphics and be mature enough so that Kojima could make a game with it, say. He could have just picked Unreal Engine.

Personally, I have been facing the same kind of decision over and over in my career. Use something already there or write something new? And writing something new is still surprisingly often a good choice.
 
Unreal Engine in its stock form produces very unimpressive visuals and performance. Its only when Sony or Microsoft 1st party completely retool the rendering that it competes. Bespoke solutions such as Decima, ND engine, frostbite, RAGE etc produce far superior visuals and performance.
 
Status
Not open for further replies.
Back
Top