Digital Foundry Article Technical Discussion [2022]

Status
Not open for further replies.

That was extremly well done. Settings were shown, including what framerate and visual impact they have. Love it.

Also as always, very interesting to see the 2060 performance with RT. It's pretty cool that it can reach a near stable 60 FPS (if that grass field is really a worst case ofc) using the transformative RTGI+RTAO effects thanks to DLSS.

At a 30 FPS target, the 2060 with DLSS Balanced in 1440p plus those transformative RT effects on, the experience is far superior compared to a PS5 even though its raw performance is a lot weaker. Maybe it could even do all effects on with DLSS Performance 1440p.
 
At a 30 FPS target, the 2060 with DLSS Balanced in 1440p plus those transformative RT effects on, the experience is far superior compared to a PS5 even though its raw performance is a lot weaker. Maybe it could even do all effects on with DLSS Performance 1440p.

Its just that you'd need DLSS with a GPU in the 2060-class. Some like it some dont, i think its a great option for those that want RT and are on a 2060, in special seeing the mining craze for the moment. The 2060/S isnt that much slower in raw rendering compared to 5700XT/PS5, yes the latter has the upper-hand obviously, but not enough to make up for the ray tracing/DLSS harness the RTX enjoys.

As per DF, incredible looking 'next gen' game.
 
Curious though. Does this game use a single bounce RTGI ? Meaning, the rays bounce once and thats it ? Cause if so, its not really that accurate. It looks more like the original release of Metro Exodus. That train carriage comparisson. There are some huge windows in there, the interior should be fully lit, not almost dark.






This is not right at all. Gigantic windows one after the other and daylight outside. Its not gonna be that dark inside. The second pic even worse. Light doesnt just utterly stops one metre from the door and everything around is pitch black. It was the same in Metro Exodus vs the Enhanced edition. What we claimed is accurate and good in the original single bounce rtgi release was not in the slightest and everything looked different in the enhanced release. All the areas were much brighter and ample lit than in the original game
 
That was extremly well done. Settings were shown, including what framerate and visual impact they have. Love it.

Also as always, very interesting to see the 2060 performance with RT. It's pretty cool that it can reach a near stable 60 FPS (if that grass field is really a worst case ofc) using the transformative RTGI+RTAO effects thanks to DLSS.

At a 30 FPS target, the 2060 with DLSS Balanced in 1440p plus those transformative RT effects on, the experience is far superior compared to a PS5 even though its raw performance is a lot weaker. Maybe it could even do all effects on with DLSS Performance 1440p.

I think its fair to say by now that despite initial reservations from much of the gaming community about the RTX featureset when Turing first launched, it turned out to be a big winner.

Certainly wish I'd got one back they were affordable even if the rasterization performance uplift over my 1070 was modest.
 
Curious though. Does this game use a single bounce RTGI ? Meaning, the rays bounce once and thats it ? Cause if so, its not really that accurate. It looks more like the original release of Metro Exodus. That train carriage comparisson. There are some huge windows in there, the interior should be fully lit, not almost dark.






This is not right at all. Gigantic windows one after the other and daylight outside. Its not gonna be that dark inside. The second pic even worse. Light doesnt just utterly stops one metre from the door and everything around is pitch black. It was the same in Metro Exodus vs the Enhanced edition. What we claimed is accurate and good in the original single bounce rtgi release was not in the slightest and everything looked different in the enhanced release. All the areas were much brighter and ample lit than in the original game
yes something is off with rtgi like for example in this scene
Im pretty sure lighting physics doesn't work this way in our dimension ;d
 
Last edited:
Great video by Alex. He goes into great depth to show the differences but as others have mentioned, something is definitely off with RTGI. Regardless, ray tracing doesn't save this game's visuals for me. It's thoroughly unimpressive imo and looks last gen even with ray tracing. Geometry quality, foliage rendering, character models and animations are some the many things that rub me the wrong way. I for one can't wait for the curse of cross-gen to end.
 
  • Like
Reactions: snc
lighting physics doesn't work this way in our dimension

I guess that not every game is going to use RT/lighting to chase realism to perfection. It can be used to enhance fidelity aswell, as seen in Dying Light 2, it looks substantionally better, be it realistic or not. I hope new tech will be used to create what we want to see, not just to chase realism other then for simulation-oriented titles and/or when intended like mirrors and reflections.
Truly next-generation looking game with the enhanced lighting for sure!
 
Curious though. Does this game use a single bounce RTGI ? Meaning, the rays bounce once and thats it ? Cause if so, its not really that accurate. It looks more like the original release of Metro Exodus. That train carriage comparisson. There are some huge windows in there, the interior should be fully lit, not almost dark.






This is not right at all. Gigantic windows one after the other and daylight outside. Its not gonna be that dark inside. The second pic even worse. Light doesnt just utterly stops one metre from the door and everything around is pitch black. It was the same in Metro Exodus vs the Enhanced edition. What we claimed is accurate and good in the original single bounce rtgi release was not in the slightest and everything looked different in the enhanced release. All the areas were much brighter and ample lit than in the original game

It’s hard as hell to determine if a scene is lit correctly outside of very basic things (like knowing you should cast a shadow if standing directly in front of a strong light). There are too many variables involved for your brain to bother to try to discriminate between incorrect versus correct.

upload_2022-2-3_16-1-9.jpeg

upload_2022-2-3_16-2-19.jpeg

If the room in the second picture was as lit as the first would your brain automatically know it was incorrect? There is a reason why dev routinely used offline ray tracing to help light scenes even before real-time RT was a thing. It’s hard to eyeball it.
 
It’s hard as hell to determine if a scene is lit correctly outside of very basic things (like knowing you should cast a shadow if standing directly in front of a strong light). There are too many variables involved for your brain to bother to try to discriminate between incorrect versus correct.

View attachment 6247

View attachment 6248

If the room in the second picture was as lit as the first would your brain automatically know it was incorrect? There is a reason why dev routinely used offline ray tracing to help light scenes even before real-time RT was a thing. It’s hard to eyeball it.
what do you think about this scene
ofDLsG8.jpeg

imo its impossible to be so dark in sunny day
 
It’s hard as hell to determine if a scene is lit correctly outside of very basic things (like knowing you should cast a shadow if standing directly in front of a strong light). There are too many variables involved for your brain to bother to try to discriminate between incorrect versus correct.

View attachment 6247

View attachment 6248

If the room in the second picture was as lit as the first would your brain automatically know it was incorrect? There is a reason why dev routinely used offline ray tracing to help light scenes even before real-time RT was a thing. It’s hard to eyeball it.


That second pic doesnt look like any room i've ever seen in my entire life :)) Its either a processed picture for the effect, or its in a place where the building is very tall and the sun is obscurred by buildings. So we only see the bright light on the street



In Metro we were gushing on the original release and how "accurate" it was, but as you can see, it was 100% wrong in every possible case. In the last pic, the raster image was better and more correct in fact that the primitive RT they had in the original release. Its the same in Dying Light 2. Its nowhere near accurate or correct. Its just a different type of image compared to it off. Single bounce RT is not cutting the mustard. Metro has the exact types of scenes we have in Dying Light 2, where there are giant sources of light, but everything is obscured
 
That second pic doesnt look like any room i've ever seen in my entire life :)) Its either a processed picture for the effect, or its in a place where the building is very tall and the sun is obscurred by buildings. So we only see the bright light on the street



In Metro we were gushing on the original release and how "accurate" it was, but as you can see, it was 100% wrong in every possible case. In the last pic, the raster image was better and more correct in fact that the primitive RT they had in the original release. Its the same in Dying Light 2. Its nowhere near accurate or correct. Its just a different type of image compared to it off. Single bounce RT is not cutting the mustard. Metro has the exact types of scenes we have in Dying Light 2, where there are giant sources of light, but everything is obscured
You are misrepresenting the lighting in Dying Light 2. The blackness you are seeing in a shadowed region in my video is 100% post-processing related and not because the game leaves shadows pure black in due to bounces.
The lighting is physically plausible - but the game is actually black crushing it in post. If it were purely single bounce GI, no amount of Gamma adjustment could reclaim detail in shadow areas (take Doom 3 for example).
gammat8k0w.png
 
You are misrepresenting the lighting in Dying Light 2. The blackness you are seeing in a shadowed region in my video is 100% post-processing related and not because the game leaves shadows pure black in due to bounces.
The lighting is physically plausible - but the game is actually black crushing it in post. If it were purely single bounce GI, no amount of Gamma adjustment could reclaim detail in shadow areas (take Doom 3 for example).
gammat8k0w.png
As I thought its not gamma problem as even after adjustment its unrealistic to have so dark places in this situation (but thx for checking and photo)
 
As I thought its not gamma problem as even after adjustment its unrealistic to have so dark places in this situation (but thx for checking and photo)
You are not speaking sense - adjusting gamma is reclaiming the dark areas. That points to the fact that they ARE lit. If they were not lit, the black would just turn to grey.

Without multibounce GI the game would not have scenes like this one: if it were single bounce the entire area where the arrow is here would be completely black.
backside25kvq.png
 
You are not speaking sense - adjusting gamma is reclaiming the dark areas. That points to the fact that they ARE lit. If they were not lit, the black would just turn to grey.
True is defenitly better now but still off, imo something more here than just gammma settings edit: maybe as Scott_Arm suggested its hdr proboem with rt, option to see this scene in sdr ?
 
Last edited:
Maybe Techland will do a gdc about their implementation here or some article down the line. I started playing the game an hour ago, i reached a few of the areas you have in the video. I just cant escape the feeling that it just doesnt look as it should. Similar scenes that we can find in Metro look completely different and even what your real life experience would tell you something should look, is too dark in here. It does look good with RT, as a game, but im sure they had to make a lot of concessions with it. I think this is the first game thats heavier a bit to run than Cyberpunk at Psycho RT
 
Status
Not open for further replies.
Back
Top