Current Generation Games Analysis Technical Discussion [2020-2021] [XBSX|S, PS5, PC]

VRS tier 2 can leverage an image mask of sorts to determine shading rates. If you don’t know what the mask looks like it would be difficult. I don’t think it’s typically grid like.

I lean to the issues being encoder related mainly because
A) developers wouldn’t ship a title looking like that, the reviews would have been plain horrible. The fact that it takes side by side comparisons for users to notice the difference is probably a measure of how far it away it looks from the YouTube. As you don’t need a side by side comparison to see how bad YouTube is.

b) I suspect with XSX versions of the game the developers still choose the best settings to present their title. We have to believe that they aren’t artificially just adding things to make it worse just for check boxes. It would be natural if them to reduce settings to gain performance and chuck VRS if it was producing the type of encoder issues you see there.

c) I don’t even know if VRS is present on nvidia GPUs. Not sure if someone can check on PC.

 

YouTube compressions has not been kind to Dirt 5 but folks like DF and NXG are capturing and analysing raw footage. When you're zooming in on raw footage and uploading the zoomed comparisons, like the image above, I'm dubious if YouTube's crappy compression could introduce artefacts that appears like geometry or depth are removed.

It just feels like Dirt 5 on Xbox has more bugs than PS5 right now because some of these differences are just weird and make no sense. None of them impact the game though, and the game is good.

 

The most reliable would be David himself saying they do use vrs for Xbox series s and x.

 

Thanks for confirming and making the gif

 

It's a DX12U feature, therefore it is supported on Nvidia hardware.
Not saying it's supported in dirt 5 on Nvidia, just that the hardware supports it.
Also for ms vrs dev videos i thought they was using Nvidia cards.

 

Thanks as this is something I've been very curious about on XS.
Did he talk about PS5 features used at all, i.e. VRS?

 

I mean for dirt 5. I’ve only seen it marketed on 6000 series cards. I’ll take another gander to see if there are benchmarks with VRS
On and off for both vendors

 

Mostly just says can't talk about it in regards to PS5.

 

That would be interesting if so. Would certainly help to explain the higher than usual AMD performance.

 

If you actually watch the vid PS5 is dropping a few frames from 60, at least in the video XSX doesn't drop any. It is the same area. Not sure why he didn't mention. But no idea if the problem is widespread or present on both since it's very short clip. Here again VGtech's data approach could help.

 

I love that numbers are published like this, but I have a curious observation. How do you get a frame time of 22.11ms on PS5? It doesn't support VRR, so it can't be an arbitrary number like it could on Xbox. At 60hz you would limited to multiple of 16.67ms, at 120hz multiples of 8.34ms. If you had a frame time of 25ms that would be about right for missing a refresh for a 60hz update on a 120hz display (3 refreshes instead of 2). But a number in between? Same thing with the minimum frame time of 12.69ms. How is this possible? It's like 1 and a half refreshes at 120hz.

 

Actually the XSX is dropping one or 2 frames in 3 scenes (there are 4 framerate scenes in total) when there is plenty of alphas. PS5 has framerate drops in one scene apparently during some data streaming.

 

I've not seen the video but one one to express a real frame time where there is tearing is to calculate the frame time at the point of the tear.

 

Hmm... So would you count the lines until you get to the tear and calculate the frame time by actually multiplying the time it takes to render each line by the number of lines? Does VG do all of it's capture at 120hz? And if so, at 4k120? And wouldn't this method create anomalies in data for the other systems? Series S performance mode has more than 10x the torn frames PS5 has, but it's max frame time is 33ms, or 2 frames at 60hz. The minimum is still odd, though. And I guess VRR can't be enabled because there are torn frames.

 

Yes.

 

Not sure why anyone is comparing performance on such a raw game, but whatever, here it is... Images from video were copied from this post on XboxEra [ https://forum.xboxera.com/t/cyberpu...for-this-game-to-release-no-spoilers/5204/279 ]



Pinned Video Comment:

The game runs in backward compatibility mode on PS5 / XSX. The NextGen patch will arrive in 2021.

The PC version runs at 4K / Ultra 30FPS with DLSS.

There are some interesting differences between versions, starting with the absence of modes on PS5. Series X has a 1800p / 30fps mode and another with a medium resolution of 1224p / 60fps, while the PS5 has to settle for only the second option. Some textures and shadows are considerably lower on the PS5 version in addition to having a lower draw distance and npcs density.

The most noticeable differences in PC are mainly the textures, the lighting (with RT) and the density of NPCs. However, I am not very happy with the DLSS in this game, as it sometimes blurs the image too much.

Currently, on consoles, the best version to play is Series X without a doubt.
We will have to wait for the 2021 patch to see if there are changes to this result.
​

 

Comparing backwards compatible games is beyond pointless. I feel for all of the people playing cyberpunk on consoles without a nextgen patch, but its not technically interesting at all.

 

I'm going to find it interesting watching the progress. Although I'll probably stick to DF comparisons mostly.
So I'm glad for comparisons