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

Status
Not open for further replies.
The versions tested were 1.0.162.766 on Xbox Series X|S and 1.01 on PS5. Only Mass Effect 1 was tested in this video. Note that all three consoles are running the game via backwards compatibility.

Timestamps:
00:00 - PS5 in Quality Mode and Xbox Series Consoles at 60fps
07:14 - Xbox Series X 120fps
14:28 - Xbox Series S 30fps
PS5 and both Xbox Series consoles in Favor Quality Mode all use a dynamic resolution with the highest resolution found being 3840x2160, the lowest resolution found during gameplay being approximately 3456x1944 and the lowest resolution found during cutscenes being approximately 3107x1748. PS5 and Xbox Series X seem to rarely drop below 3840x2160 in Quality Mode.
Both Xbox Series consoles in Favor Framerate mode use a dynamic resolution with the highest resolution found being 2560x1440, the lowest resolution found during gameplay being approximately 1920x1080 and the lowest resolution found during cutscenes being approximately 1728x972. Pixel counts at 2560x1440 seem to be common on Xbox Series X in Favor Framerate mode.
PS5 also has a Favor Framerate mode that renders at 1440p and 60fps.

The Favor Framerate mode on PS5 has improved performance compared to Favor Quality mode https://bit.ly/3uNKiv4

Favor Framerate mode is labeled as Perf Mode in the video.

Notes:
* In 60fps Quality Mode, XBSX highest framerate is 60fps, mean of 59.94fps, and the lowest being 43fps.
* In 60fps Quality Mode, PS5 highest framerate is 60fps, mean of 59.81fps, and the lowest being 30fps.
* In 30fps Quality Mode, XBSS highest framerate is 30fps, mean of 29.94fps, and the lowest being 19fps.
* In 60fps Performance Mode, XBSS highest framerate is 60fps, mean of 59.94fps, and the lowest being 51fps.
* In 120fps Framerate Mode, XBSX highest framerate is 120fps, mean of 115.44fps, and the lowest being 78fps.
* PS5 and XBSS have no 120fps Framerate Mode.


The versions tested were 2.0.0.20 on Xbox Series consoles and 1.002.000 on PS5. There is some time of day variance in some clips.

Timestamps:
00:00 - PS5 and Xbox Series X
06:22 - Xbox Series S
PS5 and Xbox Series X both use a dynamic resolution with the highest resolution found being 3840x2160 and the lowest resolution found being approximately 3360x1890. Pixel counts at 3840x2160 seem to be common on both PS5 and Xbox Series X.
Xbox Series S in performance mode uses a dynamic resolution with the highest resolution found being 2560x1440 and the lowest resolution found being 1920x1080. Pixel counts seem to often be somewhere between the 2560x1440 and 1920x1080 on Xbox Series S in performance mode.

The only resolution found on Xbox Series S in resolution mode was 2560x1440.

Xbox Series S renders the UI at 3840x2160 in both modes.

Notes:
* PS5 has a locked 60fps presentation.
* XBSX lowest frame drop was 56fps.
* XBSS has a locked 30fps presentation in Resolution Mode.
* XBSS lowest frame drop was 52fps in Performance Mode.
 

...

Notes:
* PS5 has a locked 60fps presentation.
* XBSX lowest frame drop was 56fps.
* XBSS has a locked 30fps presentation in Resolution Mode.
* XBSS lowest frame drop was 52fps in Performance Mode.

Note that in the video you can see that the PS5 version has some framepacing issues in the 30fps cutscenes.
 
Note that in the video you can see that the PS5 version has some framepacing issues in the 30fps cutscenes.
Yep, odd 16.66ms frames (no 50ms frames afterwise) occurring almost every seconds. It's very probably caused by with the way the framerate is capped on PS5.

Unsuprisingly the biggest drops on XSX (Saints Row 3) occur during the most demanding alpha effects (explosion + smoke) starting at 5:15
 
To answer that question we need to look at the CPU side of things: sound, physics, character and environment animations, NPC AI, interactibility etc.. If those aspects look PS4-ish, then chances are it will run on a PS4, albeit at a much worse image quality.

Its clearly a game designed to straddle the two generations. The game looks nice but doesn't blow me away. The ps4 will most likely render it at 1080p 30 while having everything cut way back. The ps5 will hopefully render at 4k 30 and have much nicer textures and everything else will be limited by jaguar.
 
The physics look extremly impressive when shooting down the little fortress on top of the mechanical elephant.

I don't believe this would be possible on the old Jaguar CPUs.
 
The physics look extremly impressive when shooting down the little fortress on top of the mechanical elephant.

I don't believe this would be possible on the old Jaguar CPUs.
It's probably not done by a physics simulation on the CPU. It could be GPU accelerated, or, more likely I think, an animation. Or, it could be simply changed in the PS4 version to accommodate the system.

Has anyone here played Rise of the Tomb Raider on Xbox 360? That's a cross gen port that I think might shine some light on the compromises you'd have to make to a game to get a like this on 2 platforms with this much of a power difference.
 
Gears of war 5 have impressive destruction Houdini alembic cache destruction


Starts at 08:01, very pretty destruction effects and doable on base Xbox one.
 
The physics look extremly impressive when shooting down the little fortress on top of the mechanical elephant.

I don't believe this would be possible on the old Jaguar CPUs.
I don't think that this is really cou heavy. We had precalculated destruction before. Why shouldn't it work here. Massively downscaled, maybe even some effects missing, ...
I really don't see a problem that this game gets scaled down. There is nothing really CPU intensive that is needed for the gameplay. Even destructive environment elements weren't a problem on PS4.

A good example how does sling affects image quality would be tomb raider that came for xb1/ps4 and even for the ps360 gen.
And yes it won't look that good, but it should work.
 
Last edited:
Scalability works on image quality related settings, but not on game-logic. It has to be the same game with the same AI and same physics.

If the game was made to be playable on PS4, it stands to reason some aspects of the PS5 hardware will be very under-utilized.
 
i think everything is scalable, but the main factor is dev time, and how much worse you tolerate the low end version to run.
There is an interesting channel which shows PC games running (or trying to) in "potato" mode with custom config files to remove a lot of settings.
 
Scalability works on image quality related settings, but not on game-logic. It has to be the same game with the same AI and same physics.

If the game was made to be playable on PS4, it stands to reason some aspects of the PS5 hardware will be very under-utilized.
PS5 CPU has also a dynamic clockrate. So 60fps mode might maximize the CPU performance, while 30fps might maximize the GPU performance. And according to nxgamers videos, there will be a performance mode in this game. After all the CPU is not that much faster when you compare PS4 30fps vs PS5 60fps. Scale back a bit physics and it should run just fine.
I really don't remember one game that couldn't get scaled back because of CPU limits in gameplay. Gameplay affecting physics is normally not that complex and AI is ... well more or less just better/longer scripts and not that CPU intensive for such a game. You can look at AI like on resolution for the GPU. To make it a bit better, you have to invest many more CPU cycles. And most times just more complex scripts are used which does not really need that much more cycles. There is no game so far with "real" AI. So far it is just a better "hit-detection" around the character and scripted outcomes. And we didn't even see much "gameplay"-relevant improvements since the PS2 (open worlds etc was more a streaming issue than anything else and were also possible before).
 
Scalability works on image quality related settings, but not on game-logic. It has to be the same game with the same AI and same physics.

If the game was made to be playable on PS4, it stands to reason some aspects of the PS5 hardware will be very under-utilized.
In a single player game? No, you can use simplified AI of physics, or omit things completely if you are making a bespoke version for lesser hardware. If you are shipping one version that ships across the generations that just changes some settings, sure. But there's no reason to assume that the physics simulation or AI, or anything really, is going to be 100% identical in the PS4 version as it is on the PS5.

Look at all the PS4/XBO games that got Switch releases. Plenty of them have altered game logic, simplified physics, and modified level geometry. AI is tougher to quantify but I would think it's safe to assume there were optimizations there as well. And again I would point to Rise of the Tomb Raider on Xbox 360. There are tons of things altered for that release beyond texture quality and resolution. Snow deformation, for example, is handled completely different. Instead of subtracting volume where characters touch the surface, it adds a simple simple mesh on top of the snow. And there is a bunch of physics objects missing, or objects that omit collision after the first interaction. Stuff like that.
 
Status
Not open for further replies.
Back
Top