Digital Foundry Article Technical Discussion [2021]

[snc]

Intresting is that in some cases the difference can be 36%, which is quite close to hitman 3's performance differentional, besides the XSX spending much of its time in the 50s and PS5 in the 40's fps range.

we don't know h3 perf difference only that on ps5 runs 1800p and native 4k on xsx, its 1.44x more pixels but also there is this Mendoza level near dense folliage and using sniper that ps5 has around 1.4x performance advantage edit: and iirc using camparison to rx5700xt in miami scene xsx has ~1.1x advantage over ps5

 

[Dictator]

Digital Foundry's frame rate analysis seems to be slightly off in their Control benchmark video. It looks like the tearing on Xbox Series X is being registered as unique frames which results in the frame rate on Xbox Series X being shown to be slightly higher than it actually is. This should be visible in the image here.

For this video I used a 0 compression scan to generate the framerate data - with 0 compression it cannot say the framerate is higher than the individual change in single pixels between images. It will not say the framerate is higher than it actually is unless the game is outputting a dithered output (playstation 5 does this unless you set the console to 4:2:0 limited). If I had used a 1 or 2 pixel image compression scan, then what could happen is that the framerate is registered as being lower than it actually is as small differences between frames are not being registered due to the pixel scan resolution. There is no tearing in control on console as the game uses a triple buffered vsync - at least that is what remedy has said. What happens when the game turns on photomode is that the game loses the internal 30 fps cap, not that vsync is turned off. The game is still syncing to the display.
Some of the game's screen buffer distortion affects when you run the camera through them and they intersect with the camera may look like screen tears.

EDIT: I am just gonna log off the net today - getting to involved in these discussions for my own health!

 

[Shortbread]

For this video I used a 0 compression scan to generate the framerate data - with 0 compression it cannot say the framerate is higher than the individual change in single pixels between images. It will not say the framerate is higher than it actually is unless the game is outputting a dithered output (playstation 5 does this unless you set the console to 4:2:0 limited). If I had used a 1 or 2 pixel image compression scan, then what could happen is that the framerate is registered as being lower than it actually is as small differences between frames are not being registered due to the pixel scan resolution. There is no tearing in control on console as the game uses a triple buffered vsync - at least that is what remedy has said. What happens when the game turns on photomode is that the game loses the internal 30 fps cap, not that vsync is turned off. The game is still syncing to the display.
Some of the game's screen buffer distortion affects when you run the camera through them and they intersect with the camera may look like screen tears.

EDIT: I am just gonna log off the net today - getting to involved in these discussions for my own health!

Health comes first.

 

[PSman1700]

EDIT: I am just gonna log off the net today - getting to involved in these discussions for my own health!

Wise decision. Again thanks for your input here on the B3D forums

 

[Shortbread]

Prepare the beverage of your choice, sit down and ready yourselves for everything you need to know about the new Nioh Collection for PlayStation 5. We've got detailed coverage of both Nioh and Nioh 2 here, including a breakdown of how the 4K, PS5 'quality' and 120Hz modes work on both titles - plus of course, we stack it all up against the same games running on PlayStation 4 Pro.

 

[snc]

21.44s to load level on ps4 pro and 1.1s on ps5

 

[iroboto]

I don’t mind if people disagree with DF results, but the level of discourse has gone to complete shit here. Alex and DF spend tons of hours analyzing games and video feeds for years now, if you’re going to step up to the plate, do some heavy due diligence that is equivalent to the number of hours they put in if you can find reproducible issue with their calculations.

pointing out differences and making claims without due diligence is equivalent imo to complete shit posting; you’re looking for random shit to discard their work and asking him to defend 1 thousand points they are out of his control.

this is not intelligent technical discourse that aligns with this forum.

 

[PSman1700]

this is not intelligent technical discourse that aligns with this forum.

Well, the first post clearly warns such behaviour will be dealt with, this due to other incidents where DF members got attacked and bad mouthed. Now it happened again and Alex even got called for sucking it to MS and whatever. Lets just hope the user got dealth with, abit of a show of force so this will not happen again.
We want to keep people from DF (and others) in here. Him posting to want to take a break because of those discussions isnt really a good sign.

its all facts and numbers, presented to us in the best way possible. There is no way this is BIAS, conspiracies (like Dictator noted here) or anything else like that. Numbers talking here, you cant argue with that. If you can do better, do it yourself and present the results (in another topic).

 

[cwjs]

Some of the game's screen buffer distortion affects when you run the camera through them and they intersect with the camera may look like screen tears.

It looks like the game's texture streamer is imperfect.

Sorry you had to post this because of annoying conspiracy theories, but this is a really cool detail to look into. As somebody with access to both platforms, recording equipment, and a team who understands enough about rendering to look into these, DF is in a rare position of being able to poke around and analyze stuff like this.

Would love to see more stuff like this in your videos -- I'm guess the audience that actually wants this is probably quite small though (and i guess it could just mean even more noise for people to post gotchas about)

Oh wow is RT that heavy on RDNA2.

some surprising results in this video they posted a few weeks back

 

[see colon]

The mel brooks movie we never got

Spaceballs is a real movie.

 

[iroboto]

The rest of the post was referring to handling noise in the video feed and isn't relevant to analysing footage with tearing on Xbox. When a game tears and the next frame is a duplicate then if the algorithm used can't handle tearing then it will result in the frame being registered as unique. If you step through the frames in the video at 1:33 at the point where I took the screenshot then you can see a slight inaccuracy where the frame time graph show 16.67ms when there is clearly a duplicate frame.

Btw this isn't meant to be an attack on DF and even factoring this in the Xbox Series X is still running the RT mode at a higher average frame rate than PS5.

If your frame tracking software is showing different results than DF, and you can make adjustments to replicate DF results, I think you should bring it to Richard and team. Before you take that discussion here on public forums.

It’s just feeding a frenzy right now that may be entirely unnecessary for DF to defend themselves from until it’s proven by both sides. I’m more than positive if they are wrong they’d make corrections as required.

 

[snc]

If your frame tracking software is showing different results than DF, and you can make adjustments to replicate DF results, I think you should bring it to Richard and team. Before you take that discussion here on public forums.

It’s just feeding a frenzy right now that may be entirely unnecessary for DF to defend themselves from until it’s proven by both sides. I’m more than positive if they are wrong they’d make corrections as required.

lets chill out a little ;d, one post here for sure was unacceptable but it was other user and hard to name VGA post offensive

 

[iroboto]

lets chill out a little ;d, one post here for sure was unacceptable but it was other user and hard to name vga post offensive

it lacks due diligence and is every bit as leading without verification. It’s not offensive sure, but if you can’t spend time to verify with DF the results, you shouldn’t claim their frame rate counter is wrong.

 

[snc]

it lacks due diligence and is every bit as leading without verification. It’s not offensive sure, but if you can’t spend time to verify with DF the results, you shouldn’t claim their frame rate counter is wrong.

"Digital Foundry's frame rate analysis seems to be slightly off " - it wasn't as declarative

 

[VGA]

If your frame tracking software is showing different results than DF, and you can make adjustments to replicate DF results, I think you should bring it to Richard and team. Before you take that discussion here on public forums.

It’s just feeding a frenzy right now that may be entirely unnecessary for DF to defend themselves from until it’s proven by both sides. I’m more than positive if they are wrong they’d make corrections as required.

I'm confident that what I've posted is correct, but I underestimated how bad things were for DF surrounding this. I'll send Alex a message on here about this and if the mods could delete my posts that would be great.

 

[iroboto]

I'm confident that what I've posted is correct, but I underestimated how bad things were for DF surrounding this. I'll send Alex a message on here about this and if the mods could delete my posts that would be great.

Thank you

 

[Shortbread]

I'm confident that what I've posted is correct, but I underestimated how bad things were for DF surrounding this. I'll send Alex a message on here about this and if the mods could delete my posts that would be great.

VGA can you please explain how screen tearing or torn frames can cause unique frames to register?

 

[VGA]

VGA can you please explain how screen tearing or torn frames can cause unique frames to register?

If the algorithm used to analyse the frame rate can't detect tearing then a duplicate frame immediately after a torn frame will be registered as unique as the pixels above the tear line on the previous frame will be different to those pixels on the current frame. I've uploaded an image here that demonstrates this where the white pixels are different compared to the previous frame and the black pixels are the same.

 

[cwjs]

If the algorithm used to analyse the frame rate can't detect tearing then a duplicate frame immediately after a torn frame will be registered as unique as the pixels above the tear line on the previous frame will be different to those pixels on the current frame. I've uploaded an image here that demonstrates this where the white pixels are different compared to the previous frame and the black pixels are the same.

Why are there a bunch of black pixels above the "tear line" -- is that common for screen tearing?

Edit: watched through that part of the video. The camera is panning into one of the "blood" vfx that comes out of the enemies when they're shot -- this effect distorts the frame (maybe the previous frame?) and is very visible in all videos of control. I'd guess you're looking at the frame almost entirely intersecting the effect, (and maybe the effect updating at half speed or something? not sure how it works).

 

[VGA]

Why are there a bunch of black pixels above the "tear line" -- is that common for screen tearing?

Sometimes not every pixel is different from one frame to the next which is why some of the pixels above the tear line are black. Yes, it is common.