Digital Foundry Article Technical Discussion [2021]

[snc]

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.

Are you planning to do own analysis of control photomode fps maybe ?

 

[cwjs]

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).

From a polygon article: "“So we take, for example, an enemy and we take information from the previous frame. We reproject it from the position of the camera enemy in the next frame, and we kind of draw this trail of the enemy," explains Control lead VFX artist Elmeri Raitanen. "But that would be boring to look at so then it gets advected with the fluid simulation and then, after that, it gets broken into the colors of dispersion.”"

Gonna stick with this guess -- I think remedy is not lying about no tearing, but this effect is either introducing something that looks very similar to a tear at the right angle, or some internal rendering buffer used for the effect (fluid sim or something) is tearing, not the frame.

 

[Globalisateur]

Are you planning to do own analysis of control photomode fps maybe ?

We already have such an analysis thanks to DF. I am more interested into a like for like comparisons during gameplay.

 

[Shortbread]

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.

Thank you. So you are VG Tech!?

 

[Shortbread]

From a polygon article: "“So we take, for example, an enemy and we take information from the previous frame. We reproject it from the position of the camera enemy in the next frame, and we kind of draw this trail of the enemy," explains Control lead VFX artist Elmeri Raitanen. "But that would be boring to look at so then it gets advected with the fluid simulation and then, after that, it gets broken into the colors of dispersion.”"

Gonna stick with this guess -- I think remedy is not lying about no tearing, but this effect is either introducing something that looks very similar to a tear at the right angle, or some internal rendering buffer used for the effect (fluid sim or something) is tearing, not the frame.

But wouldn't this affect PS5 as well? So far, VGA hasn't point out this being an issue on PS5 (not saying it doesn't occur).

 

[BRiT]

I don't see how any tearing would be possible using Triple Buffering as the developers indicate they are using.

 

[snc]

I don't see how any tearing would be possible using Triple Buffering as the developers indicate they are using.

maybe its only active during gameplay when there is fps cap ?

 

[Globalisateur]

I thought corridor of doom is most gpu heavy area and difference there is 3%

I think the corridor of doom is bottlenecked by plenty of transparencies (alphas) due to the scene being behind the glass. XSX has more compute but is more limited by its pixel fillrate hence the PS5 is able to reduce the gap here.

 

[Kugai Calo]

Well then explain it in the video when you are doing the comparisons. Could this also not contribute to the fps difference in a minor way. You are doing yourself no favours when showing off screenshots like these and there's differences, it just looks like you are sucking up to MS, which I am sure you are not but it looks like it. It's an impression you are giving off.

Wow you really do live up to your name... Keep fighting the good fight, soldier!

 

[Riddlewire]

Is this photomode present on the 8th gen versions? (also with unlocked framerate)
Those might be some interesting results.

Answering my own question here: Not unlocked.
I reinstalled the game on my One X. Photomode is either capped at 30 or the One X doesn't have the power to push much/any above that. Hard to tell by eye alone.

 

[iroboto]

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.

Please let me know what the results are, I performed an absolute diff on the 2 frames using their 4K H264 encode and found a lot of soft noise below the tear. It's probably not ideal because I'm working on a second encode, but without access to source, I'm not sure how else I can do better in this regard.

 

[iroboto]

I think the corridor of doom is bottlenecked by plenty of transparencies (alphas) due to the scene being behind the glass. XSX has more compute but is more limited by its pixel fillrate hence the PS5 is able to reduce the gap here.

You'll need to explain this data point here if it's alpha or bandwidth based, because IIRC, the 6800XT has it in spades;

 

[Globalisateur]

You'll need to explain this data point here if it's alpha or bandwidth based, because IIRC, the 6800XT has it in spades;

This proves AMD is bad with RT compared to Nvidia. The consoles? well:

2x ray tracing resolution on PC...These results cannot be directly compared to PS5 or XSX

 

[iroboto]

This proves AMD is bad with RT compared to Nvidia. The consoles? well:

Then the majority of this particular benchmark is RT limited and not alpha transparency limited.

 

[Globalisateur]

Then the majority of this particular benchmark is RT limited and not alpha transparency limited.

Why should it be black or white? It's grey. I'd guess some scenes are more limited by RT computation, others more by the raw bandwidth and others more by the alpha effects. When you average those scenes it gives you 16% advantage for XSX.

 

[Lurkmass]

You'll need to explain this data point here if it's alpha or bandwidth based, because IIRC, the 6800XT has it in spades;

A potential reason why AMD on PC is comparatively behind here compared to consoles is that the game isn't updated on PC ...

A good chunk of the performance deficit we see could be fixed if we use their driver DXR extension like we see in Dirt 5 where they probably are using AMD's AGS SDK ...

 

[DavidGraham]

A good chunk of the performance deficit we see could be fixed if we use their driver DXR extension like we see in Dirt 5 where they probably are using AMD's AGS SDK ...

Dirt 5 doesn't use even 1/10 of the RT workloads that Control uses.

 

[Lurkmass]

Dirt 5 doesn't use even 1/10 of the RT workloads that Control uses.

That's still performance that their leaving off the table which can be easily picked up ...

 

[cwjs]

But wouldn't this affect PS5 as well? So far, VGA hasn't point out this being an issue on PS5 (not saying it doesn't occur).

Yeah, so if my guess is correct it'd appear on every platform, but need the exact right position/angle relative to an enemys damage vfx to show. (corollary -- if i am right its also probably quite rare on xbox and only happens under these conditions)

 

[thicc_gaf]

Here we go - to sooth some conspiracy theories. I just went back to Maintanence Access Corridor after restarting the Directorial Override Mission on both consoles, waited a full minute, and made screenshots. With the HUD so people can see which console at is, and without the HUD so people can see the image in full. Here it is possible to see that the texture this time I did it looks different on the PS5 version in fact. It looks like the game's texture streamer is imperfect. A whole lot of fuss over absolutely nothing. The versions are the same.
PS5:

PS5 with UI:

XSX:

XSX with UI:

PS5 renders all games in an HDR container and then applies its own tonemaps down to SDR. So all games in SDR on PS5 will look different than normal SDR as designed by the developer on PS4 or PS4 Pro.

It is something that could possibly be producing a blacked crush look in SDR displayed back compat games.

Ah-ha! Take THAT, King Thrash Gaming!!! Seriously, some of the speculation on this stuff was just getting ridiculous.

On some level, I admire that more people want to do technical analysis but it'd help if they did a bit more research and didn't have thinly-veiled animosity towards a particular platform brand or conspiracy theories about a very specific analysis channel being a shill for said particular platform brand on the grounds of zero evidence.