Image Quality and Framebuffer Speculations for WIP/alpha/beta/E3 games *Read the first post*

[COPS N RAPPERS]

ACU even at 800P > AC4 at 1080p, fact. :smile2:

I had a look at AC4's models playing through the game, and they do seem a little less from AC3's polygon standards.

AC4

Spoiler

AC3

Spoiler

Dat arm In AC4, it can fit in a hex valve and turn it!!! ACU blows both out of the water in terms of tech. i stand by my comment.

 

[morphinapg]

The person who said AC4 looked better was talking about the picture quality, not the graphics. They just meant it looked sharper/cleaner. Obviously the lighting and shaders are better in ACU.

 

[shredenvain]

Oddball resolution but basically equivalent to 900P in pixel count. Perhaps as I read, something about needing to be 1080 to easily switch engines with the 1080p old Halo engine?

Why does the excuse 343 used that the old engine has to be switchable, make sense?

Isn't that just a RAM problem and not a GPU one?

Yeah it seems like a ram issue from everything i have read.
From what I have read it sounds like they are drawing both games framebuffers at 60+fps each at the same time for the instant switch. Halo 2 will still look great at that res with all the added graphical features. The other 3 Halos will be native 1080p even Halo ce with instant switch feature.

 

[NotTarts]

Does anyone know what's going on with DriveClub's IQ?

It's the definition of inconsistent. Some edges are nearly perfect while others appear to have no AA at all. There's even 2-pixel wide stepping on some parts of the car. Subpixel AA is similarly a hit and miss.

 

[Karamazov]

i think they had to cut on some things, because there is so much going on at the same time, while being perfectly locked, maybe they kept some room because of the incoming weather patch.
Even with some aliasing issues, in motion this is the first game i play which sometims feel real.

 

[Globalisateur]

They used a very slight post AA, in order to preserve the extra details brought by the temporal AA.

This game is really meant to be admired in motion, where the aliasing is not noticeable.

But yeah the 2 main problems of this game is the low AF and aliasing, both noticeable mostly when the car is not moving.

 

[iroboto]

Not sure if this is good enough as a source, but

Can someone pixel count DA3 for X1?

Really curious about this one since they've been ho hum.
For the sake of it for PS4 as well

 

[Globalisateur]

Not sure if this is good enough as a source, but

Can someone pixel count DA3 for X1?

Really curious about this one since they've been ho hum.
For the sake of it for PS4 as well

Quality of both videos is really really bad, very hard to tell.

Even if you could discern some approximate edges, there are too much artefacts from the (twitch?) compression and I wouldn't trust any results.

 

[iroboto]

Quality of both videos is really really bad, very hard to tell.

Even if you could discern some approximate edges, there are too much artefacts from the (twitch?) compression and I wouldn't trust any results.

Yea figured as much =( Okay we'll just need to wait for something more definitive.

 

[Sh1nn]

DA:I (PS4)

 

[iroboto]

Problem solved!
https://twitter.com/dragonage/status/520628704616407040

Confirmed: #DAI resolution is 1080p on PS4, and 900p on Xbox One. We maximized the current potential of each platform.

 

[Globalisateur]

Analysis of Quantum Break from the 16 mn Gamescon video (the direct link).

There are definitely some slight artefacts of a temporal Anti-Aliasing but it is probably completed by another post AA. Overall the AA is very effective, finding usable aliasing was not an easy task.

From 2 cases of aliasing caused by the lighting I find a 720p vertical resolution. But I am not sure it applies to the whole framebuffer even we can clearly see that the game is definitely <= 900p because of the general blurriness.

12 edges on a 18 pixels height.

Full screenshot:

Spoiler

14 edges on a 21 pixels height.

Full screenshot:

Spoiler

 

[AlNom]

I wouldn't trust specular aliasing.

 

[HTupolev]

I wouldn't trust specular aliasing.

In general, no, but a consistent 720p reading from highlights (from earlier footage, too) is suggestive.

Additionally, those aliases might be geometric in nature, especially from the first image.

 

[AlNom]

What would this be?

(tbh, I think we need to wait for something better than a crappy video compression)

 

[HTupolev]

A nearest-neighbor blow-up of a piece of a (car?), seemingly resolved at native res albeit with some edges weirdly smeared.

(tbh, I think we need to wait for something better than a crappy video compression)

Perhaps.

 

[AlNom]

A nearest-neighbor blow-up of a piece of a (car?), seemingly resolved at native res albeit with some edges weirdly smeared.


Perhaps.

Yeah, there's going to be a couple of issues (at least) analyzing this at the moment. Anything that moves will have problems (be it compression or post-processing), and anything that's static might even give false positives from edge reconstruction.

It's really not that hard to find a youtube video of 1080p games that look just as blurred in similar lighting conditions (dark + post-fx).

 

[Globalisateur]

What would this be?

(tbh, I think we need to wait for something better than a crappy video compression)

Where did you find this material? Is this from the Gamescon video?

EDIT: Ok, found it, it's a barrel in the video indeed. But I am not sure it proves anything, could be the AA + a custom upscaling analytical software doing edges reconstruction. And also this video can't be 1080p native, it's really too blurry.

 

[Shortbread]

Analysis of Quantum Break from the 16 mn Gamescon video (the direct link).

There are definitely some slight artefacts of a temporal Anti-Aliasing but it is probably completed by another post AA. Overall the AA is very effective, finding usable aliasing was not an easy task.

From 2 cases of aliasing caused by the lighting I find a 720p vertical resolution. But I am not sure it applies to the whole framebuffer even we can clearly see that the game is definitely <= 900p because of the general blurriness.

12 edges on a 18 pixels height.

Full screenshot:

Spoiler

14 edges on a 21 pixels height.

Full screenshot:

Spoiler

Is there a possibility Remedy is using dynamic resolution with Quantum Break? I noticed during the cutscenes (gameplay cutscene mixture), IQ looks more polished. I'm not talking about textures, shaders, shadows, etc... but more so, the characters finer details and surrounding looks cleaner during these periods. I'm pretty sure LOD isn't upped during those particular moments, because geometry and texture quality look virtually identical (lighting is a little different, nothing drastic though) between transitioning.

 

[Globalisateur]

Is there a possibility Remedy is using dynamic resolution with Quantum Break? I noticed during the cutscenes (gameplay cutscene mixture), IQ looks more polished. I'm not talking about textures, shaders, shadows, etc... but more so, the characters finer details and surrounding looks cleaner during these periods. I'm pretty sure LOD isn't upped during those particular moments, because geometry and texture quality look virtually identical (lighting is a little different, nothing drastic though) between transitioning.

There is more Depth of field effect during cutscenes. It could explain the difference of perceptual cleanness.

I did not notice dynamic resolution in the video.