PS4 Pro vs PS4 Graphical Comparison Thread

The problem with checker-boarding is that it's almost too good. Devs are probably asking themselves why not use it to reach 1080p on vanilla ps4 with half the samples to shade. At this point, the difference from actual native 4k (with uncompressed pixels) to what pro will be able to output will be more discernable.
Rainbow Six already used it (1080p + 60 fps). Killzone Shadow Fall used similar (but simpler) technique in multiplayer mode (1080p + 60 fps).

Checkboard works better in 60 fps games as reprojection errors are halved. 4K makes pixels 4x smaller also reducing errors of this technique. Seems that it works pretty well @ 30 fps as well (at 4K). Most likely 1080p @ 30 fps would show more errors.
 
I just wish more games would start to use it, really bummed that UC4 doesn't use it, just plain old 1440p upscaling to 4k. Horizon ZD is also another great example of it, the textures appeared so much sharper as a result.
Guerilla with Horizon is the new gem studio of Sony IMO to promote their Pro.

The future on Pro belongs to Guerilla (impeccable 4K checkerboard + 30fps with good clarity in motion so rare among the Sony first party games) and Dice (dynamic res + 60fps + realism).
 
Guerilla with Horizon is the new gem studio of Sony IMO to promote their Pro.

The future on Pro belongs to Guerilla (impeccable 4K checkerboard + 30fps with good clarity in motion so rare among the Sony first party games) and Dice (dynamic res + 60fps + realism).

I'd probably put Insomniac above Guerrilla atm in terms of pure IQ on the Pro. The technique used on R&C and the next Spiderman game is pretty damn good.
 
It's the same technique used in Killzone ? So, it's native 4K then ?

I mean, there wasn't any upscale in Killzone.
 
It's the same technique used in Killzone ? So, it's native 4K then ?

I mean, there wasn't any upscale in Killzone.
It's semantics. Rather than fight over whether something can be called "native 4K" (which by some definitions it can, and by others it cannot), we should really just say what's happening.

Also, KZSF does seem to have spatial upscale, it appears to be the fallback when temporal reconstruction fails.
 
It's semantics. Rather than fight over whether something can be called "native 4K" (which by some definitions it can, and by others it cannot), we should really just say what's happening.

Also, KZSF does seem to have spatial upscale, it appears to be the fallback when temporal reconstruction fails.

Ok, thanks.
 
I just wish more games would start to use it, really bummed that UC4 doesn't use it, just plain old 1440p upscaling to 4k.

Cerny has said that Naughty Dog is working more on Pro support. I presume they will cook something else for the release of their upcoming singleplayer DLC.
 
I'm pretty impressed with Pro, much more so in the past week or so than I was initially. And I think things will only get better in time when more developers start to use checkerboarding/temp. reconstruction rather than simply rendering at a higher resolution and using standard upscaling. Seems like developers didn't have much of a heads-up and didn't have much time to fully utilize the Pro's extra power. Respawn flat-out said they didn't have much of a heads-up.

Super-sampling provides better results than I was expecting, especially in RoTR. Framerates in high framerate mode on Infamous and RoTR are also better than I thought.

I think Sony did a very good job of putting together a $400 box here that can definitely punch above its weight.
 
The checkerboarding in TR really looks great even from a compressed YT feed, I can hardly distinguish it from 4k native except some slightly sharper textures on the PC. I'd say they're over 95% the same from pure eyeballing.

The difference looked pretty obvious to me. That's not me putting the PS4P down - some other game comparisons have been indistinguishable from native 4K on the PC IMO. However in the case of TR the difference is obvious from that video.
 
http://gifyu.com/images/8gGwkxA4.png
http://gifyu.com/images/fTiYWpdu.png
http://gifyu.com/images/SrC7MFLA.png
http://gifyu.com/images/uSevjGE0.png

I said wow.... can't freaking wait to see gt like this, this Checkerboard algo is really something

With only 4 tf hitting anything close to 4k esque image quality even by cutting out some of the raw native resolution work is really astounding.

Hats off to the engineers who made this even remotely possible

In near vertical or horizontal high contrast lines the checkboard pattern is clearly visible.
Like in the second picture, the stripes above the white wheel.
 
Also, KZSF does seem to have spatial upscale, it appears to be the fallback when temporal reconstruction fails.
Results would be absolutely horrible if you hadn't a fallback for reprojection failure. Every frame has failure areas (new revealed surface areas). This is where checkerboard beats scanline interlacing. Missing pixels have full set of neighbors in every direction (+-X, +-Y), while in scanline interlacing you only have neighbors in Y direction. In failure areas you get quality similar to 2x 1080p (= slightly above 2560x1440). It is worth noting that every surface pixel that failed reprojection is guaranteed not to fail reprojection next frame (if it is still visible). This means that you see 2560x1440 equivalent detail only for a single frame. This issue is only visible in new surface areas. Your eye & brain needs some time to focus and understand new data, thus at 60 fps it is almost impossible to see this issue. I don't know about 30 fps, since our games have always been 60 fps.
 
What is Sony thinking downclocking both the CPU and GPU when running an old code on the PS4 Pro? I can understand deactivating additional processing units (like the Polaris ALUs), as these might cause some compatibility issues, but downclocking the units as well? why, what possible reason could be behind a move like this?
 
What is Sony thinking downclocking both the CPU and GPU when running an old code on the PS4 Pro? I can understand deactivating additional processing units (like the Polaris ALUs), as these might cause some compatibility issues, but downclocking the units as well? why, what possible reason could be behind a move like this?

http://www.eurogamer.net/articles/d...tation-4-pro-how-sony-made-a-4k-games-machine

Mark Cerny said:
I've done a number of experiments looking for issues when frequencies vary and... well... [laughs] I think first and foremost, we need everything to work flawlessly. We don't want people to be conscious of any issues that may arise when they move from the standard model to the PS4 Pro.

sebbbi said:
In failure areas you get quality similar to 2x 1080p (= slightly above 2560x1440). It is worth noting that every surface pixel that failed reprojection is guaranteed not to fail reprojection next frame (if it is still visible). This means that you see 2560x1440 equivalent detail only for a single frame.

About 1527p (2716x1527)
 
Last edited:
Playing it safe doesn't explain it unfortunately, as mentioned in the article, XO Slim already does this! Any OC'ed PC can do this! Sony doesn't need to offer higher than 30 fps in these old titles on the Pro, just use the extra clocks to smooth those 30fps and maintain a higher minimum frames than the standard PS4!

But the hardware difference is much smaller with the XBS.

PS4 pro =

- +31% for the CPU (clock speed)
- +14% for the GPU (clock speed)
- +24% for the bandwidth
 
Last edited:
The difference looked pretty obvious to me. That's not me putting the PS4P down - some other game comparisons have been indistinguishable from native 4K on the PC IMO. However in the case of TR the difference is obvious from that video.
The only small amount of differences I could make out is the spider web in the cave and Lara's clothing all seemed sharper. I think checkerboarding can't resolve small alpha coverage details as well as native 4k can, but then again PC Ultra by default has slightly higher res textures to begin with which also helps. But overall the two look astonishingly similar.
 
Back
Top