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

It's full screen, every pixel. What kind of AA solution indiscriminately blends the previous frame
Wasn't that use in killzone to double the resolution. In a fast game it might result in such behavior.
But you're right, it is more likely that it was the capturing device. But than it should be the same for all captures... well if nobody did something wrong
 
Wasn't that use in killzone to double the resolution. In a fast game it might result in such behavior.
But you're right, it is more likely that it was the capturing device. But than it should be the same for all captures... well if nobody did something wrong

No not at all. Killzone used motion vectors to predict where pixels should be, reprojected them in the alternating fields, and the interpolated what was left. The result of that should be better than every deinterlacer out there, and should look identical to full 1080p most of the time, and always looks identical when stationary.
 
perhaps it's simply an artifact from a capture device capturing at 30fps and blending frames together.

however (while this isn't motion blur), motion blur still makes sense at 60fps because there is still motion that happens between frames, so motion blur will still make motion look smoother even at 60fps.
Even so, just look at this real life comparison where you can clearly see the blur produced from a fast moving object at 30 fps vs 60 fps. From this article:

http://blog.logicalincrements.com/post/116319047721/does-fps-matter-decide-for-yourself

"This slow-motion video comparison shows how much smoother a fast-moving object will appear at 60 vs. 30 FPS".

 
Well yes obviously higher frame rates would result in smaller blur. The motion blur produced should be directly representing the amount of pixels an object has moved between the previous frame and the next.

Also, btw, that "30fps" video isn't actually 30fps. They shot at 60fps with an incredibly fast shutter (hence very little motion blur) and then frame-blended it down to 30fps. If they had shot the 60fps video with a 1/60th shutter, then frame blended down to 30fps, it would have been a better comparison, because then both motion blurs would be natural.
 
Apparently many people noticed the weird ghosting (and excessive blurring) of Project Cars, PS4 edition, day one patch rumored.

http://www.thegamescabin.com/projec...ne-patch-and-how-quality-is-fcking-important/

Because about what I’ve said about the hundreds of people that have looked at our game in the QA process etc etc before it was viewed by the public, there is a small possibility here that people might have spotted a geninue bug, so despite it being a minor thing in the scheme of things, I want to get to the bottom of it. Call me OCD but a few hours of feedback conversion to perfect a possible visible error to a few percent of people who pay for our game is f*cking important.

Gfwa3uC.gif

9wPO2Og.gif
 
As a last-frame whatever, whether 'motion blur' or bugged AA, it'd be easy to remove and the game would be better for it.
 
I don't like the effect, but the slowed down GIFs make it look A LOT worse than it actually does in motion. Just watch the gamersyde videos. Still, hopefully SMS can address it.
 
I don't like the effect, but the slowed down GIFs make it look A LOT worse than it actually does in motion. Just watch the gamersyde videos. Still, hopefully SMS can address it.
Even if the gifs are slowed down, if the blur is continuous in any version, wouldn't you notice it? Knowing how important IQ and calibration is for you, I have a feeling you'd see it. I for one noticed the temporal AA in Halo Reach, and hopefully they will remove it in the TMCC version
 
I do notice it slightly at 60fps, I'm just saying it's not nearly as pronounced as the GIF above makes it seem. Knowing how open SMS are to feedback, I think this will (hopefully) get addressed. If not, it's not a deal breaker IMO.
 
Last edited:
Response from SMS regarding the ghosting on PS4:
SMS said:
It's not a bug. It's a side effect of the way we do AA on PS4, and is arguably a good thing (hear me out!). It's there, and very noticeable from static images of non-direct feed footage especially, but also undeniably in the game itself (mainly when paused). I spotted it and reported it as soon as it got checked in long ago (so we didn't "miss it" as many would have you believe).

However, I also "spotted" that the PS4 version has really lovely image quality, a very smooth and slightly soft (not low res!), "non-gamey" look, which is very natural, and (imo!) rather fantastic, due to this very clever AA approach. It also greatly minimises the distant "shimmer" you often see in console racing games, which is a side effect of low quality AA on thin vertical objects such as fences.

So yes, the ghosting effect is there, if you look hard enough. The advantages far outweigh this small disadvantage however. Notwithstanding that, if customer feedback demands it, we have a plan to add a UI slider in a future update, which will enable you to tune out the ghosting, at the expense of more aliasing and distant shimmer.
So basically both console versions use EQAA but they added this (temporal?) AA to the PS4 version which they think overall improves the IQ. They will add a slider if enough people complain about it.

Not sure what AA was used in the DF comparison, but the PS4 version did indeed seem to have considerably better IQ, even considering the resolution differences.

Again, in stills it looks horrible, but in motion it's not that noticeable to my eyes. If it does considerably improve the IQ, then I'd be fine with it. A slider would probably be a good idea though.
 
Last edited:
Again, in stills it looks horrible, but in motion it's not that noticeable to my eyes. If it does considerably improve the IQ, then I'd be fine with it. A slider would probably be a good idea though.
Well, that's basically how we should be viewing games these days. Methods that make use of the temporal variance in games won't translate well to still. Likewise artefacts of temporal variance like texture shimmer won't appear in stills.

As an AA technique, I'm not sure what's going on. One possibility is that the very slight delta on distant changes means blended frames works very well. If so, they could mask the blending based on depth (or even just centre of screen!) and reduce the trails while still maintaining the blend where its needed.
 
So it is indeed ghosting from temporal AA...I am curious now, how this game looks in real life?! Temporal AA when working well really gives a good IQ imo.

I did not plan to buy it, but I really like how the developers react to the community and am thinking about supporting them just because of this (also, I like racing games, but just don't have time for those GT type of games anymore and hence typically don't buy them).
 
I'm unconvinced that their temporal AA is a true temporal AA as we are used to talking about it. Blending a previous frame may reduce temporal aliasing but it's not really an AA technique.
 
I'm unconvinced that their temporal AA is a true temporal AA as we are used to talking about it. Blending a previous frame may reduce temporal aliasing but it's not really an AA technique.

DMC 3 had the same technique, I think. Not that many people liked it then either. If I am not mistaken, GT5 has it as an option (you can choose 3 different AA methods I think)
 
That's blending a full frame at 50% with the current frame. This can work as temporal aliasing, because it makes aliasing from motion 50% less visible, but it's not traditionally what we mean when we use the term. I'm playing MGS2 (2001) right now and it's using the same effect. It's incredibly outdated and it's ridiculous that any game would still use it today.
 
Back
Top