Digital Foundry Article Technical Discussion Archive [2011]

Status
Not open for further replies.
I must say that there are several great graphical set-pieces in this demo already, but did anyone else have a genuine wow moment when you see the cargo hold of that massive tanker? With the bleak sun and the various wind/smoke effects, and at the same time an immensely crisp and detailed image? The sense of scale is much better than in most games I've been in. Must be an awesome scene in 3D (though of course you'd lose some of the crispness).

This time I've also been paying more attention to the textures, and they look just so much better, it's not even funny. Thank you MLAA. So much going on though, I think I'll be playing this one hour at a time in the final game just to not overload my brain! :LOL:
 
1/4 res effects are indistinguishable from full buffer effects as long as they don't take up more than 1/4 of the screen though.
 
1/4 res effects are indistinguishable from full buffer effects as long as they don't take up more than 1/4 of the screen though.

Nah, the low res effects are plain as day, impossible to miss. I would even challenge the assumption that they are 1/4 res, some look lower to me.
 
I think a little of the talk of the buffers is disingenuous. It's used because it's very hard to tell the difference while in motion. Still shots will obviously make the problem more glaring.

That said, Nebula those shots don't look super problematic.
 
I think a little of the talk of the buffers is disingenuous. It's used because it's very hard to tell the difference while in motion. Still shots will obviously make the problem more glaring.

One could says the same about texture res in games etc but at some point one will still notice it and it will stand out. Especially stationary effects like fires and smoke.
 
Yeah it's blatantly obvious no mather playing or seeing videos/shots. Extremly low-res effects and while some dont stand out to much due to design those that have more detail on the particle textures like explosions and fire directly reminds about PS2 esque effect quality. Also since game uses MLAA and has cleaner and sharper IQ it stands out more than it would in KZ2 where the whole image was greatly blurred minimizing disparity.

:???: Having not played the game I don't see how you can say this. Playing the beta the low res effects aren't noticable during gameplay at all, certainly nothing like PS2 quality. By no definition was Killzone 2 "greatly blurred" either, quincunx gets far too much of a bad rap.
 
:???: Having not played the game I don't see how you can say this. Playing the beta the low res effects aren't noticable during gameplay at all, certainly nothing like PS2 quality. By no definition was Killzone 2 "greatly blurred" either, quincunx gets far too much of a bad rap.

Well, it's true in kz2 but in kz3 not so much. The explosion are good enough even in the long distance where in kz2 were a complete disaster. About the QAA vs MLAA well... MLAA is exceptional to cover the jaggies but leaves a lot of shimmering in the fine edges; kz2 it's more clean but it's very dark compared to kz3. I don't know honestly if the game could be so better with QAA with so much high contrast, even without HDR.
 
:???: Having not played the game I don't see how you can say this. Playing the beta the low res effects aren't noticable during gameplay at all, certainly nothing like PS2 quality. By no definition was Killzone 2 "greatly blurred" either, quincunx gets far too much of a bad rap.
If you make a Hong Kong PSN account, you can download the Killzone 3 SP demo without PSN+.

I've played through it already. The low res particles are definitely noticeable. They aren't all that bad in my opinion, but they are still blatantly low res.
 
Umm, so what are we talking about? I guess not cutscenes in general, so what type of cutscene are you referring to? Ones that apply only to fps games, or only in between level changes, or what? I was speaking generally, but sounds like you had a very specific case in mind.

I'm talking about the cutscenes that you think are (or can be) used to mask loading, not MLB 2k7 stadium loading, or most intralevel stuff.
 
I'm talking about the cutscenes that you think are (or can be) used to mask loading, not MLB 2k7 stadium loading, or most intralevel stuff.

Ok so indeed cutscenes used to mask loading. So then back to where you said this:

The best way of course is that you start streaming post-cutscene data before the cutscene as seen in Uncharted games.

How can you load post cutscene data before the cutscene? If you are using all the memory in the machine to display a section of a level, where would you stream these new assets to in memory? That confused me until you posted this:

A linear fps should be designed around a streaming solution.

Which makes more sense now because then you are not speaking in the general sense, you are talking about games that are designed to only show a small subsection of a level at a time and hence will have spare memory into which to stream new assets to. That is far too limiting though in the general sense of gaming because it very much limits your design. Hence why most people just play a video to mask the loading. That way they can design the game as they like and use the machines entire available memory to display it (minus a buffer for video), then start playing a video during a transition while streaming new assets in the background, then when video is done user is back to a new section that is using all available memory again to display what's there.

What you are describing is far more limited in both scope and design, and all but requires that every game be designed such that during a transition the player needs to be shoveled somewhere in the level that only needs a small subset of memory to display so that the rest of the assets can be discarded from memory and new ones loaded in there. These new assets likewise also have to also represent a small subset of the new level otherwise they won't fit into memory. This is all fine and dandy if your priority is streaming and not game design, but it simply won't work with so many types of games out there.

Which brings me back to my original point where if a video is indeed used to mask the loading of a new section then it shouldn't hog all the bandwidth of the storage device.


:???: Having not played the game I don't see how you can say this. Playing the beta the low res effects aren't noticable during gameplay at all, certainly nothing like PS2 quality. By no definition was Killzone 2 "greatly blurred" either, quincunx gets far too much of a bad rap.

What size tv do you play on, and at what distance? I'm genuinly curious because KZ2 is the only game I've ever played that got me to take my glasses off and wipe them thinking that they were dirty, only to realize that no, the game actually looks that blurry. It's the blurriest game I've ever played by far. Likewise on particle effects in KZ3, they really stand out as being low res, but I noticed that really easily in many PS3 games. Like Uncharted 2, it's particle effects look bad to me as well and really stand out like a sore thumb even though they tried to smooth them out. For refernce, my eyes are 12 feet away from a 65" screen, hence why I'm curious what your viewing setup is. Maybe that would explain why stuff that is patently obvious to me is totally not noticed by others.
 
Last edited by a moderator:
What size tv do you play on, and at what distance? I'm genuinly curious because KZ2 is the only game I've ever played that got me to take my glasses off and wipe them thinking that they were dirty, only to realize that no, the game actually looks that blurry. It's the blurriest game I've ever played by far. Likewise on particle effects in KZ3, they really stand out as being low res, but I noticed that really easily in many PS3 games. Like Uncharted 2, it's particle effects look bad to me as well and really stand out like a sore thumb even though they tried to smooth them out. For refernce, my eyes are 12 feet away from a 65" screen, hence why I'm curious what your viewing setup is. Maybe that would explain why stuff that is patently obvious to me is totally not noticed by others.

Just a 19 inch and I'm quite close. Generally claims of extreme blur don't seem to be visible to me. I even tried the Ghostbusters demo on PS3 with it's 540p+QAA and it wasn't that offensive (aliasing noticable but not blur).
 
Just a 19 inch and I'm quite close. Generally claims of extreme blur don't seem to be visible to me. I even tried the Ghostbusters demo on PS3 with it's 540p+QAA and it wasn't that offensive (aliasing noticable but not blur).

Ok so perhaps different sensitivities. I freely admit that blur drives me bad, I hated it on the C2 demo as well and was pleasantly surprised to see it gone in the KZ3 demo. But low ress particles and effects at least to me are really easy to spot, and they stand out more in KZ3 than KZ2 because the rest of KZ3 is actually sharp whereas all of KZ2 was blurry.


Load it to the HDD rather than the RAM Then stream it to the memory at faster than optical media speeds. Simple, yet genius.

Except that you still then have to get it into to ram at some point, run code to prepare/process all the data, etc... So you would not get a no loading screen transition that way. Look at existing PS3 games that use the hdd + optical drive for loading assets and note how they can still be relatively slow to load. Most games that do what betan suggests will funnel people into "load zones", load assets, then gradually funnel people out of load zones. That works for some games, but it's not a general solution. Hence why good old fashioned cutscene videos are still in use to mask loading.
 
Ok so perhaps different sensitivities. I freely admit that blur drives me bad, I hated it on the C2 demo as well and was pleasantly surprised to see it gone in the KZ3 demo. But low ress particles and effects at least to me are really easy to spot, and they stand out more in KZ3 than KZ2 because the rest of KZ3 is actually sharp whereas all of KZ2 was blurry.

I have to agree that the low res particles wherever present stand out like a sore thumb especially next to areas where the visuals shine. Its probably the biggest visual drawback in KZ3. They dont fit at all considering how well everything else looks. Ruins the "flawless" look expected from a game such as this.

Also is it me or there is noticeable "noise" in the animation during the cutscenes? I think there is some of it during gameplay too. Also when a scene changes, some parts of the next scene dont render instantly. Some objects appear a fraction of a second later.

But overall I am pretty impressed by the amount of stuff going on, the detail and texture fidelity of the overall image (when there are no explosions or fire effects present)
 
Nah, the low res effects are plain as day, impossible to miss. I would even challenge the assumption that they are 1/4 res, some look lower to me.
By 1/4 res, I thought 640x360. Those textures were most likely even lower resolution as you said. Taking the 1/4 res of 512x512 is going to be a lot worse than taking 1/4 of a higher res.
Yeah... NO..
KZ2 effects did not look this bad, I saw it myself when I played the demo where the waves are hitting the oilrig is another highlight. They may indeed be lower than 1/4 res to look this bad, because KZ2 effects aren't blatantly low res like this.
 
Last edited by a moderator:
I am playing Halo Reach right now after some k3 beta multiplayer session. Can you elaborate a bit more since I cant see what you are seeing?

Psssst, that's why it's called an opinion.

I know other people that think Reach looks better, but I know them in person, my surprise was just seeing someone post it in a forum. Usually comments like that aren't seen since the "masses" generally jump on said individuals.

Reach and KZ are too different in design approach to be compared evenly. Both are tech rich games so I have no issue with anyone saying either one looks better.

KZ3 does look great and I love what it does but everything appears to be low-res but with a high-res snow texture on it.
Nearly everywhere I look on halo reach has a great texture and with the theatre mode and more open environments, I think it looks the better game. Also the player models on reach look better and when you consider that the player is the same that is built when using the armoury and wearing your clan symbol it's pretty amazing.
K3 definately is a very big step above K2 but it just doen't amaze me the way halo reach does.
I really wish these 2 game engines were used in other games too.

That's cool man, I think Reach is one of the most gorgeous shooters too. :smile:
 
In most cases I don't think quarter-res buffers for effects and particles are such a bad compromise. Doing z-buffer testing at 1/4 rate can definitely be a problem since visibility is high-frequency, but if what you're rendering is relatively low-frequency then the lower sampling rate won't cause as much error for the particles/effects themselves. A lot of fog and smoke stuff can fall into that category. The problems arise when you have high-frequency detail in a texture that you want to show up.

Either way whether or not it's "noticeable" or "bad looking" is completely subjective and dependent on the viewer, which is the same as any other graphical technique. The only technical aspects we can argue are the error it causes, and the potential performance gains. Personally I don't really notice it when I'm playing the games that use, however sometimes I do notice how good it looks when I see a really nice high-res explosion in a game like Halo Reach. But of course I can notice it in stills or quickly pick out the games that use it, if I decide to actively look for it.
 
Sorry for going OT but can anyone explain why bayonetta on PS3 runs so poorly compared to 360 version?
I bought it today on sale and I can't believe the difference.
I understand the devs farmed it out to sega to port but why?
I have never seen such a big performance difference between two versions.
The 360 version is one of my top ten games of all time and I can honestly say the PS3 version may well be one of my bottom ten.
Is this an example of bad dev tools or a game that truly shows the difference between these two consoles?
Thanks and sorry for taking the discussion OT.
 
Status
Not open for further replies.
Back
Top