• Hey, guest user. Hope you're enjoying NeoGAF! Have you considered registering for an account? Come join us and add your take to the daily discourse.

Next-gen Racing Graphics Face-off | (Next-gen means current-gen)

You said Driveclub graphics. Last I checked, fps is an important part of graphics. Driveclub is 30fps, which is not at all suited for simulation and good physics (but sometimes ok for sci-fi arcade racers, like Driveclub), so no thanks to that as well. It does have good weather effects, I give it that, but let's be honest here, for "the perfect car racing game" everything else is quite a few steps back.

This is not how a game engine works. The final visual output is determined by numerous processes that happen within thousandths of a second and you have hundreds of layers of data that run at anywhere from 30 to 360+ cycles per second. Forza, for instance, does not update it's physics data once each frame, but ~6 times per frame as they (and most other racing games) have to process what is happening between the contact points of the vehicles and ground as they are traveling anywhere from 50 to 200+ feet per second. That is a lot of height and traction changes that can happen in a single second.
 

RayMaker

Banned
I guess it must be unpatched versions, in store setting and uncalibrated TV's.

Kinda of sucks how when trying a game in store is not representative of the true game , like u would play at home.
 

adelante

Member
then im rly not sure what else to say

It's actually simple really, just say PS3-style shadows, and don't include the term PCF. Doesn't matter whether your posts were quoted in order. Because that term PCF? It isn't what you think it means.

As for shadows not appearing blocky (never mind the fact that it wasn't my original point of contention at all), I was actually gonna use these photos to refute that:

blockystuffmks8o.jpg

But then again, blockiness of some degree can exist in real-life too:


But hey, for argument's sake lets shift our focus to your statement about PCars shadows being "completely devoid of ANY softness whatsoever". That's not what these screenshots seem to indicate (as a bonus, I've included a comparison between portions of the photos - not zoomed - so you can see better):


My point? Nitpicking quality doesn't really get you anywhere. With a game like PCars, I'd imagine a higher graphical setting means better looking filtered shadows, hence the disparity in asset, lighting quality in every shot of PCars we've seen (something inherent to PC games in general). I would rather compare actual visual features or tricks, like DriveClub's screenspace reflections that's implemented in ways you don't see in PCars or any other racers at all. Or bounced/indirect lighting from headlights in DriveClub. Or how the game has water droplets behave on your windshield in a convincing manner, moreso than other racers.

Or, on the flipside, how the headlights on opponent cars in PCars actually casts shadows instead of just the player's like in DriveClub (I could've sworn I saw a video showing off this effect in PCars but I can't find it right now).
 
When lightning strikes!


----

Is it actually noticeably better (do you have any pics detailing the difference)? I've been playing it both pre and post patch, and can't say I've noticed any real improvements.

No, it is quite hard to prove as temporal AA is not kind to screenshots.

I guess you could take my comment about the AA on release:

I am really unhappy with the IQ overall. Not only is it jaggy, but also blurry. The worst of both worlds. If you can't get rid of all the aliasing, at least make it crisp. Use SMAA! FXAA should really be a thing of the past. Game would look so much better.

This really does not apply anymore.
 
I guess it must be unpatched versions, in store setting and uncalibrated TV's.

Kinda of sucks how when trying a game in store is not representative of the true game , like u would play at home.

Even if you play at home people have different tvs and settings that can make games look bad.
 

RayMaker

Banned
Even if you play at home people have different tvs and settings that can make games look bad.

true, Ive just been watching some gamersyde driveclub footage on my tv via my PC and stood up and watched about a 1ft away, and its loads better then the driveclub instore, most odd.
 

HTupolev

Member
No, it is quite hard to prove as temporal AA is not kind to screenshots.
Temporal AA is extremely kind to screenshots, if it's implemented correctly.

Temporal AA works by reusing pixels from previous frame(s) as samples in new ones, generally with subpixel shifts in the pixel grid between frames to ensure that those old pixels aren't redundant with pixels in the current frame. Reprojecting said previous-frame pixels into their correct locations in the current frame is hard to do when things are moving quickly and/or in complex ways.

Some implementations which use accumulation buffers (i.e. KZSF) have "less aliasing" in motion, but that's because they're being liberal with reprojection accuracy, smearing high-frequency details (whereas in still imagery, KZSF shifts between just 2 "correct" sample locations between frames, resulting in an image that's sharper but has some residual aliasing).

But in non-accumulation implementations, it's usually clearly evident that the TAA breaks down in motion. The most obvious example I've experienced is Halo Reach; the TAA doesn't use any sort of sophisticated reprojection, so Bungie had it just turn off for anywhere in the frame where significant movement is detected. But you can also see it in games like Infamous Second Son, where going from slow to medium-speed rotation can cause the size of some stairsteps (i.e. in specular aliasing) to double.

The only way the TAA would "work" in motion and not in still would be if you were using an accumulation implementation, but not actually shift sample locations between frames. In that case, the TAA processing would be totally wasted for still images, and would operate vaguely like a crappy small-scale motion blur for moving stuff.

//=======================

tl;dr The notion that TAA is ideal in motion and doesn't work in still imagery is more or less backwards.
 
My point? Nitpicking quality doesn't really get you anywhere. With a game like PCars, I'd imagine a higher graphical setting means better looking filtered shadows, hence the disparity in asset, lighting quality in every shot of PCars we've seen (something inherent to PC games in general). I would rather compare actual visual features or tricks, like DriveClub's screenspace reflections that's implemented in ways you don't see in PCars or any other racers at all. Or bounced/indirect lighting from headlights in DriveClub. Or how the game has water droplets behave on your windshield in a convincing manner, moreso than other racers.

Or, on the flipside, how the headlights on opponent cars in PCars actually casts shadows instead of just the player's like in DriveClub (I could've sworn I saw a video showing off this effect in PCars but I can't find it right now).

PCars is set in the STALKER universe?
 
Temporal AA is extremely kind to screenshots, if it's implemented correctly.

Temporal AA works by reusing pixels from previous frame(s) as samples in new ones, generally with subpixel shifts in the pixel grid between frames to ensure that those old pixels aren't redundant with pixels in the current frame. Reprojecting said previous-frame pixels into their correct locations in the current frame is hard to do when things are moving quickly and/or in complex ways.

Some implementations which use accumulation buffers (i.e. KZSF) have "less aliasing" in motion, but that's because they're being liberal with reprojection accuracy, smearing high-frequency details (whereas in still imagery, KZSF shifts between just 2 "correct" sample locations between frames, resulting in an image that's sharper but has some residual aliasing).

But in non-accumulation implementations, it's usually clearly evident that the TAA breaks down in motion. The most obvious example I've experienced is Halo Reach; the TAA doesn't use any sort of sophisticated reprojection, so Bungie had it just turn off for anywhere in the frame where significant movement is detected. But you can also see it in games like Infamous Second Son, where going from slow to medium-speed rotation can cause the size of some stairsteps (i.e. in specular aliasing) to double.

The only way the TAA would "work" in motion and not in still would be if you were using an accumulation implementation, but not actually shift sample locations between frames. In that case, the TAA processing would be totally wasted for still images, and would operate vaguely like a crappy small-scale motion blur for moving stuff.

//=======================

tl;dr The notion that TAA is ideal in motion and doesn't work in still imagery is more or less backwards.

Thanks for the explanation, but aren't there also different kinds of TAA? In my experience the temp SMAA in Watch Dogs always looked better in motion. Possible that I am off the mark here, but I recall hearing this multiple times.
 

HTupolev

Member
Thanks for the explanation, but aren't there also different kinds of TAA?
Yes. As I mentioned, for instance, some methods use accumulation buffers, and others reproject directly from the last N frames.

Possible that I am off the mark here, but I recall hearing this multiple times.
Tons of gamers claim that temporal AA is a technique intended to clean up moving content. I think this is simply a misinterpretation; temporal AA with high-quality reprojection can sometimes clean moving things up relative to having no reprojection at all (particularly for shader aliasing in the middle of surfaces), but it's generally more accurate for static imagery (especially on geometry boundaries). The blurring effect that accumulation implementations sometimes has perhaps contributes to the confusion, since an image can look "less aliased" in motion in those cases (and this can arguably be aesthetically preferable in some cases, even if it's not very accurate).
The claim certainly doesn't reflect discussions from developers, in which keeping TAA valid under moving imagery is the big difficulty; getting TAA valid for static stuff is trivially easy, you just shift your sample grid between frames and blend.

In the current run of games, Driveclub's photomode probably offers the best illustration of TAA. When you hold the camera still, the game "supersamples" by blending tons and tons of frames with tiny offsets into the on-screen image, and you can watch the image becoming cleaner and cleaner as this happens. Then when you move the camera, coherency is lost and the temporal samples are rejected and the image immediately becomes aliased again.
 
Ubis recent tsmaa implementations do only function when moving. I wonder if they just turn off all temporal aspects when standing still. Nvidias txaa is more interesting. It functions when standing still but movement improves it further.
 

Javin98

Banned
Ubis recent tsmaa implementations do only function when moving. I wonder if they just turn off all temporal aspects when standing still. Nvidias txaa is more interesting. It functions when standing still but movement improves it further.
While TXAA may offer the best results in motion, don't forget its biggest drawback. It blurs the hell out of textures!
 
it does come with some blur, but people blow it way out of proportion. everything on the screen shimmering and flickering like crazy is much more of an eyesore in motion than having a slightly soft image.
 
I was banking on Project Cars to be the game to make me upgrade my extremely modest GPU (AMD 6950 Unlocked) but in it's current state, at least, thats simply not happening.

I'm a graphics and image quality whore but as it stands....no amount of 4k and other bells and whistles can save Project Cars compared to DRIVECLUB right now. Its not even close.

Can anyone show me better graphics in any other racing game than this?:

https://www.youtube.com/watch?v=duUU0CcAl1E

When Project Cars got delayed in November last year, I was actually quite happy and excited because I thought the extra development time could at least make them catch up somewhat to DRIVECLUB and that still remains to be seen......even though I feel it's a tall order.

Forza Horizon 2 looks nice sometimes but I dont think it's in the running imho.
 
I was banking on Project Cars to be the game to make me upgrade my extremely modest GPU (AMD 6950 Unlocked) but in it's current state, at least, thats simply not happening.

I'm a graphics and image quality whore but as it stands....no amount of 4k and other bells and whistles can save Project Cars compared to DRIVECLUB right now. Its not even close.

Can anyone show me better graphics in any other racing game than this?:

https://www.youtube.com/watch?v=duUU0CcAl1E

When Project Cars got delayed in November last year, I was actually quite happy and excited because I thought the extra development time could at least make them catch up somewhat to DRIVECLUB and that still remains to be seen......even though I feel it's a tall order.

Forza Horizon 2 looks nice sometimes but I dont think it's in the running imho.

Honestly they are very different games and if you are literally after a game purely for graphics then its not going to be PCars. I've put hundreds of hours into it and whilst the graphics are top notch (I'm not one of these who crawls round the track at 1mph hoping to find a flat texture or something!) the rain effects are not a patch on DC, frankly nothing is.

If you are after a deep track racing simulator with huge starting grids, good physics and very good graphics then PCars is ideal.

Edit: I should be more specific, driving in cockpit view the rain effects in DC are a vast improvement over what's in PCars. That being said the cockpit view itself in PCars is better mainly because its adjustable, DC makes me feel like I'm sitting in the boot!
 
Driveclub wouldn't have these graphics if it was 60fps+ and wasn't a closed circuit arcade racer, that was my point. I'm sorry for mentioning Driveclub when it wasn't to go down on my knees and praise it. Now, continue circle-jerking and have a happy new year.

But it's not 60fps and it is a closed circuit racer. It's no use saying 'bu bu but, if X and Y were different, it wouldn't look like it does'. No shit, Sherlock. Still doesn't change the fact that it's arguably the best looking racer on any platform as it stands right now.

Now, accept that and have a bitter new year.

Oh, and fucking lol at the old 'sorry for not joining in with the circle-jerk' excuse. You aren't being called out for valid criticisms or having a different opinion than the positive majority, you're being called out for shitposting.
 

bj00rn_

Banned

I'm so grateful that the Driveclub community is so tolerant and level-minded. You really used restraint in your post and showed me examples of proper behaviour. As I wrote; I'm so sorry for writing something not perfect about Driveclub instead of going down on my knees before it. I've seen the global illumination light through the raindrops. So I'll just go burn my opinions in a big fire now, because Driveclub is god, nothing else matters, let it be heard. Happy new year to you too sunshine.
 

ShamePain

Banned
I hope Forza 6 is 30fps
Get out, a sim is not a sim if it's not 60 fps. You've got Horizon to see what 30 fps gives. 500 Gflops difference between Xbone and PS4 is not going away no matter the framerate. The gulf between PS3 and X360 was very small yet the difference in graphics between Forza and Gran Turismo is massive for example. In fact GT5/6 push some tech that even Forza 5/Horizon 2 don't.
 
I'm so grateful that the Driveclub community is so tolerant and level-minded. You really used restraint in your post and showed me examples of proper behaviour. As I wrote; I'm so sorry for writing something not perfect about Driveclub instead of going down on my knees before it. I've seen the global illumination light through the raindrops. So I'll just go burn my opinions in a big fire now, because Driveclub is god, nothing else matters, let it be heard. Happy new year to you too sunshine.

Lol...

You don't have to write something "perfect", you just have to not shitpost. Like what you're doing now. Not sure why you're finding that so hard to understand.
 

Caddle

Member
Get out, a sim is not a sim if it's not 60 fps. You've got Horizon to see what 30 fps gives. 500 Gflops difference between Xbone and PS4 is not going away no matter the framerate. The gulf between PS3 and X360 was very small yet the difference in graphics between Forza and Gran Turismo is massive for example. In fact GT5/6 push some tech that even Forza 5/Horizon 2 don't.
I think its time you stop posting your bullshit.
 

bj00rn_

Banned
Lol...

You don't have to write something "perfect", you just have to not shitpost. Like what you're doing now. Not sure why you're finding that so hard to understand.

Thank you sir, I'll be forever grateful for showing me the right path. You're a true role-model for forum posters everywhere. May I kiss your feet now?


Anyway, what the next big thing coming up in racing games in 2015 except Project Cars? GT7 is listed as 2015/2016, so definitely 2016 or beyond then, right?
 

wapplew

Member
Thank you sir, I'll be forever grateful for showing me the right path. You're a true role-model for forum posters everywhere. May I kiss your feet now?


Anyway, what the next big thing coming up in racing games in 2015 except Project Cars? GT7 is listed as 2015/2016, so definitely 2016 or beyond then, right?

That open world race everything game from EA? They show some early footage at E3.
 

bj00rn_

Banned
I'm looking forward to that 90's Arcade Racer.

ihgnKo8wdSwRY.jpg

Oh yeah, heard a lot about this when it was announced, then it kinda vanished.

That open world race everything game from EA? They show some early footage at E3.

Didn't catch that at that time. Ok. So Need for Speed is also EA, perhaps that's the reason we haven't heard a NFS announcement (wasn't there like 4 NFS games during two years in the past..).

Forza 6 should hit in 2015. AKA this year...

Ah yes, should be a good title, dev tools and optimizations locked in. With that said, I don't have a Xbox One, but I was surprised when I saw how much better some of the real world tracks look in videos than I thought. Next one should look pretty good. Unless they bite over too much.. I mean how wise would it be to cram in night-day cycles and weather effects in a 60fps sim-like game.
 
Top Bottom