DriveClub by Evolution Studios [PS4]

Because it uses an accumulated sample implementation, it gives incredibly accurate results, but if you set up extraordinarily heavy DoF, photomode can actually fail to take enough samples to eliminate the noise, resulting in a grainy appearance (despite the tremendous rendering time!).
What's the longest time for a photo render? Are we talking seconds or minutes?
 
i'd say around 10sec for the progress bar to reach 100%

adds AA, softer shadows, and can draw more details in the scenery (fences in the distance etc...)
contrary to other games photo mode using a black screen, it does that in realtime on the screen so you can see the process in real time.
 
Yep. And if you play around with the shutter speed to create some more interesting effects, you can see the engine creating the samples more easily.
 
I would say that most of the picture finetuning happens druing first second or two. After that it doesn't make noticable changes, at least from my experience with photomoding.
 
What cutscenes?

I consider cutscenes to be anything that's not actual gameplay. For example, the shop mode where you are seeing the car. The car doesn't look anywhere near that quality when you are driving.

A lot of screenshots are photomode captures which are heavily supersampled and enjoy superb MB and bokeh DoF, but "cutscenes" isn't the word I'd use to describe that.

I get what you are saying.

At some level I think it's sort of surprising that you're praising the SSR for being a high-quality implementation in its own right;

Because I can't find an implementation that looks better when compared to other games. Can you?

I have no idea what you're talking about. I know Karamazov has said that some things have better LOD in photomode (thus explaining why photomode does seem to occasionally drop frames), but the aliasing on fine details is exactly the same except when the game is actually going through the screenshot-rendering process, during which time there's not much of a meaningful reference for "framerate" since everything is still (and during which time it doesn't really matter because you're not looking at something which even remotely represents realtime IQ and perf anyway).

When you turn on Photomode, the framerate drops. Even though you haven't taken a picture yet. That's what I'm talking about.

If you're talking about realtime visuals, I'm not aware of any DoF in Driveclub.

Hmm.. there looks to be DOF just before a race as the many cameras are viewing the tracks at different angles.

Actually, the system uses geometric displacements "everywhere", not just on the bumpers. Contrast this shot of a beat-up car with this shot of the same car at the start of the race.

Those displacements are hardly noticeable along the sides of the car. I even took the camera and rotated it about to see if there was any drastic change to the vertices along the side. I couldn't tell..
 
VFX_VETERAN you should have talked about the volumetric clouds in your analysis, because that's cetainly a ressources heavy feature, and adds a lot the the graphics. Watching some timelapse videos really shows how impressive they are, with cloud shadows casting on the scenery; and sun rays passing through. They can even create fog on some tracks when racing through them..

Sorry, missed this one because the clouds didn't stand out to me in the moments I played. While the clouds may be truly volumetric, they seem to affect lighting more so than anything else (which was covered under the weather system). I can modify the article to mention the clouds though.
 
In photomode framerate really Drops a little only if i zoom in with all the cars on screen, but keep in mind that in this mode all cars are highest LOD everytime. but the absence of motion blur may make people feel like framerate is lower overall ?
 
In photomode framerate really Drops a little only if i zoom in with all the cars on screen, but keep in mind that in this mode all cars are highest LOD everytime. but the absence of motion blur may make people feel like framerate is lower overall ?

Nah. I am pretty sure the framerate drops. I wish I had a measuring device like Digital Foundry does.
 
overall it's still a great accomplishment for that $400 box. Of course there are some rough spots but as a whole package it still feels amazing while playing, i it the most detailed racing game i've ever played as of now.

You could add to your negative points that wind affects clouds and particles movement, but not flags, it's not rare to see clouds and drops moving in a direction and flags in the opposite direction :p
 
Why would anyone care about photo mode in a game?

Well Sony and the publisher cares because the share feature via Twitter and Facebook is free advertising. The player cares because they get to capture some really nice shots and share with their friends. Photo mode has been a staple of car games for a while now, people like cars and treat their shots as art.
 
I consider cutscenes to be anything that's not actual gameplay. For example, the shop mode where you are seeing the car. The car doesn't look anywhere near that quality when you are driving.
Okay, I see what you're saying.

I'm not sure that it's true that the car is nowhere near that quality during driving... doesn't that depend on mode? For instance, the car looks pretty fantastic in single-car time trial.

Because I can't find an implementation that looks better when compared to other games. Can you?
As I said, in it's own right. I think most of why it looks better than other implementations is the same reason I noted that it looks awful in photomode: racing game cameras are extremely kind to screen-space reflections.

When you turn on Photomode, the framerate drops. Even though you haven't taken a picture yet. That's what I'm talking about.
I know. Karamazov's claim was that you can get drops if you look at a huge number of cars in photomode simultaneously, and explained this by suggesting that photomode seemed to always used LOD0 on cars.

Regardless, it's obvious that the realtime AA doesn't change in photomode. If you stop, look at some aliases, and switch to photomode, the artifacts don't so much as budge.

Hmm.. there looks to be DOF just before a race as the many cameras are viewing the tracks at different angles.
Not sure about that, but I noticed that there was DoF in the "garage" view, so I see what you're talking about now.

Those displacements are hardly noticeable along the sides of the car. I even took the camera and rotated it about to see if there was any drastic change to the vertices along the side. I couldn't tell..
Did you look at the pictures I posted? It's pretty noticeable there.

Chances are that you didn't beat up your cars very badly; you have to get into pretty nasty collisions for significant displacements to appear on the sides.

I would say that most of the picture finetuning happens druing first second or two. After that it doesn't make noticable changes, at least from my experience with photomoding.
It doesn't make all that huge of a difference if you're just using it for supersampling, but it can need the full progress if you've got heavy DoF and MB and such.
 
overall it's still a great accomplishment for that $400 box. Of course there are some rough spots but as a whole package it still feels amazing while playing, i it the most detailed racing game i've ever played as of now.

Well, I bought Assetto Corsa last night on Steam and while it has no day/night cycle, no volume clouds, no global illumination, and no weather affects or collision parts, its still a better gameplay experience than DC. Firstly running at 4k resolution @ 36fps with 16x anisotropic fitlering displays a much much cleaner image than what DC can muster. I am able to see details way into the distance while driving and the cockpit is much more detailed. Even playing at 1440p @ 60fps offers a tremendously smooth playing experience but the aliasing starts to kick in. 1080p looks significantly worse than DC however. I would hope that the next DC can tone down the fancy graphical features for higher res textures and less aggressive MIPMAP algorithms and perhaps better AA running @ 60fps. Truth be told, we really don't need GI or dynamic day/night cycles to enjoy the driving. Weather FX is debatable I guess.

You could add to your negative points that wind affects clouds and particles movement, but not flags, it's not rare to see clouds and drops moving in a direction and flags in the opposite direction :p

Nah.. it's ok. This game is obviously not trying to mimic real physical reality.
 
for assetto crosa you mean greater IQ, there is no question about it, like Pcars and even forza horizon 2 and it's 4X MSAA, but about the amount of details on the trackside i'd day DC renders more things in the scenery.
 
Did you look at the pictures I posted? It's pretty noticeable there.

Even in this picture:

http://i.imgur.com/bJ4bXyj.jpg

It's very hard to tell that the side is being displaced. The camera doesn't allow you to view the car from any angle so it's going to be hard to judge. With the bumper though, you can clearly see it.

Chances are that you didn't beat up your cars very badly; you have to get into pretty nasty collisions for significant displacements to appear on the sides.

I'll play it again tonight and beat it up pretty good to see if I notice them. Not I'm not relenting on my claim yet. ;)
 
for assetto crosa you mean greater IQ, there is no question about it, like Pcars and even forza horizon 2 and it's 4X MSAA, but about the amount of details on the trackside i'd day DC renders more things in the scenery.

Yes, of course. DC was made with scenery as something important to the driver.. AC seems to concentrate resources on the actual track and cars.
 
Well, I bought Assetto Corsa last night on Steam and while it has no day/night cycle, no volume clouds, no global illumination, and no weather affects or collision parts, its still a better gameplay experience than DC. Firstly running at 4k resolution @ 36fps with 16x anisotropic fitlering displays a much much cleaner image than what DC can muster. I am able to see details way into the distance while driving and the cockpit is much more detailed. Even playing at 1440p @ 60fps offers a tremendously smooth playing experience but the aliasing starts to kick in. 1080p looks significantly worse than DC however. I would hope that the next DC can tone down the fancy graphical features for higher res textures and less aggressive MIPMAP algorithms and perhaps better AA running @ 60fps. Truth be told, we really don't need GI or dynamic day/night cycles to enjoy the driving. Weather FX is debatable I guess.



Nah.. it's ok. This game is obviously not trying to mimic real physical reality.


I disagree weather, lightning system and day and night cycle in Drive Club are very important and are making this game unique. It affects the visibility and made each race unique.

It is THE technical feature of Drive Club and it affects gameplay too...

I remember one of my first event race in India where I lost control of my car because I was not knowing very well the circuit and the visibility of one part of the road was affected by sunset...

And it is dynamic and change during the race...

In Norway some part of a race are under heavy snow and other not...

They choose to not use Forward rendering like many racing game (Forza, Forza Horzon, PCars,GT...) and sacrifice MSAA... But it is the only racing game where I am impressed by dynamic lighting system...

I just hope for next Drive Club game they can optimize the game engine and improve weather system(fog, snow accumulating on the road, melting snow...). Maybe with optimization they can improve texture filtering and AA (for example share with Naughty Dog and see if they can implement U4 AA tech...)
 
Last edited:
Back
Top