• 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)

bj00rn_

Banned
I'm curious. If they don't mean the quality of car models, then what?

_Accurate_ car models, not necessarily detailed car models. This has been harped on for years on gaf, so it shouldn't come a surprise. F.ex. Forza infamously had some very inaccurate car models in their first couple of games, while GT has always been pretty consistent.
 

Noobcraft

Member
Yea... well this looks bland to me... and not even close to realistic.

driveclub_20150103110tzu89.jpg


Look at those...hills... or whatever they're supposed to be? Looks terrible.
I think that's more a result of PBR making the wet dirt look like clay.
 

Corine

Member
A lot of these tracks ARE actually quite bland in real life. The thing about DC is that the environments look like clay half the time and have that added benefit of being 'based on and taken inspiration from real life regions around the world.' So they take a lot of liberties that racers based on real life racing tracks which need to be licensed and thus sadly don't really sport all the gorgeous vistas/eye candy that the DC devs throw in to make their game look good.

Yeah Digi posts a lot of real life vs PC vids and the environments are pretty spot on. Just real life courses aren't always the most visually exciting places to see.

https://www.youtube.com/watch?v=l2pAdWJfmS8

https://www.youtube.com/watch?v=stiFsh-1aH4
 
Yea... well this looks bland to me... and not even close to realistic.

Look at those...hills... or whatever they're supposed to be? Looks terrible.

Ok...? We were talking about PC's environments, though. Not quite sure why you're replying to me with DC shots.
 

twobear

sputum-flecked apoplexy
Driveclub actually does switches models between gameplay and photomode but it's subtle enough that people don't notice it. But it is completely false when people say the gameplay model and photomode model are exactly the same. The game switches to a higher LOD model as soon as you go into photomode, even when you do not apply any processing.

Proof?

gameplay:


photo mode:



Open the images side by side and flip between gameplay and photomode to see the changes. Image credit goes to Karamazov on Beyond3D.
How do you explain this difference other than model swapping ?

EDIT: Note that in Time trial mode the poly count is the highest out of any game mode and close to the photomode.

Honestly I'm always a bit suspicious when people say 'there's no difference' between gameplay and non-gameplay models. In fact it's almost bad design on behalf of the game's creators if it's true, because most of the time you won't notice a difference and those polygons could be better spend elsewhere.
 
Yeah Digi posts a lot of real life vs PC vids and the environments are pretty spot on. Just real life courses aren't always the most visually exciting places to see.

https://www.youtube.com/watch?v=l2pAdWJfmS8

https://www.youtube.com/watch?v=stiFsh-1aH4

This isn't the issue I have with PC, though. It's not the real life tracks/locales that I think make the environments look bland, it's the assets used for the environments in those tracks/locales. The grass, the tree models, the ground textures, etc.

And the car models being as good as they are only highlights the bland environments even more to me.
 

RankoSD

Member
Yes but if it was possible to run it at full LOD in gameplay it would have been running it. But since it isn't it can only be attributed to performance reasons. The argument was that the game has the exact same models ingame as the ones you see in the photomode photos.
True.

Gameplay -> Lower LOD -> Realtime
Photomode -> Highest LOD -> Realtime
Pic taken in photomode -> Highest LOD -> Post-processing
 

bj00rn_

Banned
Yea... well this looks bland to me... and not even close to realistic.

driveclub_20150103110tzu89.jpg


Look at those...hills... or whatever they're supposed to be? Looks terrible.

That does indeed look terrible. Almost looks like a tsunami, or some other liquid, coming over some hills. DC has some of the best and worst ups and downs.
 
Yes but if it was possible to run it at full LOD in gameplay it would have been running it. But since it isn't it can only be attributed to performance reasons. The argument was that the game has the exact same models ingame as the ones you see in the photomode photos.



I don't think it's ever "identical" but I can't say for sure. I don't have the time to check it out myself, I'll probably drop a message to Karamazov on Beyond3D asking about it.

It might be that the maximum LOD kicks in during gameplay when zoomed in. Photomode could just be that bit more aggressive and trigger it earlier.
 
Honestly I'm always a bit suspicious when people say 'there's no difference' between gameplay and non-gameplay models. In fact it's almost bad design on behalf of the game's creators if it's true, because most of the time you won't notice a difference and those polygons could be better spend elsewhere.

I think Gran Turismo did it best with its tessellation. There isn't technically a defined LOD model for each stage of detail on a car, the game just adds more polygons to smooth out details wherever the camera is closest to. I can imagine Forza possibly adopting this for their next game.
 

nOoblet16

Member
Honestly I'm always a bit suspicious when people say 'there's no difference' between gameplay and non-gameplay models. In fact it's almost bad design on behalf of the game's creators if it's true, because most of the time you won't notice a difference and those polygons could be better spend elsewhere.

Yes that is correct it's a waste to spend resources on areas where it won't be noticed. Say the game used photomode models for gameplay then it could have used even higher detailed models for photomode...you draw the line somewhere basically. And once you do you can go higher than than in photomode or not do that at all to please the people so that they can say "Look exact same models !".

If I was a developer I'd rather have my prettier photomode models for screenshots than use the exact same gameplay model in photomode just to stroke my ego.
 
You aren't sure why I'm countering the notion that pCARS alone has bland environment assets with pics of Driveclubs bland environment assets?

It was just strange to me how we were talking about PC, then you come out of nowhere with an pretty childish-sounding post mentioning another game entirely. You can mention DC all you want, but you'll be talking to yourself, because I was talking about PC's environments.

But just to be clear, I agree with you. The environments in those DC shots look pretty bad to me.
 

Game4life

Banned
I am excited to see which racer which will top Driveclub graphics wise. I think GT7 will look stunning but I doubt given the 60fps target it will dethrone Driveclub. Maybe some multiplat maxed out on PC will eventually topple Driveclub. Eyeballs will pop out when that happens that's for sure.
 
I'm just gonna be honest, the reason why some people think DriveClubs environments look so nice is because of it being physically based rendered, and because when you are playing the lighting, motionblur combined with the 30fps framerate makes the game look somewhat realistic. When you stop and look at the environment, you see it's pretty clay-like due to the fact that the PS4 has terrible AF and that there's still limitations to their rendering engine.

The gifs look nice... but when you play the game on a huge TV and stop and look at the environment.. it breaks down big time. IMO, pCARS is the same, and I do realize that it has lots of low quality assets... I've backed the game since pretty much the start, so I've also seen how far it's come over time.. however, it looks incredible in motion, especially with everything maxed and at 4K. It's on another level.
 

nOoblet16

Member
Alright just confirmed it, Time trial mode in Driveclub = same as photomode.
Photomode in normal races = highest LOD for all the cars in race.

https://forum.beyond3d.com/posts/1816563/


I'm just gonna be honest, the reason why some people think DriveClubs environments look so nice is because of it being physically based rendered, and because when you are playing the lighting, motionblur combined with the 30fps framerate makes the game look somewhat realistic. When you stop and look at the environment, you see it's pretty clay-like due to the fact that the PS4 has terrible AF and that there's still limitations to their rendering engine.

The gifs look nice... but when you play the game on a huge TV and stop and look at the environment.. it breaks down big time. IMO, pCARS is the same, and I do realize that it has lots of low quality assets... I've backed the game since pretty much the start, so I've also seen how far it's come over time.. however, it looks incredible in motion, especially with everything maxed and at 4K. It's on another level.

You know those leave shots people throw around? They are beautiful and immaculately rendered indeed, but they are actually 2D sprites that are nicely shaded.

Driveclub is the best looking racing game out there right now, the game uses a plethora of high tech visual effects with high quality. But it should come off as no surprise that it cuts corners in a lot of areas. However, it is very clever about them and manages to pull through most of the time giving an appearance of a flawless look. A game that doesn't cuts corners in visuals doesn't exists, because if it did we'd arrive at the end of the line for visual effects.
 

Game4life

Banned
Yeah Digi posts a lot of real life vs PC vids and the environments are pretty spot on. Just real life courses aren't always the most visually exciting places to see.

https://www.youtube.com/watch?v=l2pAdWJfmS8

https://www.youtube.com/watch?v=stiFsh-1aH4

Stuff like this is not impressive anymore in 2015. GT5 on ps3 looked like this - https://www.youtube.com/watch?v=-Mi_xKz3WLw

Like I am not sure what is impressive about these PC videos. We still have not received a true next generation sim jump yet is what I can gather for the above videos in terms of visuals. Hopefully Forza 6 or GT7 bring it.
 
It was just strange to me how we were talking about PC, then you come out of nowhere with an pretty childish-sounding post mentioning another game entirely. You can mention DC all you want, but you'll be talking to yourself, because I was talking about PC's environments.

But just to be clear, I agree with you. The environments in those DC shots look pretty bad to me.
First of all, I don't ever remember naming you specifically when I said that 'people say pCARS looks bland' which isn't an opinion specific to you. And either way, people are comparing pCARS and Driveclub in this thread...this IS the Next-Gen Racing Graphics Face-off thread, after all.
 

ShamePain

Banned
I think Gran Turismo did it best with its tessellation. There isn't technically a defined LOD model for each stage of detail on a car, the game just adds more polygons to smooth out details wherever the camera is closest to. I can imagine Forza possibly adopting this for their next game.

Models have to be setup in particular way in order for tessellation to work without artifacts. It looks fantastic and makes cars perfectly smooth and round in GT6, but it's only applied on a really small amount of cars, not even all premiums. Even Forza's models while boasting 1 mil polygons still have some obvious polygon edges in places.
 
Models have to be setup in particular way in order for tessellation to work without artifacts. It looks fantastic and makes cars perfectly smooth and round in GT6, but it's only applied on a really small amount of cars, not even all premiums. Even Forza's models while boasting 1 mil polygons still have some obvious polygon edges in places.

They definitely aren't 1 million in gameplay. With a 16 car race in Forza 5 that'd be 16 million polygons roughly, without even considering the track.

Since apparently even Driveclub's cars have 250,000 average on the more powerful hardware, with fewer cars per race and at half the framerate, there's no way that Forza pushes 1m per car at 60.

If I had to give an estimate for Forza 5, I'd say that a Forzavista model is around 1 million, since it has all of the interior and engine fully detailed and very few polygonal edges are visible (and since it's only rendering one car it can "afford" to be that detailed). I'd guess that a photomode model is around 250,000, and a gameplay car is around 80,000 - 100,000. For some reason, with AI cars using lower LOD models (I'd say around 40,000 maybe), they don't get more detailed in photomode, so you can see all their rough wheel arches and lights, which is weird.
 
I'm just gonna be honest, the reason why some people think DriveClubs environments look so nice is because of it being physically based rendered, and because when you are playing the lighting, motionblur combined with the 30fps framerate makes the game look somewhat realistic. When you stop and look at the environment, you see it's pretty clay-like due to the fact that the PS4 has terrible AF and that there's still limitations to their rendering engine.

Is there something I don't understand about how AF is implemented by the PS4, or does this not make any sense? How can the PS4 not have good AF? Isn't that like saying 'the PS4 has terrible lighting'?

Plus, plenty of PS4 games have good AF. So surely if the quality of the AF in a PS4 game is terrible, it's because the dev has chosen to implement it that way, not because of the PS4 itself?
 

ShamePain

Banned
They definitely aren't 1 million in gameplay. With a 16 car race in Forza 5 that'd be 16 million polygons roughly, without even considering the track.

Since apparently even Driveclub's cars have 250,000 average on the more powerful hardware, with fewer cars per race and at half the framerate, there's no way that Forza pushes 1m per car at 60.

If I had to give an estimate for Forza 5, I'd say that a Forzavista model is around 1 million, since it has all of the interior and engine fully detailed and very few polygonal edges are visible (and since it's only rendering one car it can "afford" to be that detailed). I'd guess that a photomode model is around 250,000, and a gameplay car is around 80,000 - 100,000. For some reason, with AI cars using lower LOD models (I'd say around 40,000 maybe), they don't get more detailed in photomode, so you can see all their rough wheel arches and lights, which is weird.

I wasn't talking gameplay, just the highest quality models a game contains. It's not gonna be until the next ten until 1 mil can be run in-game, and even then better lighting makes a more dramatic difference to the look of the game than simply boosting polycounts.
 
Is there something I don't understand about how AF is implemented by the PS4, or does this not make any sense? How can the PS4 not have good AF? Isn't that like saying 'the PS4 has terrible lighting'?

Plus, plenty of PS4 games have good AF. So surely if the quality of the AF in a PS4 game is terrible, it's because the dev has chosen to implement it that way, not because of the PS4 itself?

Poor choice of wording on my part. It obviously can support 16xAF however yes, it is indeed up to the developer... sadly, many PS4 games suffer from this since devs tend to want to save bandwidth and opt for 1-4x AF or trilinear filtering in many cases.
 
and even then better lighting makes a more dramatic difference to the look of the game than simply boosting polycounts.

Definitely. Horizon 2's cars are noticably lower detailed than their Forza 5 counterparts, but they look so much better because they're under a dynamic day/night cycle and can be rained on. A car looks nicer with an orange glow of a sunset on it than a boring midday setting, regardless of the detail (to an extent)
 
Poor choice of wording on my part. It obviously can support 16xAF however yes, it is indeed up to the developer... sadly, many PS4 games suffer from this since devs tend to want to save bandwidth and opt for 1-4x AF or trilinear filtering in many cases.

Ok, got ya.

I wish more devs prioritized good AF to be honest. All that work into nice ground textures gets more or less wasted only a few feet in front of the player. It doesn't really bother me in DC, or any other racers, though due to the nature of the genre, you're hardly ever going to be at a stand-still.
 
you guys are wasting your time with the warrior. he goes on and on about DC photomode stupossedly looking vastly better, yet tries to pass this off as a representation of what pcars actually looks like

project-cars-screen-9.jpg


when we all know it actually looks like this

LSeYiDj.png


hes already been shown 10 to 15 comparison shots showing how minor the differences between photomode and gameplay are in DC.
That in-game screenshot doesn't look so good imo. I mean, some rough aliasing can be easily noticed in some parts of the car such as the spoiler... I think IQ wise, PC might be better than DC but DC has a big advantage that opponents don't have. It is weather effect and it makes DC look absolutely amazing in motion.
 
when we all know it actually looks like this

LSeYiDj.png

Is this your own picture? If so you clearly aren't using a very good AA method. And with PC games you can't really go off of a picture, obviously it varies from rig to rig and the game can look stunning if the right settings and ToD are used.

Here I have a similar angle with better AA and a more overcast setting
pcars642015-01-0420-26musc.png


The game also looks a lot better in the rain, similar to Driveclub.

pcars642015-01-0420-2m9uow.png

pcars642015-01-0420-2ucu5q.png

pcars642015-01-0420-42juh9.png


Now the game can look better than my pictures as I only have a GTX 760 but I really feel the game is stunning at times and while Driveclub is clearly superior overall, the IQ of Pcars and weather effects can give sometimes give Driveclub a run for its money. I really think you guys are too harsh on it.
 

ShamePain

Banned
Is this your own picture? If so you clearly aren't using a very good AA method. And with PC games you can't really go off of a picture, obviously it varies from rig to rig and the game can look stunning if the right settings and ToD are used.

Here I have a similar angle with better AA and a more overcast setting
pcars642015-01-0420-26musc.png


The game also looks a lot better in the rain, similar to Driveclub.

pcars642015-01-0420-2m9uow.png

pcars642015-01-0420-2ucu5q.png

pcars642015-01-0420-42juh9.png


Now the game can look better than my pictures as I only have a GTX 760 but I really feel the game is stunning at times and while Driveclub is clearly superior overall, the IQ of Pcars and weather effects can give sometimes give Driveclub a run for its money. I really think you guys are too harsh on it.

I wonder if they'll keep Nvidia banners in the console versions lol.
 
I really feel the game is stunning at times and while Driveclub is clearly superior overall, the IQ of Pcars and weather effects can give sometimes give Driveclub a run for its money. I really think you guys are too harsh on it.

Couldn't agree more but the truth is I have not seen a screenshot that captures how impressive it seems whilst actually playing! Even when I take shots myself I can't believe they are from the same pictures I am seeing whilst playing.

That said the rain effects (especially the windscreen effects) are a ton worse in project cars.
 
Couldn't agree more but the truth is I have not seen a screenshot that captures how impressive it seems whilst actually playing! Even when I take shots myself I can't believe they are from the same pictures I am seeing whilst playing

Yea that happens with a lot of games, when I'm playing AC Unity it looks incredible yet in the screens I take it looks like they've been smudged, weird.

That said the rain effects (especially the windscreen effects) are a ton worse in project cars.
Yes the windshield effects are quite bad indeed, I remember being more impressed with Gran Turismo 5's! Hope they improve those before launch.

pcars642015-01-0420-42ju67.png
 
Yes the windshield effects are quite bad indeed, I remember being more impressed with Gran Turismo 5's! Hope they improve those before launch.

pcars642015-01-0420-42ju67.png

It's the worst graphical part of the game by a country mile, i really hope it gets improved or this thread may become a bloodbath! ;)
 
Ok, got ya.

I wish more devs prioritized good AF to be honest. All that work into nice ground textures gets more or less wasted only a few feet in front of the player. It doesn't really bother me in DC, or any other racers, though due to the nature of the genre, you're hardly ever going to be at a stand-still.

Yeah for sure. We have to remember that these are really still the first generation PS4 titles.. things will get a LOT better from here.

I personally think Gran Turismo 7 will blow these current racers out of the water. I can't wait for it!
 
Yea that happens with a lot of games, when I'm playing AC Unity it looks incredible yet in the screens I take it looks like they've been smudged, weird.


Yes the windshield effects are quite bad indeed, I remember being more impressed with Gran Turismo 5's! Hope they improve those before launch.

pcars642015-01-0420-42ju67.png

Yea, this is one area where it's undeniable that DC's implementation is miles (dat pun) better. And yea, even GT5's implementation is more realistic at this point. pCARS' looks quite bad, especially in certain lighting conditions (at night)

Here's hoping it gets better.
 
Yeah for sure. We have to remember that these are really still the first generation PS4 titles.. things will get a LOT better from here.

I personally think Gran Turismo 7 will blow these current racers out of the water. I can't wait for it!

I have faith in PD, but I wonder if 60fps will limit them in surpassing Driveclub. Personally if they match Driveclubs visuals at 60fps I'll be very happy.
 

Mullah88

Member
Driveclub actually does switches models between gameplay and photomode but it's subtle enough that people don't notice it. But it is completely false when people say the gameplay model and photomode model are exactly the same. The game switches to a higher LOD model as soon as you go into photomode, even when you do not apply any processing.

Proof?

gameplay:


photo mode:



Open the images side by side and flip between gameplay and photomode to see the changes. Image credit goes to Karamazov on Beyond3D.
How do you explain this difference other than model swapping ?

EDIT: Note that in Time trial mode the poly count is the highest out of any game mode and close to the photomode.

show me ingame with hud and then photo mode with the post processing meter not yet applied

edit: and one with everything applied
 
I have faith in PD, but I wonder if 60fps will limit them in surpassing Driveclub. Personally if they match Driveclubs visuals at 60fps I'll be very happy.

True. It all depends on the strengths and weaknesses of the renderer and what they prioritize. And I'm sure doing more physics simulations is gonna be costly, but I think given what they've been able to do with the hardware before (PS2,PS3) they will be able to pull off some incredible things given the relative ease of use of the PS4 compared to the PS3.

We'll see. They've got a high bar to hit.
 
True. It all depends on the strengths and weaknesses of the renderer and what they prioritize. And I'm sure doing more physics simulations is gonna be costly, but I think given what they've been able to do with the hardware before (PS2,PS3) they will be able to pull off some incredible things given the relative ease of use of the PS4 compared to the PS3.

We'll see. They've got a high bar to hit.

Agreed, and don't forget about Forza Motorsport 6, they've had more time with the hardware and now they have more CPU to play with, I think they could impress after the slightly underwhelming FM5. Forza 3 was a very nice looking game at the time, and it looked sublime next to FM2. Let's hope we get the same leap.
 
Agreed, and don't forget about Forza Motorsport 6, they've had more time with the hardware and now they have more CPU to play with, I think they could impress after the slightly underwhelming FM5. Forza 3 was a very nice looking game at the time, and it looked sublime next to FM2. Let's hope we get the same leap.

For sure. That's one thing I gotta give MS credit for. They've been improving the hell out of the SDK for developers and improving the interface/functionality for consumers at a steady rate since launch. That's a good point about the increased CPU usability. It'll be interesting to see what Turn10 says about that for Forza 6, since for us it would be hard to tell what was improved due to that specifically or simply due to being more familiar with the system.

If they can get a similar jump from 5 to 6 as there was from 2-3, then we're in for a treat.
 

Hawk269

Member
Yea... well this looks bland to me... and not even close to realistic.

driveclub_20150103110tzu89.jpg


Look at those...hills... or whatever they're supposed to be? Looks terrible.

Yeah, that pics looks pretty horrid. The only that would make sense if that was in Hawaii and those mountains were black lava rocks. I was in Hawaii not long ago and while it was raining some of the black lava rocks had that look.
 

nOoblet16

Member

Its entirely false if one's talking about time trial where the models are photomode models. Semantics....how do they work ?

Because it does use lower quality models in races with other cars...the proof is right there. How else can one even explain the existence of the lower quality model so close to the camera ?'

i have the game and the difference has never been that drastic going from gameplay to photomode (without post processing added)....

What do you mean it has never been that drastic without post processing added?
This is about additional polygons, post processing would not add those additional polygons.
 
Its entirely false if one's talking about time trial where the models are photomode models. Semantics....how do they work

Because it does use lower quality models in races with other cars...the proof is right there. How else can you explain even the existence of the lower quality model so close to the camera ?'



What do you mean it has never been that drastic without post processing added?
This is about additional polygons, post processing would not add those additional polygons.
I'd wait to see his explanation.
 
Top Bottom