Exclusive photos from the studio Polyphony Digital from Japan

  • Thread starter Dionisiy
  • 199 comments
  • 26,296 views
What's wrong? Because I wrote a . after preposterous? For your information, if that's the case, it's not incorrect.

I surely won't be given a warning because of the grammar I used.


EDIT: Or do you feel offended because I wrote "last page" and apparently only your pic is on the last page? For your information, at the time I was writing that particular post, page 7 was actually the last page.
I had to do something else in the meantime (while writing) and that's why it eventually landed after your post.

Actually, it doesn't matter.

EDIT2: Or maybe you just don't understand the thread title.
There were no ill-feelings involved. I was just acting stupid. You wrote an improper sentence though and that actually made sense, but it wasn't grammatically correct. I do it too, and that's how I was acting stupid when I told you your grammar was incorrect. Honestly, I meant nothing serious by it and am not trying to make a fool out of you. It's the other way around if anything. I'm acting a fool :dopey:.
 
The entire underbelly thing is kind of a moot point, in a game where the car only rolls over once in a blue moon. I rather have those polys in the body.

When you model the whole car, you get access to a better damage model, "autovista"-like modes/better photomode, etc. Just having the pretty mesh is less immersive IMO, but overall, it's a tradeoff. I'd prefer a full model like the picture of the Ford Capri shown from Project CARS.
 
When you model the whole car, you get access to a better damage model, "autovista"-like modes/better photomode, etc. Just having the pretty mesh is less immersive IMO, but overall, it's a tradeoff. I'd prefer a full model like the picture of the Ford Capri shown from Project CARS.

I would prefer that too, but not at the expense of more cars in the game. Only so many resources will be attached to car modelling and if it means 150 new cars but a cool damage model, or 200 cars but the same damage model we have now, I'll take the 200 cars any day. Unless PD is willing to buck up the team to get that kind of thing done, which so far they haven't shown a willingness to do as far as I can tell, then I'd rather not trade off cars to drive for a better damage model. Others may feel differently but I've always felt that damage modelling is not that big a deal to me, more of a gimmick suited to making cool Youtube videos and the "wow" factor would wear off after the first few crash replays.
 
I would prefer that too, but not at the expense of more cars in the game. Only so many resources will be attached to car modelling and if it means 150 new cars but a cool damage model, or 200 cars but the same damage model we have now, I'll take the 200 cars any day. Unless PD is willing to buck up the team to get that kind of thing done, which so far they haven't shown a willingness to do as far as I can tell, then I'd rather not trade off cars to drive for a better damage model. Others may feel differently but I've always felt that damage modelling is not that big a deal to me, more of a gimmick suited to making cool Youtube videos and the "wow" factor would wear off after the first few crash replays.

I'm guessing you're just referring to visual damage. Visual damage I agree, it's nice but I could take it or leave it.

Mechanical damage, on the other hand, has significant effects on gameplay and I think that is important. It's not necessarily linked to the quality of the model though.
 
I'm guessing you're just referring to visual damage. Visual damage I agree, it's nice but I could take it or leave it.

Mechanical damage, on the other hand, has significant effects on gameplay and I think that is important. It's not necessarily linked to the quality of the model though.
At lease they should make visual damage on aerodynamics and tire, this has significant effects on cars performance.
 
(awesome headlight piccie)
:drool: The headlight:crazy: I wonder how much detail it will be in GT7.
I do too, but I'm fascinated to see what kind of next-genness PD are putting into GT6. That headlight pic in your quote is insane. And then there's that article from GamesRadar:

‘Adaptive tessellation’ actually shouldn’t be possible in such a game on PS3. It may not be the sexiest phrase, but the tech has never been used on any PS3 game before. It’s so CPU-intensive, it’s supposed to be workable only on high-end PCs and next-gen. Each model is made up of millions of tiny geometric shapes fitting together to create a car. The 'adaptive' bit comes from the game’s programming automatically splitting each shape into smaller and smaller ones depending on viewing distance and position. This means that no matter what you’re doing, Scuderias and Skodas alike remain stonkingly - and accurately - sexy to the last millimetre.
This is a hardware feature on PS4. Most likely, it's paired down a bit from that monster system and an SPU assigned to manage it. But still, this is incredible use of technology that even Scaff and I thought would have to be cut back in almost every area including resolution to achieve. And yet, PD evidently hasn't cut back a thing. Instead, they tore open the entire GT5 Engine, something that was worked on for around four years, and made it even better - the resolution was increased for crying out loud! This is some kind of dark magic the team is doing, because most developers have accomplished all they can in a systems fourth and fifth years. This is an unreal testament to how powerful this "obsolete" seven year old hardware really is. And what quantum wizards PD are.

Next gen games (Fm5 and driveclub) can not be compared with current gen.
Well... and yet they are - at least one game is. And leave it to Polyphony Digital to make that game. :D
 
I would prefer that too, but not at the expense of more cars in the game. Only so many resources will be attached to car modelling and if it means 150 new cars but a cool damage model, or 200 cars but the same damage model we have now, I'll take the 200 cars any day. Unless PD is willing to buck up the team to get that kind of thing done, which so far they haven't shown a willingness to do as far as I can tell, then I'd rather not trade off cars to drive for a better damage model. Others may feel differently but I've always felt that damage modelling is not that big a deal to me, more of a gimmick suited to making cool Youtube videos and the "wow" factor would wear off after the first few crash replays.

Beyond the damage model, we could have better, more accurate suspension animations. There's also the possibility of having more accurate aero models once the car is damaged. Of course, there's no way that's going to be done in real time with the current and upcoming hardware situation. But, it's basically too late for PD to go that route. How strange would it be to have 1000 shells and 200 fully detailed cars?
 
There were no ill-feelings involved. I was just acting stupid. You wrote an improper sentence though and that actually made sense, but it wasn't grammatically correct. I do it too, and that's how I was acting stupid when I told you your grammar was incorrect. Honestly, I meant nothing serious by it and am not trying to make a fool out of you. It's the other way around if anything. I'm acting a fool :dopey:.
Ok, but I'd still like to see what was wrong.

I can't improve without seeing the mistake.
 
I've found something interesting

tilgl3p8.png


Looks like Route X With Mountains!

Maybe an early build or a GT6 version of Route X?
 
The GT5 version of Route X does have those. They're not mountains, they're small hills on the side of the track.
 
Saw a bunch of pics on the last page, saw the title... preposterous.

Like I said, it wasn't "preposterously" wrong and I said that I wasn't being serious, but here goes.

"I saw a bunch of pics (or pictures - I COULD go there) on the last page, then saw the title. Preposterous, it is (or 'It's preposterous)."

Again, I was being anal about it and didn't mean to make you look like a fool, especially since you thought my intention was to target you. Didn't have anything to do, saw the post...jackpot. :lol:
 
The entire underbelly thing is kind of a moot point, in a game where the car only rolls over once in a blue moon. I rather have those polys in the body.
Same goes to the tires, theres a lot of polys in a single rim just for the outside. You have to double the amount if you do the inside too. To even this out, you have the to reduce the details. So you end with a double sided rim that is only half as detailed on the side you see like 90% percent of the time. In my view that would be a bad decision.

There is also another little thing to a 3D model you like to ignore. Mesh quality.

Just compare this picture to the one posted by you.

Gran-Turismo-6_2013_05-15-13_021.jpg


Look how beautiful and even the mesh is, almost no triangles. This gives a very clean smooth and very few distortion. A mesh like this can also be upscaled easily with the press of a button, futureproofing it. Same method on the other mesh will most probably mess it up.

Actually the picture you posted is before it gets tessellated.With adaptive tessellation, it will look similar to the picture Samus posted... well change between what the 2 of you posted.
Gran-Turismo-6-63.jpg


You want to avoid triangles before smoothing the mesh not afterwards to my knowledge.But they aren't always avoidable as you can see here.
gran-turismo-6-20.jpg


I do however believe PD are unique because they kinda set the bar high with the amount of detail they put in cars in video games, but now more games are starting to reach that bar and some are setting it higher.
 
Actually the picture you posted is before it gets tessellated.With adaptive tessellation, it will look similar to the picture Samus posted...
Do you understand how adaptive tesellation works and how are modeled the cars in GT6?

Adaptive Tesellation => convert subdivision surfaces to polygons for rendering high-quality curved surfaces without visible polygon artifacts.

Subdivision surfaces are curved surfaces (or higher-order surfaces) described by a simple polygon control mesh and some subdivision rules. The subdivision rules define how to perform one subdivision step on the surface. Each subdivision step creates a new mesh with more geometry and a smoother appearance than the previous mesh.

Adaptive Subdivision = The subdivision adapts to the surface curvature and the viewpoint. The closer the surface is or the more curved it is, the more it gets subdivided.

Read this for more info:
http://http.developer.nvidia.com/GPUGems2/gpugems2_chapter07.html

gt6conceptmovie22013eu4k2x.gif
 
Actually the picture you posted is before it gets tessellated.With adaptive tessellation...

I think you confuse adaptive tessellation and subdivision surface.

This is one of my parts for my Wartburg 353 project. How I modeled it and how it looks after adding subdivision surface:
zfz3.jpg

ig5p.jpg

pcy5.jpg


As you can see this looks quite different to the render from Gran Turismo, as there are lots of polys in my model where you don't really need them (its not for a game). So this is most probably not the exact technique PD uses to make their master models. Or they clean op the meshes after using it. The finished master model on the other hand looks to be ideal for further subdivision.

If using Subdivision of any form, you should avoid triangles, sure they get converted into quads but they ruin the mesh flow.

The model posted by Samus would not be suited for subdivision and will probably turn into a mess because of the many triangles and the many undefined faces. Its simply made in a different style of modeling.
 
Last edited:
I didn't say it gets tessellated with adaptive tessellation.There's a period there for a reason :P.I don't know much about subdivision ,tessellation and adaptive tessellation, but i'm just going off of what happens when I tessellate a model.It looks like the picture from PD prior to tessellation and after I tessellate the faces it looks like the model Samus posted.The picture showing the adaptive tessellation also looks like the picture samus posted.
 
Like I said, it wasn't "preposterously" wrong and I said that I wasn't being serious, but here goes.

"I saw a bunch of pics (or pictures - I COULD go there) on the last page, then saw the title. Preposterous, it is (or 'It's preposterous)."

Again, I was being anal about it and didn't mean to make you look like a fool, especially since you thought my intention was to target you. Didn't have anything to do, saw the post...jackpot. :lol:

...Is this a real post?
 
Hmm ok, I try to make myself more clear.


249204_493940813970185_363642372_n.jpg


This, is a very poly effective mesh, its good for what it is. But its not verry suited for subdividing.

Gran-Turismo-6_2013_05-15-13_021.jpg


This mesh is not so poly effective but its better suited for subdivision.

gt6conceptmovie22013eu4k2x.gif


This "adaptive tessellation" is a real time process, it subdivides the faces close to the camera (upscaling) while at the same time reducing polys that are farther away from the camera (downscaling). Upscaling is a "clean" process, its produces quads and downscaling is a "dirty" one it produces triangles.

So basically the first mesh is designed to more or less stay like this, while the other one can be manipulated in real time, making parts of it better and other parts worse.
 
...Is this a real post?
Well, I AM a real boy.

Actually, all I tried to say was that certain people went awfully off topic.

It's alright though.
Thank you. It's like I said earlier. "I'm just being anal..." :lol: and so on and so on. I was in the wrong and was only being a bit of a prick (yet, I have to reiterate that it wasn't my intention to be a jackass to you.)
Time didn't stand still.
 
Back