Kazunori Yamauchi Pre-E3 2017 Interview:

Damage model and modelling the parts beneath the surface is something that can be done to existing cars, again no need to remodel anything. I've imported FH3 and F4 models to Blender before, internal structures of the cars are separate pieces (even though you don't notice it in game) and not part of the car's exterior. If PD wants to add internal structures of the car they can do it without remodelling what they currently have. Same thing with damage.

Perhaps. Perhaps not. It depends how it's implemented. If they go to a soft body style system then I rather think it would require some fairly substantial changes. If they're calling whatever they've done to go from GT6 to GTS models "remodelling", I think that would qualify.

And let's be honest, If PD started modelling engines and trunks of cars, the fanbase (me included) would complain about wasting precious modelling time on useless things instead of more cars or tracks.

Perhaps. Perhaps not. People seem to quite enjoy Forzavista. I would have said that a feature like that would have suited the Gran Turismo "car museum" style quite well. But I guess that's not Gran Turismo any more.

The point is, it doesn't matter if the assets are 2 or 5 or 10 years old if they look good.

If you're trying to be an also ran, sure. If you're trying to be a graphical showcase for a specific set of hardware and have the best looking game on any platform, it really doesn't help if you're not working with the best available assets.
 
Perhaps. Perhaps not. It depends how it's implemented. If they go to a soft body style system then I rather think it would require some fairly substantial changes. If they're calling whatever they've done to go from GT6 to GTS models "remodelling", I think that would qualify.

Soft-body damage in Gran Turismo. Yeah, it would require redoing everything but I no chance you will see that anytime soon. For a number of reasons. We don't even know if it would be allowed by manufacturers. Even if such a system would become feasible in the next ten years (and it won't) I doubt PD would use the tech, as damage is hardly a focus in GT.

And BTW, regardless of what PD says about "all cars remodelled from scratch for GTS", it's pretty obvious that the "super-premiums" from GT6 were simply ported over. Which is a logical thing to do, but it pretty much proves PD's statement false.

Perhaps. Perhaps not. People seem to quite enjoy Forzavista. I would have said that a feature like that would have suited the Gran Turismo "car museum" style quite well. But I guess that's not Gran Turismo any more.

Forzavista is alright, and T10 can afford it because they outsource. For PD it would be a waste of resources. You could spend a 100K polys modelling an air filter housing but if it's almost never seen what's the point. I don't think Forza even uses the internal parts during game play (though I could be wrong). But as I said, if PD decides to add these parts they can do it without having to redo the whole car.

If you're trying to be an also ran, sure. If you're trying to be a graphical showcase for a specific set of hardware and have the best looking game on any platform, it really doesn't help if you're not working with the best available assets.

Yeah, but that's assuming that a competing developer will somehow create digital cars that put PDs car models to shame. I doubt that will happen, because we've pretty much reached the ceiling I think. In terms of geometry, at least. It's why I said, GT5 photo travel cars still look great even now, and Forza is still reusing cars from like F4 and they look fine despite being from an older console. Even the "super premiums" originated on the PS3, in GT6. And they look better far better now, even though though it's the same model, because of lighting, materials etc..

Even in the hypothetical PS5 GT game 10 years from now I don't see how do you make these look "not good enough" or dated? What would be the point of redoing them? The shapes are accurate, curves are smooth, the details are there like on the real car. How much closer to the real thing can you really get? Where would you spend more polygons?
43823ba4gw1f6zgtx3swtj21hc0u04ar.jpg
Gran-Turismo-Sport-Porsche-GT3-headlight-close-up.jpeg
Gran-Turismo-Sport-BMW-interior.jpeg
 
And BTW, regardless of what PD says about "all cars remodelled from scratch for GTS", it's pretty obvious that the "super-premiums" from GT6 were simply ported over. Which is a logical thing to do, but it pretty much proves PD's statement false.

Oh, I agree. But if they're "remodelling" for GTS, then they can and should be "remodelling" for every new generation.

Forzavista is alright, and T10 can afford it because they outsource. For PD it would be a waste of resources.

You make it sound like Polyphony is underfunded compared to T10. You might remember that at the time GT5 was one of the most expensive video games ever made, until GTA V crushed everyone and everything. Polyphony has only gotten bigger since then, and I doubt it's gotten any cheaper to make video games.

Even in the hypothetical PS5 GT game 10 years from now I don't see how do you make these look "not good enough" or dated? What would be the point of redoing them? The shapes are accurate, curves are smooth, the details are there like on the real car. How much closer to the real thing can you really get? Where would you spend more polygons?

OK, even though I was clear that there's more to a model than simply polygons, let's do this as a thought exercise. What more could you do with polygons that would improve the model? Assuming that we're not adding a damage model and no one will ever want to look at any parts of the car that aren't already modeled.

Would it be worthwhile to go back over and add imperfections into the model? Many of these are taken from CAD models as a base, but a real car isn't so perfect. Is it worth adding variation into the panel structure and how they're positioned relative to each other? It's not uncommon, particularly for GT, for the cars to be accused of looking "too perfect".

Is it worth having modeled paint or sticker layers that are separate from each other? A decal is not physically on the same layer as the paint, although currently the engines treat them as though they are. Is that valuable to have separate physical layers of base coat, metallic, and clear? Depending on how your lighting engine works it could be.

I'd say there's almost certainly materials that could benefit from an unlimited poly budget. Carpet. Fabrics of all types. Rough metal or plastic. The way that these things are dealt with currently are ingenious hacks, and with increasing hardware power it may be preferable to simply have them modeled as the textures materials that they are.

Games look great now, but they don't look real. Still shots are getting there, but there's still a gulf between something like a Scape and real gameplay. It's only in the last handful of years that CGI has really gotten good enough to seamlessly integrate cars into things like car commercials, and that's with essentially unlimited hardware power. You don't think that there are parts of the modelling that add to that? Up to you. Bit of a lack of imagination on your part though, to think that we're somehow at the pinnacle of what can be done to make something look real.
 
You make it sound like Polyphony is underfunded compared to T10. You might remember that at the time GT5 was one of the most expensive video games ever made, until GTA V crushed everyone and everything. Polyphony has only gotten bigger since then, and I doubt it's gotten any cheaper to make video games.

I meant more like T10 has no problem outsourcing assets and PD does, Obviously PD should either outsource or hire more staff to remain competitive.

OK, even though I was clear that there's more to a model than simply polygons

Well true, but not really. Materials, lighting, textures etc all improve the look of a model but are not directly related to geometry, or polygons (whatever your preferred term is). And I'm strictly talking polys because i was responding to the topic of having to remodel the cars again for next gen.

let's do this as a thought exercise. What more could you do with polygons that would improve the model? Assuming that we're not adding a damage model and no one will ever want to look at any parts of the car that aren't already modeled.

Would it be worthwhile to go back over and add imperfections into the model? Many of these are taken from CAD models as a base, but a real car isn't so perfect. Is it worth adding variation into the panel structure and how they're positioned relative to each other? It's not uncommon, particularly for GT, for the cars to be accused of looking "too perfect".

Funny you mention that, as i think PD paid more attention to that than other devs. In GT6 it was quite noticeable how the old muscle cars, especially the GT350 had some slightly misaligned panels and such compared to modern cars. Can't say I've noticed that in other games really. And while I agree that tiny imperfections improve overall realism, all of the things you mentioned can be added to current models if necessary. If PD wants to misalign the door on a Porsche they can do it, it's not necessary to remodel the car.

Is it worth having modeled paint or sticker layers that are separate from each other? A decal is not physically on the same layer as the paint, although currently the engines treat them as though they are. Is that valuable to have separate physical layers of base coat, metallic, and clear? Depending on how your lighting engine works it could be.

But that's the purpose of material shaders and not the geometry, isn't it? Many recent games' shaders simulate base coat, clear coat etc. For extremely thin things like a coat of paint it's far more effective to use layered materials. Materials are a separate thing from the geometry. And if you really want for decals to be made with geometry it can be done. In fact FH3 models actually have decals as geometry and not textures. It's not necessary to redo the car for that.

I'd say there's almost certainly materials that could benefit from an unlimited poly budget. Carpet. Fabrics of all types. Rough metal or plastic. The way that these things are dealt with currently are ingenious hacks, and with increasing hardware power it may be preferable to simply have them modeled as the textures materials that they are.

Even if you are totally nuts (and you have the hardware power) and want to have every little bump in the plastic on a dashboard, or every strand of fabric in a carpet to be actually modeled and not textured, these things are usually not done manually but with an algorithm that would spread these shapes across a surface. Again, it's not necessary to remodel the car and PD can easily add that when there is enough hardware power.

Games look great now, but they don't look real. Still shots are getting there, but there's still a gulf between something like a Scape and real gameplay. It's only in the last handful of years that CGI has really gotten good enough to seamlessly integrate cars into things like car commercials, and that's with essentially unlimited hardware power. You don't think that there are parts of the modelling that add to that? Up to you. Bit of a lack of imagination on your part though, to think that we're somehow at the pinnacle of what can be done to make something look real.

Whoa now, I never said that. Scroll up and notice I simply said that car models are so detailed now, you won't achieve much by pumping more polygons into them. With the whole point being, that I don't see a reason why PD to remodel the cars anytime soon. PS1 to PS2 was a huge jump, PS2 to PS3 too, but PS3 to PS4 not nearly as much. The next evolution in realism will be in lighting (like ray tracing), materials, textures, animations and such. But PD is PD, so who knows what they'll do.
 
Many of these are taken from CAD models as a base, but a real car isn't so perfect.

But are the cars in GTS as detailed as those found in Project cars? When we talk about 3d cad models you should also talk about how intricate/detailed the models are when transferred from the cad/3d design program to the game. I dont talk about the amount of polygons of the "shell" but what is underneath the chassie/car body itself if the car itself is even structurally and mechanically resembling the real car. How can a hollow body with animated wheels in a game even be called a sim? :P I dont think I have seen any game up to date to be that 3d cad model:ish as those found in Project cars. Would be so cool if GTS really did have as advance models as in PC and not only a realistic looking body shell.
 
Last edited:
How can a hollow body with animated wheels in a game even be called a sim? :P

Like that :

Assetto Corsa is a sim.

Just because a game is missing something you consider important (or something other game is doing) doesn't automatically disqualify it as a sim.
 
Like that :

Assetto Corsa is a sim.

Just because a game is missing something you consider important (or something other game is doing) doesn't automatically disqualify it as a sim.

Well it is a "sim" but yet it does not even try to simulate the car properly that I drive. the car should simulate all those forces in motion, at least the bigger ones, like how the engine moves, freewheel, drive shaft, body roll/twisting, the centre of gravity of the car and of course the tires as closly as they possible can as it is the medium that propels the car forward by interacting with the ground. And we should not be forgetting about the importance of air resistance :P

Maybe I am bit over the top because of what it was possible to do with simple 3d cad software in 2000s and because I would see in a real sim and a mere console would probably not stand a chance to even calculate al those factors but seeing how much money a simracer is spending on his rig yet often use a crappy pc to run it makes me wonder.. But yeah if there was a proper sim they would gladly upgrade to a machine capable of running such an demanding title because they would know that it would be as close to driving a real car as possible.
 
Back