NASCAR Tracks and Roster...?

  • Thread starter Thread starter Saiyavenger
  • 49 comments
  • 4,488 views
I really doubt GT will have nascar drivers. GT never had anything like that. I believe it will be limited to the cars and races only. But then again I can be wrong. GT never had anything like nascar before.
 
red70mach1
From looking at the list. Only the four Chevy Imps of Hendrick Motorsports' (Hardcore Ford Guy but also drive Toyotas, so anything but Chevy) Sprint Cup Team is listed. I'd like to see Gibbs Racing and even Red Bull Racing's Toyota Camrys, Roush Fenway's Fusions, Penske Racing LLC's Chargers also added.
Click to expand...

There is the #11(Denny Hamlin,Toyota) and #55(Michael "insert sponsor plug here" Waltrip, Toyota).
 
It really doesn't. You are thinking about the wrong thing. If we were talking about something like duplicate lines of code being eliminated because they could instead refer to the one instance where it pops up, it would apply.
Having multiple instances of the car model doesn't allow that, because the game still needs to render all of those assets individually (in essence, as far as the game engine works, the cars are not the same model), so there are no performance benefits to loading just one model and pasting it around unless the model itself is lower detail than a different car would be (for example, if a NASCAR is less detailed than a Corvette by itself, two NASCARS would be less taxing than a NASCAR and a Corvette only because the total poly count is less).

Just about the only difference it would make resource-wise is that it wouldn't have to load the models individually at the start of the race, and PD could implement a way of loading the single model and duplicating it to the number/textures needed instead of loading the model multiple times so race load times would be faster; but even then I doubt the benefit would be that great.
Click to expand...

This sounds wrong to me. In my experience creating 16 instances of a model will run faster than 16 unique models of equal polycounts. I`m not a programmer, but I do love making models in 3DSMax and just from using that it definitely feels like this is the case. From what I understand freeing up the additional RAM does speed up your frame rate, quite like what chepe371 described.

Polygon count isn`t the overriding bottleneck in the system, there are other variables limiting what is happening on screen. Proof of this is making a bolt out of a chamfer cylinder. If you leave it a primitive it will render faster than if it is converted to a polygon object with the back faces deleted. Thus it has more polygons and renders faster.
 
Last edited:
Unfortunately, no matter how similar two shapes are, the fact is they are going to be in different positions relative to the viewpoint. That means the renderer is going to have to position vertices in different places, cull different backfaces, apply different shaders, etc etc. Here's a quick test you can do in 3ds max. Make an object, a really complex one. Render it. Now add another to your scene, and render it again. See what happens to you render time.

Toronado's explanation of where you could gain performance with identical cars looks pretty accurate to me. Other than maybe a gain in loading times, and the possibility of some really funky texture sharing optimization to save some VRAM, (unlikely), it won't save you much. Even if the polygon count were lowered, there's no guarantee that the PS3 could handle the extra physics calculations, audio mixing, etc.
 
If Polyphony Digital got some time, they should add older NASCAR cars from '70 to '90 and some legend driver like Richard Petty and Dale Sr.
 
My argument is that rendering x instances is faster than x different shapes.

An instance in Max is a copy where if you alter the master that alteration will be carried through to all the other instances or copies.

Aspect8 I understand what your saying about the differing angles for each object, but that`s only one set of variables in a very large porridge. Your test doesn`t apply to what I`m arguing. Does a cube render faster than 2 cubes, yes. More importantly does 2 instanced cubes render faster than a cube and a rectangular prism, yes again. The latter being the more profound for Polyphony Digital and there 16 identical cars on a track.

Why this happens I can`t explain in any detail, like I said I`m not a programmer or engineer. I just know from tutorials and practical experience that instancing is quicker.

So the question is, how much boost to your framerate do you get by having 16 identical car models on track? No idea, maybe room for one more? haha, yeah I`m pretty sure it won`t be 40 :o) Crossing my fingers.
 
Polys are polys, regardless of whether they're on identical shapes. The renderer still has to render them. 3ds max instancing is a tool to make changing properties and animating easier. I am quite sure it doesn't affect render times.

It's a sad sad situation, but hey. They may find some way to bump up the numbers. Even if they don't, 16 is better than 8 right?
 
^ There are optimizations that can free you some memory though.

For example, when you have 16 different cars in the track you need 16 different sound files in your memory to be able to play them. With 16 copies, you need only 1 sound file. You will be playing 16 different parts of the same file, but you will be freeing 15 sound files worth of memory space. The same goes for car models. Instead of 16 different models it will be replicating 1 model what allows for more memory space.

Of course, as you said, to the renderer that doesn't matter. I don't know exactly how renderers work (hopefully in a few years I will =D) so I can't tell if there are some other kinds of optimizations you can pull off but essentially, even though it's 16 copies, it's still 16 things to render.

The only way I can tell that it would be possible to add more cars is if their shape is significantly simpler as Toronado noted (and I think they are). If somehow PD gets to do a stock car with 100k polygons, they can roughly double the number of cars as far as the GPU is concerned. Sure this means more physics calculations but trust me, car handling physics are really easy for these kinds of processors to deal with. That's not a problem at all. Audio mixing is also really easy to deal with so that's also not a problem at all. AI calculations should be fairly easy too unless PD is going for something crazy, what I think is unlikely.

There are a lot of other things to take in consideration like the damage engine and how intensive it is in the processors but I believe that if PD at least tries to, they can get about 20 cars in the track due to their simpler shapes more than anything else.
 
red70mach1
From looking at the list. Only the four Chevy Imps of Hendrick Motorsports' (Hardcore Ford Guy but also drive Toyotas, so anything but Chevy) Sprint Cup Team is listed. I'd like to see Gibbs Racing and even Red Bull Racing's Toyota Camrys, Roush Fenway's Fusions, Penske Racing LLC's Chargers also added.
Click to expand...

+1 👍
 
AtD94-[GT];3604072
If Polyphony Digital got some time, they should add older NASCAR cars from '70 to '90 and some legend driver like Richard Petty and Dale Sr.
Click to expand...
50's and 60's as well. Imagine taking a '57 Black Widow for a spin at Daytona...

Black%20Widow%202.jpg


... or a '65 Galaxie

0603_07z+Ford_Galaxy+Top_Side_Track.jpg
 
A way to fix this lack of drivers is to simply let us skin our own cars.

And I'm not just saying that because I'm a wannabe NASCAR designer.

Also, I'm still curious about possibly running Sprint Cup cars on tracks like Laguna Seca and Tokyo...
 
The interiors of NASCAR stock cars are also rather complex. There is a lot of detail that can be put into the cockpit. The damage even plays a part. My point being you aren't rendering any less with a corvette as opposed to a stock car due to the added damage.

Even if we only have 16, we will still have amazing racing. Plus the only rise of a starting grid I would want is if we could fit 100+ cars at the ring.
 

Latest Posts

Back