Is this what the track surface actually looks like?

So, iRacing has their mythical "bump map" which makes their cars float but how does AC translate their details scans to its track surface?

Apparently (unless things have changed), the answer has been available for quite some time:

gBsXNxp.png


Perhaps I'm the only one who never got around to watching this video, but I nearly jumped out of my seat when I saw the image. iRacing has never shown anything like this, to my knowledge!

I think the detail is quite impressive, if this is the end result.

Using the photo mode app, I tried to zoom around the track to find any bumps, but it was really difficult to see (I would had no idea that the surface was so many polygons without the above picture). When you get up close to the white lines, though, some evidence of it is revealed:

CoR4sxm.png


One thing that is cool with iRacing, though, is the long videos and articles they produce that show some of what is going on behind the scenes (although, imo, indies like Shining Rock Software and Jonathan Blow have them beat by miles and miles when it comes to revealing technical stuff)...

...I wish Kunos would produce something similar for us nerds. :)
 
iRacing usually show off previews of their scan data when working on a new track, like this:-

iracing_spa3.jpg


But you don't get anything to physically roam around and check the finer detail.

That is what I mean! That is NOT the surface in the game. That is the raw scan data, which Kunos also shows (in every menu background in fact):

1yBMB8I.png


We still have NO idea how it actually looks/works in the game in iRacing since it's "invisible".

Kunos are the first ones who now have possibly given clues as to how that scan data is actually used in the simulation, with that plain polygon image.
 
That is what I mean! That is NOT the surface in the game. That is the raw scan data, which Kunos also shows (in every menu background in fact):

1yBMB8I.png


We still have NO idea how it actually looks/works in the game in iRacing since it's "invisible".

Kunos are the first ones who now have possibly given clues as to how that scan data is actually used in the simulation, with that plain polygon image.
I remember an article where they said the vertices are spaced ca 0.3 m apart in iRacing. Wouldn't help much to go finer, because you can't resolve smaller bumps with the current timestep/tickrate of current sims.
 
I remember an article where they said the vertices are spaced ca 0.3 m apart in iRacing. Wouldn't help much to go finer, because you can't resolve smaller bumps with the current timestep/tickrate of current sims.

Yes, again, that's the scan data. Now what do they actually use in the game? They never say beyond that it's some form of "bump map".

The tech tracks have the actual point cloud on the trackside objects, but not on the surface itself, so again, we don't know.

But let's focus on Assetto Corsa now! :)

What are the chances that a modder can recreate such a high polygon track surface, I wonder?
 
This might give a little more insight to how iRacing work in comparison to Kunos:-


This excerpt is taken from the iRacing site:-

In addition to its sight-picture visuals, iRacing.com replicates the precise physical features of each track’s racing surface. Our laser-scanning technology produces a mathematical ‘bump map’ of the track’s camber, cracks, undulations and patches – recording every millimeter of the surface. A series of “point clouds” capture the three-dimensional profile of the track surface and adjacent curbing.
 
Ok. Damn. I should not have mentioned iRacing. My fault.

I have more than one thread on this topic on the official iRacing forums, and again:

iRacing has not yet shown this bump map! They only talk about it and we only see floating cars.

We have no idea what kind of detail it actually has, though it might very well be more detailed than the AC polygons, but until they actually show it to us, let's keep this thread about AC. :p
 

That is NOT the physical tarmac you're driving on. That image was released long time ago and the tarmac was the first pass of the visual one, eventually optimized.

This is actually a mesh derived from the physical vertices you're driving on:
surface-imola.PNG


Edit: the 2 images from Mugello track also show only the visual track, the physical one is more detailed.
 
That is NOT the physical tarmac you're driving on. That image was released long time ago and the tarmac was the first pass of the visual one, eventually optimized.

This is actually a mesh derived from the physical vertices you're driving on:
surface-imola.PNG


Edit: the 2 images from Mugello track also show only the visual track, the physical one is more detailed.
Thank you for this! It's so amazing how far simulations have come. I really don't know how we could ever drive on a surface more real than this. I am actually surprised the poly-count is so dense. It really gives me a whole new appreciation for whats going on under the hood of this sim.
 
That is NOT the physical tarmac you're driving on. That image was released long time ago and the tarmac was the first pass of the visual one, eventually optimized.

This is actually a mesh derived from the physical vertices you're driving on:
surface-imola.PNG


Edit: the 2 images from Mugello track also show only the visual track, the physical one is more detailed.

Holy ****!

Wait, are you also now saying that in Assetto Corsa there is a physics based mesh to drive on that is separate from the visual polygons, similar as to iRacing?
 
If the graphics engine has to render that amount of polygons, we'd have 5fps...

It doesn't render it. Why render a static thing? It's precomputed and only loaded to memory when you see the loading track screen. This is the representation of the data structure holding the mesh, the physics is performed on that precomputed structure.
 
I know very little about graphics, so maybe I should better shutup ;).

But, I was replying to eobet who seems surprised that the physical mesh is separate from the visual polygons. To my understanding, the GPU has to render all 3d objects we see, correct? That also means the static objects like pitbuildings etc.

To optimize graphics, the visual road is less complex than the actual physical mesh. Ie; not every miniscule bump you can feel on the road is visual on your screen. As the laserscanned mesh is very dense and high poly, rendering it all would be to GPU heavy. iRacing does exactly the same AFAIK.

But as I'm not very knowledgeable on this subject, please correct me when I'm wrong.
 
I know very little about graphics, so maybe I should better shutup ;).

But, I was replying to eobet who seems surprised that the physical mesh is separate from the visual polygons. To my understanding, the GPU has to render all 3d objects we see, correct? That also means the static objects like pitbuildings etc.

To optimize graphics, the visual road is less complex than the actual physical mesh. Ie; not every miniscule bump you can feel on the road is visual on your screen. As the laserscanned mesh is very dense and high poly, rendering it all would be to GPU heavy. iRacing does exactly the same AFAIK.

But as I'm not very knowledgeable on this subject, please correct me when I'm wrong.

I was not ramming on you, just explaining :p. Yes, the GPU must render all 3D objects. In fact, it only renders what you see from your point of view, this means it does not render occulted objects. Moreover, it only renders what you see from an object, so if your facing a cube the GPU will only render the faces that you see.

To render an object it must be loaded to DDR memory and then passed to the GPU. This takes time and it's roughly proportional to the detail of the object. To optimize this process my guess is that there is two track versions; one with all the visual details but with a simpler mesh for stuff that we are less effective in detecting, such as small road bumps, so that the total number of vertexs to render is small; other with only the detailed information of the track surface and gravel traps etc. The latter never gets transferred to the GPU to be rendered, as it's only rendered the simplified track with more eye-candy and less physically accurate. The second version is for the physics engine to use so it can provide us the amazing physics that we feel on Assetto Corsa :p

Note that this is only a very simplified explanation of how GPU processing works, I can talk a bit more if you want :D
 

Latest News

How long have you been simracing

  • < 1 year

    Votes: 380 16.2%
  • < 2 years

    Votes: 258 11.0%
  • < 3 years

    Votes: 248 10.6%
  • < 4 years

    Votes: 181 7.7%
  • < 5 years

    Votes: 305 13.0%
  • < 10 years

    Votes: 262 11.2%
  • < 15 years

    Votes: 168 7.2%
  • < 20 years

    Votes: 130 5.6%
  • < 25 years

    Votes: 101 4.3%
  • Ok, I am a dinosaur

    Votes: 306 13.1%
Back
Top