Car. Material properties

[Surzh]

Hi guys,

I want to play with material properties (first of all for car model).

Where I can find that?

Thanks,
Ivan Surzhenko

[NaN]

Material is the misc1 texture. Red and green channel are the interesting ones.

Code:

float gloss = tu1_2D_val.r;
float metallic = tu1_2D_val.g;

[portets]

am i right in saying that, after the deferred rendering refactor, the green and red channels never got separated?

making a misc-1 blue or green does nothing. red causes gloss and metallic.

[NaN]

Did a quick check myself. Looks like a bug to me. Maybe Joe can comment on this.

[fudje]

Is there a normal map for cars?

[NaN]

The UV mapping of the most cars is unusable for normal and ambient map creation. We would have to redo them. Just have a look at the diffuse textures to get an idea.

[portets]

i created a normal map for the xs to make seams between body panels.

it turned out pretty bad, a bunch of shiny squares followed the lines.

doesn't vdrift compress textures in-game? because the normal maps have jpeg-like compression artifacts, but shiny.

and it's not just me; the normal maps look good when loaded into different programs.

[fudje]

I reckon that's more likely to be stretching. Although the lowest quality filter in VDrift seems to be Trilinear mip-maps, which should more or less hide such things.

[NaN]

DXT compression artifacts I guess. All textures > 512 are compressed. We should either disable it for normal maps or implement normal map compression.
http://developer.nvidia.com/object/real-...ssion.html

[portets]

yeah, i think disabling normal map compression sounds good. then we can figure out what to do later.

i've partially been stalling on making new normal maps because they don't look right. like my rouen tarmac maps; they look pretty bad up close.

[fudje]

Haha DXTn. I never noticed. Without compression (my driver doesn't do S3TC off the bat) I can quite happily fit all the textures for Ring 2007 and 4 cars in 256M of memory for IGP....

[joevenzon]

am i right in saying that, after the deferred rendering refactor, the green and red channels never got separated?

making a misc-1 blue or green does nothing. red causes gloss and metallic.

That's true, the deferred rendering rewrite changed the lighting model being used. The lighting model takes a specular color and specular exponent (shininess), so I think the best change here is to go with a specular map where the specular color is in RGB and the exponent is in A. However, this requires adjusting all of the existing miscmap textures....