Diffuse vs Specular

[AlphaAsh]

I'm finally making some progress in (re)learning modelling and texturing, but I'm scratching my head on a few things. One being, when to use a diffuse texture or a specular texture, or more specifically what's the difference in terms of HD footprint and CPU/RAM/GPU performance. Is there a performance benefit of diffuse over specular?

[NoMrBond]

A diffuse texture is the visible (RGB) texture of a part, the specular is (usually) just a layer on the image (the alpha layer in this case) which specifies which areas of the base (diffuse) texture are shiny.

So a 'plain' diffuse texture is RGB, and a diffuse+specular texture is RGB(A)

A part which is matte (i.e. not shiny or reflective at all) could just have a diffuse texture (and it's mesh renderer attribute for shader type in Unity would /KSP/Diffuse and the texture would be an RGB image)

A part which is shiny (i.e. a polished or finished surface) could just have a diffuse+specular texture (and it's mesh renderer attribute for shader type in Unity would /KSP/Specular and the texture would be an RGB(A) image)

Cost wise, specular textures are bigger because they're four layers instead of three for plain diffuse. Performance wise specular parts probably cost more to render but unless your video card is a potato I would assume this cost to be almost insignificant?

Edited June 19, 2014 by NoMrBond
Accidentally a letter

[AlphaAsh]

Cost wise, specular textures are bigger because they're four layers instead of three for plain diffuse. Performance wise specular parts probably cost more to render but unless your video card is a potato I would assume this cost to be almost insignificant?

I'm guessing it's dependent on just how many specular textures and light sources are on screen at any given time. Thanks NoMrBond, I'm feeling like I'm on track with texturing even more now.