Texturing with components vs. Texturing using full textures, your preference and why.

General Forums Blender and CG Discussions
1

Now before we begin, I am starting this thread with the awareness that people will take different approaches to texturing a scene, in this case, I would like the see the community gather together information on the pros and cons to texturing using either fully-detailed textures or individual components (in my case usually black & white masks) that together make up a texture.

I’ll start with the fact that I’ve more or less used both types when texturing objects, so I can start this thread with a list of observations I’ve made over the years.

   <i><b>Pros of fully-detailed textures::</b></i>
    • Ease of use, just slap a single copy of the different map types on the model and you’re good to go.
    • Memory savings, less textures are needed which means less memory (especially if you’re rendering with the GPU which for myself I am not).
    • Allows for more possibilities in raw color detail

Cons::

    • Requires a larger library of textures due to needing new images for variations
    • Requires the creation of a new texture file if you need to make a unique material
    • Less possibilities in editing and customizing the texture through the use of nodes or other material tools in the 3D program.

<-------------------------------------->

       <i><b>Pros of texture components::</b></i>
    • Mix and match, modern workflows like the Cycles node system can really make this shine, for example, if you need to add pockmarks or weathering to a texture, you just use the appropriate component from your library so you don’t need to make an entirely new texture file.
    • Saves time due to not having to create a library of similar textures with variations that you can otherwise add with texture components.
    • The component masks can be used to reduce the visibility of tiling in various scenes

Cons::

    • Doesn’t work for all models, a creature model for instance might need precise detailing
    • Not optimal when using high resolution images in a GPU render due to memory constraints and the CUDA texture limit.
    • More setup time needed due to the fact that you’ve offloaded more of the work from the paint program to the 3D app.

My opinion to start us off, I think there’s a place for both of these techniques and one shouldn’t necessarily completely dominate over the other for all cases, as they both have cases where one is preferred over the other.

Your thoughts?

2

Ah, it’s been a while since there’s been an interesting technical topic! I’m looking forward to reading about other people’s texturing techniques! Nice one Ace Dragon. :slight_smile:

I haven’t seen that naming convention before so I hope I understood correctly.

I use lots of component textures while working, but before I actually use the asset I simplify the shader by baking the model onto itself with a unified UV map, getting the standard unique diffuse/spec/normal/displace set that way.
Then I put the spec in the diffuse’s alpha channel and the displacement map or alpha map (if I make one) in the normal map’s alpha - that way all the textures for the asset fit in two RGBA image files.

It all takes a little time to do and your points about needing to make new images for uniqueness are correct but it’s not so bad since you can tweak the component textures of the original setup and just do a new bake.

This is one of the reasons I’m looking forward to Cycles baking. BI nodes tend to make a mess with materials. :slight_smile:

This is of course for assets with medium to complex geometry. For large simple surfaces like grounds/walls I tend to just use components and tiling textures with several uv maps because otherwise I’d need really really large textures.