KatsBits Community

Computer Workload

0 Members and 1 Guest are viewing this topic.

Offline ACDmvmkr

  • Newbie
    • Posts: 47
With all other things being equal; Which is easier for processing and rendering; an object with a surface appearence applied using a texture mapped image, or one where all the design is applied by assigning material color patterns formed by vertex groups (selecting the various faces and applying material colors from the color picker)?

Example -

A multicolored Soda Can with many colors, but a relatively simple pattern of right angles, etc....

- 1) If you wrap a photographic image around a cylinder, or

- 2) if you break the can up into vertex groups and color those?

- 3) What happens if you take the top example, but with the mesh broken up into the same number of faces as the bottom one?

Offline kat

  • Administrator
  • Hero Member
  • *
    • Posts: 2721
    • KatsBits
That's actually difficult to answer because the Material and general geometry processing and rendering aspects of an engine might each have different 'bias'. For example, raw polygons are by far the cheapest aspect of rendering, BUT, it depends on the organisation of the geometrical data, a messy mesh takes longer to process and render than a clean one - by that we mean a mesh where triangle data is organised and flows logically versus a similar object where triangles are flipped any-which-way.

With respect to your question though there's a problem because all things should not be equal. What's meant by that is that the mesh used for either/or should be tailored specifically to the requirements of decoration. For example. If you had a cube with a simple two-toned texture image assigned - one half red, the other blue - the cube would only need to be composed of six sides (6 faces, 8 vertices and 12 triangles). Doing the same thing using vertex colours or material assigned patterns means adding more geometry in order to replicate the same effect - using material colours you could just divide the mesh in two and assign a red material to the top half, and a blue one to the bottom; using vertex colours might be a bit more complicated because you have to isolate each vertex set attributed to a particular face and the only way to do that is either add additional loops, or separate the geometry (using "Split", "Y").

The simple cube treated for texture, material and vertex colour assignments for example means that whilst the latter is quicker to process because it's just mesh data, vertex colours being a component of that directly, this doesn't mean overall it's more efficient or less resource hungry because to get the defined colours the mesh has to be split into segments that constantly break the math calculations the engine is using (cf. poly-striping linked to above) to render the structure of the mesh - splits are expensive.

Textured = mesh + Material + texture, + simple structure
Materials = mesh + Materials, + additional structure
Vertex Colour = mesh + vertex, + increase on additional structure

Hope all that helps ;D