Just wanted to know why cube maps often (but not always) comes in this sort of a cross texture?
Is there a standard to it or is is -"it has always been done like that son."?
It's easier to visualise it that way, the horizontal faces all in a line producing a panorama type result.
This also makes it easier to produce in the first place
When I was messing about with cubemaps I packed the faces to conserve texture-space, something like
but, as FMX suggests, it is all a matter of how you encode your cubemaps, and how you access them later. That panorama setup has one big advantage when you are sampling consecutive pixels, it can be tiled in the direction where you have all the samples together.
"There is nothing truly useless, it always serves as a bad example". Arthur A. Schmitt
Well but don't you like 'always' (except with those "weird" shaders that don't use 6 textures :p ) do more or less like this:
Load up texture T
Create 6 new ones (Up, Left, Front, Right, Back, Down)
Discard T
Right?
Which means the savings are just for disk space (or load-up time as compressed white usually doesn't take very much disk space).
I'm asking because I've done a cube mapping thingy and I'd like to go with the most used / best format or
something but I don't have a clue why I'm not just sticking those 6 textures one after another in a predefined way ?
Do what you think is best. Or you can have 6 textures, load them, and arrange them in memory how you think is best. Think, not obstant, that those arrangements are done with the sampling process in mind, and samples which are not together in the image will lead to seams, or incorrect samplings. I really don't know how the storage of those textures is done in disk, but i guess they go correlative, and are arranged when they are loaded.
"There is nothing truly useless, it always serves as a bad example". Arthur A. Schmitt
