Image vs Float Image

[rosinho12]

well, i'm here with a little question about these two parameters
so, i was using, and, question myself: "whats the diference betwen these two?"
so, whats the diference betwen then, and when should i use each of them?
thanks people

[Refracty]

float image is grayscale only, so use it if you don't need color information (bump maps,...)

[matej]

If you open an image file as image it will load as RGB(3 bytes), if you open it as float image it will load as grayscale (1 byte) even if the image file is stored in RGB, if you open it as alpha image it will load only the alpha channel (if the image has one) as grayscale (1 byte).

Yes, the term float image is kinda misleading / confusing (since all the data is stored as floats) and unnecessarily technical. It should be called simple "grayscale image" instead.

[rosinho12]

hmm, now i understando. quite simple hehe
but, whats the advantage of having it in differents "tipes"
most of the programs use the two as one
(hope you understand me :X)

[matej]

The benefit is saving VRAM. Why load an image in RGB(A) if you need it only to map an intensity.

RAM is something we don't have in truckloads around here

[rosinho12]

wow, now it make sense to me...so images like bump, specularity, etc... is a way better to open in float image...
hmm
thanks guys
helped me alot
specialy with the bad feeling of "what the hell is that?"

[Proupin]

There sould be only one image node and in the node's options be able to choose what you want to use, rgb, grayscale, alpha or whatever... right?

[matej]

@Proupin, yes that would be a step forward, just one thing; in the graph editor it's important to see at glance which type of data (RGB / grayscale / alpha grayscale) a node produces - ie. the icon of the node should change accordingly when you change the "filter" of the said node.

This improvement + the ability to duplicate nodes and we won't need anymore to browse a (the same) file a thousand times in one project