Understanding RGB888

[scarleton]

I know there is one RGB format that pads each scan line to 4 bits with zeros. I assume RGB888 does not do this, correct?

[Talei]

After RGB there is bit representation for each channel. 888 means 3x8bit = 24bit, 444 -> 3x4 = 12bit. http://en.wikipedia.org/wiki/RGB_color_model
If You see something like RGB 4:4:4 that means subsampling for one of channels in YCbCr: http://en.wikipedia.org/wiki/Chroma_subsampling

[wysota]

I know there is one RGB format that pads each scan line to 4 bits with zeros. I assume RGB888 does not do this, correct?

(8+8+8) mod 4 = 0 thus there can be no padding here.

[scarleton]

(8+8+8) mod 4 = 0 thus there can be no padding here.

My bad, I ment bytes, not bits. I think it is Windows native image format that does this. I am guessing it is because 32-bit OS's work the quickest in 4 byte chuncks, though it is a pain to save out Further more I take it that this is not how RGB888 works, correct?

[wysota]

My bad, I ment bytes, not bits. I think it is Windows native image format that does this.

Native Windows format is BMP which is vertically mirrored BGR 8b or 24b without any padding, as far as I remember.

I am guessing it is because 32-bit OS's work the quickest in 4 byte chuncks, though it is a pain to save out

Honestly I never heard of such padding, it would be a pure waste of memory space. Padding usually occurs only at the end of some data, not in the middle of it. Of course I don't deny such formats may exist, I just never had any experience with those Image formats tend to have multiple of word length as their bit per gun anyway. Of course the word length might be different for different architectures.