Il nome non tragga in inganno: si tratta di un profilo ICC, non di un formato grafico. Precisamente di un profilo a tabella di tipo 'spac'.

Viene presentato anche con altri nomi (GettyOne, PhotoDiscRGBl, EBU-RGB dove EBU sta per European Broadcasting Union). Il nome interno del profilo è ColorSpace Conversion Profile EBU RGB. Si noti che LinoColor visualizza il nome esterno (Finder) del profilo, mentre Photoshop il nome interno.

Lo spazio su cui si basa è identico allo spazio di lavoro PAL/SECAM di Photoshop: gamma 2.2, punto bianco D65 e primaries EBU/ITU.

In Photoshop il profilo TIFF RGB si può usare per fare delle conversioni (Assegna profilo, Da Profilo a Profilo) e si può usare anche come spazio di lavoro RGB (in Photoshop 6, non in Photoshop 5).

Questo è un interessante anche se lungo commento di Joseph Holmes tratto dalla mailing list di ColorSync (6 luglio 2000):

Heidelberg (Linotype Hell at the time) wanted to offer some kind of standard RGB storage space for files being saved out from Linocolor during the scanning process. TIFF RGB is indeed a lookup table type profile, which has a gamut very roughly similar to a typical monitor, but which clearly does not coincide to any monitor. It is no larger than a typical monitor space.

Being a LUT type space, it has at least the potential to map colors that are outside its modest gamut more gently and intelligently than a matrix type monitor profile does. I would argue that it is better to use a profile for storing your RGB information that is large enough that none of the colors are out-of-gamut in the first place, however.

I happen to have done some tests yesterday with a circular color gradient, sending a slice of the full Adobe RGB 1998 gamut into each of TIFF RGB and Apple MultipleScan 17 D50 (which is internally the standard Trinitron profile at gamma 1.8 and D50 i.e. it uses the published Sony chromaticity values for R, G and B). As one would expect, there were differences in the artifacts produced by the two mappings, but I was a little disappointed by the mapping into TIFF RGB. The smoothness of the reds for example was much better with TIFF RGB (meaning that the transition from out-of-gamut reds to in-gamut reds was gradual), but the smoothness on the other side of the circle, in the neighborhood of Cyan, was not as smooth at the Apple result, with its simple matrix (or at least was not very good).

Matrix profiles should never, in theory be used as a destination for converting colors that are outside of their gamuts, though one might never notice a tiny bit of damage done to tiny bits of color in an image, if the out-of-gamut colors are insignificant. One can clearly see why this shouldn't be done when looking at the result with the reds in the above case going into the Apple MultiScan 17 profile. The out of gamut reds have lost all visible differentiation. Again the object of building TIFF RGB was to prevent this and hue shift problems, and it is partially successful, but I would argue that this profile is now essentially out of date, given the rise of RGB working spaces that we have seen on the list, stimulated primarily by support for them in PS 5. In fact, if I were Heidelberg, I would add a new choice or two for saving into RGB, right inside Linocolor and Newcolor, one of which would be my own space and the other Adobe RGB 1998.

Also, TIFF RGB uses a lot of data points and 16-bit precision in the tables, and it has always struck me as a good effort at making a useful RGB space in certain respects, but I'm very glad I never used it to store my archive, the way some archives have been stored. A lot of color information was arbitrarily altered, but only in those cases where the images contained significant, very saturated colors (out of TIFF RGB gamut colors).

The matter of it being labelled internally as a SPAC (or color space) type profile is partly just a matter of choice for the makers of it. What I mean is that the matrix type working spaces are functionally simply monitor profiles with different numbers put into them, though they do not describe the color space (or gamut) of any actual monitor, therefore in theory they should be labelled SPAC profiles too, but it doesn't hurt that they are not, and often applications impose limits on which types of profiles can be used for various tasks, so that has to be taken consideration and makes monitor type profiles more versatile.

Remember that out of gamut colors get clamped in a mathematically brutal way entering matrix type spaces, of necessity, given current technology. But this does not mean that the way colors will be treated going into LUT type RGB profiles will be somehow ultimately gentle either. Inevitably the relationships between original image colors will be altered, and most likely imperfectly. The safest bet, to protect the integrity of the image, is to leave the colors alone as much as possible and avoid the inexorable buildup of what I call color chaos, which is introduced by many photographic and imaging processes, especially the chemical processes upon which photography has necessarily relied from its beginning until the transition which is occurring now.