On Sun, 2008-02-03 at 03:49 +1100, Graeme Gill wrote: > Leonard Evens wrote: > > > Can some one remind me of the trade offs among the -a options for a > > display. I presume the primary considerations are speed and whether or > > not a particular application can handle it. Also, I'm presuming the > > vgct tags in the resulting profiles would be essentially the same. Is > > that correct? > > Essentially the choice is between flavours of shaper/matrix > and a clut based profile. Shaper/matrix is generally smoother > since there is no table, but can less accurately model an > arbitrary device response. The flavours are to accommodate > various application limitations. Gamma is a simplified > form of shaper. Using a single curve is a simplification > of independent curves per channel. > Some application may be faster using a shaper/matrix. > > Only clut supports the full range of intents within > a profile. Because clut uses a table for the multi-dimension > transform, it can have noticeable stepping in its response, > depending on the clut resolution and nature of the > device response. Thanks for the response. I'm not sure what "stepping" mean, but I presume it means seeing distinct boundaries between adjacent colors. I see one such prominent boundary in the timage test image with the profile produced with -qm -as and to a lesser extent with the one using -qh -as. Would my shaper/matrix profile be smoother in general but still produce what I see in the test image? Is there some other way to produce a smoother test image response with a shaper/matrix profile? > > Graeme Gill.