Graeme Gill wrote: > Alexander wrote: > >> "The monitor profile * appears to be defective." When I say use anyway, it >> seems to be >> OK, but something must be a bit off if Photoshop is balking at it. >> > > OK, the response from Adobe is: > > ACE contains some monitor profile validation code to attempt to filter out >> bad profiles. >> One test it performs is to convert Lab white (100,0,0) -> RGB to check >> that it’s at >> least (f7,f7,f7). It also converts RGB (ff,ff,ff) -> Lab to test that it’s >> reasonably >> close to white. >> >> With this particular profile, GDM-F520_2011-01-14.icm, ACE got RGB >> (f1,f6,ef), which >> caused it to report that it was a “bad” monitor profile to Photoshop. >> > > So it is a result of a not very accurate B2A white point mapping. This is > likely > a result of the D50 white not falling exactly on a grid point, given that > D50 XYZ is 0.9642, 1.0000, 0.8249, while the PCS range is to 1.999969, so > that unlike Lab PCS, the white point falls somewhere in the middle... > > I'll see if I can do something about this for the next release. > > Graeme Gill. > > So if I'm understanding correctly, use of this profile would result in a reduced luminance white point and increased banding from clipped values (~94% vs >=97% usable)? Any manual tweaking I can do in the meantime with my ti3 measurements or colprof in order to forcibly generate a XYZ profile ACE finds acceptable?