[argyllcms] Re: colprof problem
- From: Graeme Gill <graeme@xxxxxxxxxxxxx>
- To: argyllcms@xxxxxxxxxxxxx
- Date: Tue, 23 Jun 2015 11:03:21 +1000
Robert Hansen wrote:
Have you tried qh instead, I made a profile and the peak delta was 2.1 with
an average .8 with the high quality setting using 1800 patches
The self fit error metric doesn't give much indication of how accurate
the profile is, just how consistent the data points are to the interpolated
surface. Assuming your data points don't conflict, you can get a zero
delta E self fit if you increase the grid resolution and relax the smoothing
to zero - and this is known as "over fitting", and probably makes your profile
less accurate. So take the self fit error values with a grain of salt...
Graeme Gill.
Other related posts:
