[argyllcms] Re: scanin glitch -and- very noticeable color differences between profiled scanners
- From: Ben Goren <ben@xxxxxxxxxxxxxxxx>
- To: argyllcms@xxxxxxxxxxxxx
- Date: Thu, 14 Oct 2021 09:48:18 -0700
On Oct 13, 2021, at 9:02 PM, Graeme Gill <graeme@xxxxxxxxxxxxx> wrote:
Scanin has improved a lot over the years,
and seems to be quite robust whenever I use it.
Some ago when I was using it, it was already insanely robust. I was throwing
under-developed linear not-white-balanced skewed un-cropped photos of a
ColorChecker Passport at it and, as often as not, it recognized the chart just
fine — even when it was visually not all that obvious that the photo was of a
chart. It wouldn’t work every time, of course, but often enough that I’d always
try scanin first and only when it failed worry about fixing the image — and
then the fix was almost always a quick-n-careless application of the crop tool.
For really nasty instances (I was doing some well-off-the-beaten-path
experimentation), I had a vector-based template of a ColorChecker Passport. I’d
stretch the photo to the right dimensions and then overlay the border (with
cutouts for the patches) of the vector template on top. Not once did scanin
have a problem with that.
I suspect adding some fiducial marks or the like to the chart recognition file
would probably be the most that scanin would need. Or, if the file doesn’t have
expected CIE values, that would explain it.
b&
Other related posts:
