Re: ICC specification bug
Re: ICC specification bug
- Subject: Re: ICC specification bug
- From: Henrik Holmegaard <email@hidden>
- Date: Sun, 10 Dec 2000 08:25:21 +0100
There seems to be an implied agreement to consider Perceptual the defult,
both when intent is undefined, and to label an undefined intent... isn't this
what is happening here? Wouldn't AtoBn bust a lot of existing software?
The concept of selecting an intent doesn't apply to link profiles.
The only thing you can do with a link profile is select it or
deselect it in a user interface.
According to the spec, the place you learn what the conversion type
is, is the alphabetic name ('Perceptual', 'Relative Colorimetric'
etc.). The alphanumeric device profile LUT naming scheme doesn't
apply, so why carry over this scheme unchanged, if the consequence is
that the alphanumeric name is misleading nine times out of ten?
That is, if you build link profiles, it isn't to separate RGB into
CMYK. You can't embed the darn thing because embedding for
repurposing isn't supported cf the spec. You could prebuild
softproofing links and load them, I suppose, the same way you could
hard convert a CMYK raster image into monitor RGB and load it from
disk into QXP as softproof, but who wants to when device profiles do
the job with economy and ease of use.
The only use links have is CMYK to CMYK relative or absolute
conversions, and for that you need all the information you can get
about the original source and destination spaces and the intent.
An ICC profile is only as good as your specific record of the device
setup of which there are here a minimum of two, and only as good as
the calibration of the setup(s), too, of course.
So...
--
Henrik Holmegaard
TechWrite, Denmark