Re: BCP or no BCP to epson without rip
Re: BCP or no BCP to epson without rip
- Subject: Re: BCP or no BCP to epson without rip
- From: Gerhard Fuernkranz <email@hidden>
- Date: Thu, 11 Jun 2009 14:57:07 +0200
Graeme Gill wrote:
> Marco Ugolini wrote:
>> I think you're being a bit disingenuous here, Graeme. The Perceptual
>> conversion *procedure* first maps source black to PCS black, then PCS
>> black to destination black. While strictly speaking the two profiles
>> never "see" each other directly, the *procedure* certainly knows how
>> to go about it.
>
> That's not quite what you said, and many source profiles don't operate
> this way at all (consider all the matrix profiles out there for a start,
That's a good point since it is common practice to use perceptual intent
when converting from RGB working space profiles to output profiles (e.g.
printer profiles). But eventually this leads to a transformation
"working space -> relative colorimetric [since matrix/trc is relcol only
according to the ICC spec] -> PCS -> perceptual [B2A0 table] -> printer
color space", which is no longer a true perceptual transformation from
source to destination color space, but actually a mixed intent
transformation.
But the eventual solution should likely not be to make any assumptions
about the "other" profile, but one should alternatively use profiles
which do include perceptual tables, even for the RGB working color
spaces [I think to remember that on the ICC web page there is a V4 sRGB
profile available which does include perceptual tables, mapping between
sRGB gamut and PRMG].
> and even LUT ones may do whatever they want in regard to the black,
> including behaving colorimetricaly or map to some proprietary
> conventional intermediate black point).
IMO the basic ICC philosophy is indeed that the A2B0 tables of the
source profile should map the source device gamut to the PRMG, and the
B2A0 tables of the destination profile should do an inverse gamut
mapping, from the PRMG back to the destination device gamut, in order
that source and destination profiles do not need to know anything about
each other.
But this was clearly an unsolvable problem in former days, due to the
lack of a standardized reference medium and PRMG. So any V2 profile
unfortunately has to make proprietary assumptions regarding the PCS
gamut and PCS black point. And somehow we still have to cope with old
profiles already existing in the field...
I also think there are no doubts that this "universal" approach (going
through a common PRMG) may not be optimal, and that a direct gamut
mapping from source device gamut to destination device gamut will likely
give superior results in many cases. But that's not topic of this
discussion.
[Btw, I have generalized "black point" to "gamut", since perceptual
intent gamut mapping implies black point mapping anyway.]
Regards,
Gerhard
_______________________________________________
Do not post admin requests to the list. They will be ignored.
Colorsync-users mailing list (email@hidden)
Help/Unsubscribe/Update your Subscription:
This email sent to email@hidden