Re: GaMapICC
Re: GaMapICC
- Subject: Re: GaMapICC
- From: Martin Orpen <email@hidden>
- Date: Sun, 19 Oct 2008 22:47:08 +0100
On 19 Oct 2008, at 00:48, Graeme Gill wrote:
I'm not sure exactly why Martin sees a color count increase when doing
color conversions if he is doing 8 bit -> 8 bit, but dithering
cannot be
the explanation if Argyll is being used to do the pixel conversion.
I was confused by that myself until I parsed my Photoshop log file and
remembered that the separated image had been sharpened because it is a
live job that is going through our system at the moment. Interesting
that an amount of 149% with a 1.9 pixel diameter and 0 threshold can
add so much colour to this image :-)
Note that Argyll profiles and device links may or may not include
colorspace
specific and image specific gamut mapping, so his distinction between
"device link conversion" and "gamut mapped conversion" is not
something I follow.
Apologies for not making this clear. And to make sure that I haven't
confused myself by comparing images that have received different
treatments I've done the following:
1. Downloaded a fresh copy of Klaus' GaMapICC and set it to convert to
ISO Coated v2 (ECI):
2. Taken a crop from the master RGB file which measures 1428x1141 pixels
3. Processed it through GaMapICC using "Source Profile specific" gamut
mapping strategy
4. Processed it through GaMapICC using "Image specific" gamut mapping
strategy
Visually I see the same problems that I had with the versions that I
posted in the previous email -- a lack of detail in the "image
specific" gamut mapped version.
A count of colours (without unsharp masking this time) gives:
RGB: 3468
Source Profile specific mapping: 1786
Image specific mapping: 1690
My expectation was that Image specific mapping would give a better
result -- one that would be worth the additional time taken to convert
the images. But profile specific mapping seems to give a better result
on *difficult* images and the differences on *regular* images are near
impossible for the human eye to detect in the tests that I've run here.
Regards
--
Martin Orpen
Idea Digital Imaging Ltd
_______________________________________________
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