Re: Article on Photoshop CS4 and DeviceLink profile
Re: Article on Photoshop CS4 and DeviceLink profile
- Subject: Re: Article on Photoshop CS4 and DeviceLink profile
- From: Chris Cox <email@hidden>
- Date: Wed, 3 Dec 2008 13:13:38 -0800
- Acceptlanguage: en-US
- Thread-topic: Article on Photoshop CS4 and DeviceLink profile
Karl;
Who did you check your facts with? Even the Photoshop CS4 documentation would have corrected most of your mistakes.
Other applications may tolerate profiles that don't work - that has historically been the case. Photoshop has to be more picky, because our users have rather high expectations of things working. We fixup some profiles with well known issues, but have to reject many because they are incomplete or corrupt. Yes, we find a LOT of those in our testing. I'm still amazed at how some software can use a truncated table for color conversion...
Multichannel mode does not have a document profile - it's just a collection of channels, previewed using spot channel logic. Yes, we always have the display color space - but multichannel does not have a document colorspace to convert from (because we can't generally color manage them, and haven't special cased the few 3 and 4 channel profiles that could be color managed).
Abstract profiles are normally just "in-place" conversions. Applying them as a filter during another conversion is, well, kind of a bizarre use. If you want to use abstract profiles for creative color transformations - then the way Photoshop does it makes more sense (as sort of an adjustment, or adjustment layer). If you want to use them for color science experiments, then the way Photoshop does it makes perfect sense. Only if you wanted to use abstract profiles are a pre-correction to an output profile does your description make sense - and then, why haven't you baked that into the output profile?
Again, the standards for destination profile specs in device link profiles is too new - we had no samples to test with (other than a few generated in house specifially for testing). We've tried releasing untested code in the past - and won't do it again. One thing you have to remember - we had this feature coded and finished almost two years ago. We've been busy with other things since then (like testing, and fixing bugs, dealing with compiler and OS vendors to fix their bugs, etc.). As much as I'd like to support standards as soon as they're completed - there is some delay in getting acceptance of new standards, working out problems in standards, getting testable samples based on new standard,s and testing interoperability. And users had a need for the features now, not when the standards were ready.
Yes, there are limitations to the new profile support in CS4 - but please explain the limitations correctly without accusing Adobe of incompetence or some other agenda. And if you don't understand why some things are the way they are - ASK.
Chris
On 12/3/08 7:02 AM, "Koch Karl" <email@hidden> wrote:
Chris,
Am 02.12.2008 um 20:22 schrieb Chris Cox:
Wow. Someone should have done some fact checking before sending that out.
I did quite some fact checking before sending this out and I can back all of my factual statements. My opinions, though, are subjective, as is the nature of opinions ;-)
Yes, there are a lot of corrupt multichannel profiles out there - mostly from older profile makers that didn't test interoperability. Adobe applications do check for profile validity before offering the profiles for use. So, not all of your installed multichannel profiles may show up in Photoshop - because they may not be valid profiles.
Strangely enough, these profiles work well in other applications (not only those which created them). ColorSync Profile Repair doesn´t mark them as invalid or corrupted, either. I admit that these were old v2.0.0 profiles, not really state of the art.
The PhotoYCC profile is interpreted correctly - but the result is YCC, displayed as a multichannel document.
I never doubted that the profile is interpreted correctly in the B2A direction, it´s just A2B that sucks or in other words, is not suppoerted.
There currently is no way to provide realtime previews or display of multichannel profiles with more than 4 channels - so Photoshop opted to get users the basic functionality requested (making separations) and use the existing multichannel mode. The preview during the conversion is limited by what is in the multichannel profile plus a few limitations of how Photoshop creates the preview (it could be improved, but it'll take a lot of work).
Accepted.
Um, yeah - most multichannel profiles don't have enough information to correctly convert back to the original data.
Agreed. You will never get back to the original data. But there are users out there who have the crazy wish to print proofs from multichannel data, so they need to go from multichannel to proofer CMYK (or RGB).
And you're in multichannel mode without any information about the destination colorspace.
This I don´t understand. As far as I thought I had understood the innards of Photoshop, you always have the information of your monitor color space which acts as a destination colorspace.
The author should spend some time reading about how multichannel mode in Photoshop works (it is not a new thing).
OK, so it´s the same old mechanism, just pimped with a multichannel profile conversion.
Abstract profiles convert through the PCS. The source and destination can easily be the same profile (like the current document profile).
Agreed - "can" but doesn´t necessarily have to, except in Photoshop 11 where the user doesn´t have the choice.
Abstract profiles are used to "change the look" -- that's what they're for.
I think I understood that much.
The author seems to think they have some other purpose, but does not explain what that purpose might be.
Maybe it´s a problem of not being a native English speaker. I thought I had made myself clear. And I asked some precise questions which you didn´t bother to answer:
"When you select a rendering intent, where is it applied? On the way from working space to PCS, on the way back, or in both directions? I haven´t found out yet. If you want to apply the abstract profile on the way from your working space to an output space, you need to transform another time - you lose flexibility and each transformation costs image quality."
Nothing was goofed up for device link profiles. The primary demand for them was CMYK->CMYK (for print), and RGB->RGB (for non-print folks). Adobe has a limited number of engineers, a limited number of testers, and a limited amount of time to develop and test features. Rather than delay support of device link profiles until time became available to test every possible combination of input and output color mode, Adobe focused on what was requested by the users to solve real world problems.
Also, yes, after a device link conversion document remains in the original colorspace, because currently devicelink profiles do not contain information about the destination profile so that Photoshop could assign the correct profile after applying the device link.
AFAIK it was Adobe who requested a new tag in ICC profiles (pseq) that provides exactly this information. So, why don´t you use the information for applying the correct profile after applying the device link WHEN it is there?
The ICC has recently approved standards to improve the situation, but there were no samples available to test that before Photoshop shipped.
Maybe you should have asked basICColor ;-) (this was not meant as advertising!)
The author seems to have drawn several bad conclusions from a severe lack of knowledge, and a failure to do background research on the topic. This article is worse than useless, it is seriously misleading.
I´m always eager to learn. Unfortunately your anwers to my article didn´t really enlighten me. There is no new information and nothing that would make me take anything back. I´m not quite as dumb as you seem to believe.
What magazine was this published in?
It´s not yet published, I wrote, I had written an article, not that I had had one published.
And was it correctly labeled as advertising for basICColor?
Because I mentioned, that there is a solution out there? That doesn´t have Adobe´s name on it? Being the CEO of basICColor GmbH, I take the liberty to mention solutions that are working, although they bear the basICColor brand.
The article was not about Adobe-bashing. I wanted to warn the Photoshop 11 users of some pitfalls when working with the "new" type of profiles. Especially the device link conversions are error-prone. Not every user out there will understand that she/he will have to assign a different profile after a device link conversion. Converting again (e.g. to a proofer space) without having done so will wreak havoc!
The multichannel support is only half-hearted and the abstract profile support is a pragmatical approach - at best.
Best regards,
Karl Koch
______________________________
basICColor GmbH
Maistraße 18
82377 Penzberg - Germany
phone: +49-(0)8856-932505
fax: +49-(0)8856-932503
email: email@hidden <applewebdata://B9C78DB9-A913-4E15-A94F-0C3411B06B19/email@hidden>
http://www.basICColor.de
Managing Director: Karl Koch
Registered Office: Penzberg
Commercial Register: 172485, AG München
VAT No. DE814946213
______________________________
free profiles for standard printing conditions:
http://colormanagement.org <http://colormanagement.org/de/isoprofile.html>
______________________________
_______________________________________________
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