Re: Adding printer profiling support



(I'm CC'ing this to the list to keep the discussion there, hope it's ok)

On Fri, Feb 19, 2010 at 3:36 AM, Pascal de Bruijn <pmjdebruijn pcode nl> wrote:
> Some applications (even commercial ones) don't support LUT display
> profiles properly.
>
> Argyll can generate Matrix+LUT profiles, but then you have the problem
> that one app uses the matrix and the other the LUT, this can give odd
> inconsistencies... So it's not recommended for defaults...

Makes sense. Do you know if I'll be ok with a LUT using
GIMP/f-spot/ufraw? I should probably try g-c-m again to see what
profile it comes up with.

When I did the LUT one with 2000 patches the result seemed to have a
fair amount of extra contrast that looked good. Don't know where that
came from.

> This is true, however, 210 is already a _big_ improvement over 90...

Sure. Although I need to test it again as it was too little for me
before. But here is Graeme Gill's opinion on number of needed patches
(600-1000 for a "well-behaved" printer):

http://www.freelists.org/post/argyllcms/Printing-using-an-ICC-profile,22

I guess this one is easy to expose to the user by asking him how many
sheets to print.

> Please also note, that you're passing all kinds of options with
> unknown effects...

I just passed the options recommended by argyll itself (the mailing
list and the docs).

> Please try again with exactly the options I gave...
>
> Also make sure you have an up to date version of Argyll (1.1.0-final)...

Yes, this I did use in the last run but not the previous ones, so that
could be the problem with 210 patches.

> -q has little do to with actual quality with LUT profiles, it has to
> do with detailedness of the generated profiles... Generating a high
> detail profiles with low patches seems silly at best.

Makes sense, thanks.

>> Again the scenarios and the mailing list tell us to use a source
>> profile. Milan had an explanation for it.
>
> Do you have a reference for that?

Sure. Here are the docs:

http://www.argyllcms.com/doc/Scenarios.html#PP5

And Alastair Robinson's (of PhotoPrint fame) opinion on it was to try
different profiles there:

http://www.freelists.org/post/argyllcms/Printing-using-an-ICC-profile,3

It could be that we can actually get away without -S for a decent
default. We should probably start a thread in the argyll mailing list
about what to use as sensible defaults for profiling printers (and
monitors/scanners/cameras too maybe).

> We aren't catering to the masses... But even advanced hobby
> photographers and professional photographers see color management as a
> necessary evil, and most don't want to bother with patch counts and
> all kinds of settings, they just want a reasonable match by clicking
> on "Next" a couple of times...

Yes, I agree completely. My point wasn't that the GUI shouldn't start
as a straight-through "get a decent enough profile and be done with
it". I was just arguing that the user shouldn't be dumped into
learning all these argyll command line options every time he wants to
do something even moderately advanced.

> Again, the big question would be how often you _really need_ the
> advanced settings... Liking to fiddle with them and needing them are
> two different things...
>
> Trying to control to much was a mistake I made in the past as well,
> sticking to Argyll's defaults has generally improved my results.

I totally agree there. Only settings that are actually needed should
be exposed. I don't yet know what those are though considering all the
advice to fiddle with different settings I got from the argyll list.
Maybe most of that advice can actually be the default and not need
changing.

>> I think the underlying assumption behind telling users who want
>> advanced stuff to go to the argyll tools is that someone that wants
>> advanced calibration/color in Linux is also an advanced Linux user.
>> Whereas in reality it would be good to be able to support advanced
>> photographers/artists that are also non-geek computer users. It's an
>> uphill battle right now with GIMP not supporting 16-bit images, f-spot
>> not very complete and raw converters still not particularly good.
>> We're advancing in all three fronts though and g-c-m is a much needed
>> complement.
>
> Patience is key here...

No doubt. I wouldn't be using Linux to do photography work if I wasn't
happy to put up with and work around the shortcomings.

>> why -g?
>
> Adds greyscale patches.

So why deviate from the argyll standard there? This one didn't get
mentioned in my thread. Is it to improve the tone curve for B&W
printing?

>>> # printtarg -v -i CM -h -t 300 -p A4 test
>>
>> Does -t 300 help? I was getting some non-sharp edges on my patches
>> maybe it was because of this.
>
> This doesn't particularly matter much, you're not measuring the edges.

I wasn't too worried about it although I guessed it could mess up the
patch detection and maybe sometimes interfere with a reading,
especially when using the smaller patches.

Cheers,

Pedro


[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]