[argyllcms] Re: Question regarding White Point and spotread
- From: <robert@xxxxxxxxxxxxxxxxxx>
- To: <argyllcms@xxxxxxxxxxxxx>
- Date: Sun, 9 Nov 2014 12:00:22 -0000
Graeme Gill [graeme@xxxxxxxxxxxxx] wrote:
robert@xxxxxxxxxxxxxxxxxx wrote:
> Which is exactly what the tutorial provides.
Yes, your tutorials are an excellent first step, or pointer on where to go
next. They do assume significant knowledge though (it may not seem so to
you, but to someone new to the subject, the very notion of profiles and B2A
tables etc., is incomprehensible). Yesterday I spoke to a professional
photographer (a very good one actually), who has been sending sRGB files to
his printer since he went digital and wondered why he wasn't getting the
vibrant colors that his printer was supposed to produce.
>> All-in-all this represents perhaps a dozen Argyll commands ... but, more
>> importantly, their use requires no knowledge of color management once the
>> batch files have been made, so that anyone should be able to use them
>> without having to delve into the complexities of your software.
>Right, but the batch files have been configured to suit your particular
>details.
Up to a point because I haven't tried to generalise them. However the batch
file prompts me for specifics, like numbers of patches, whether I want OBA
compensation or not, whether I want a conditioning profile, what the target
illuminant is etc. With a little work (especially if using Perl) the batch
files could cover the bulk of everyday tasks that photographers need for RGB
printer profiling. There's no reason that I can see why similar batch files
couldn't be produced for CMYK profiling, camera profiling etc.
>I've invested 10+ years of forgone salary. It has cost me very much more
>than you.
>But that seems not enough - you want more, and all for free!
I certainly don't expect anything for free and would be more than happy to
pay. After all, I have purchased i1Profiler Publish and loads of other
software that has cost me (and continues to cost me) thousands, so paying a
few hundred for Argyll would be entirely fine by me.
I don't know why you've developed Argyll, but I assumed that it was out of
interest, and also because you do license it out, presumably for a fee. At
any rate you now have the opportunity to make money out of it if you so
wish, and my suggestion is that providing an infrastructure that would make
it usable/understandable to the common man would be entirely chargeable.
I've had endless discussions on Luminous Landscape extolling the virtues of
Argyll but I've found the interest to be nil. People just don't want to go
the command-line route, especially if they want to have to figure out the
intricacies of every individual command (and which commands to use in the
first place).
>> But I feel that the software is
>> inaccessible to most of us and I am disappointed that the users of Argyll
>> don't get together to do something to make it less so (of course
DispCalGui
>> is a notable exception).
>And I'm quite content with the situation. Why should I help or even
>encourage the development of something that cuts me out of any possible
>compensation for what I've done ?
Well, if you're planning to do this development yourself then I for one will
be delighted (and very happy that you make some money out of it). I would
like to use Argyll because I think it's very good, but I don't have the time
to spend learning it. If more comprehensive support were available (at a
fee, no problem) or if an infrastructure was available (even better and
happy to pay for it) that meant that a lower-level of support was enough,
then that would be wonderful.
>> I do understand that YOU cannot because you don't have the time ... and
also
>> probably because you know too much.
>Lets see you and every other person who has ever used ArgyllCMS donate even
>half as much as the cost of Profile Maker, and I might get more enthused
>about helping or encouraging more wide spread use. And such assumptions
>about
>my awareness and understanding of the situation, is really not appreciated.
>By all means suggest improvements, but don't assume that in some way I am
>inherently incapable of understanding.
Well, I have donated a reasonable amount, considering that you make the
point that you can only accept moderate donations because of licensing
issues.
I developed some tools for watercolour painting using Photoshop that I used
to give away for free ... and I found that thousands of people were
downloading them and I was getting deluged by questions. So I now charge
for them and the downloads have dropped from tens of thousands a year to
about a thousand a year, I get some money for my work and have to provide a
lot less support because the people who do download the tools are serious
about using them. Because I get paid for the tools it also makes it more
interesting for me to keep developing them, adding instructional videos
etc., all of which adds to my revenue stream. So I would be the very last
person to expect you to do this for free. However, you've made Argyll
available under Open Source so you can't change the model overnight ... but
of course you can add tools and support around the software that you would
be free to charge for.
To ask people to pay hundreds of dollars for a possible future and
unspecified product isn't going to attract too many donors, unfortunately.
I would say that you are probably one of the most intelligent people I've
come across (and I've employed hundreds of very intelligent software and
hardware engineers in complex areas like telecommunications, robotics and
instrumentation) so I intend no disrespect whatsoever when I say that you
don't appear to understand the needs of people like me. It's a common
problem in software (and science, philosophy ...): when we know a great deal
about a subject it makes it difficult for us to understand how little others
do, and so we use terms that make complete sense to us, but make no sense to
our audience.
>> But, the fact is that what I need is a simple:
>> "If you want to validate your profile, this is how you do it", or "If you
>> want to check that your printer is printing colors correctly, this is how
>> you do it". I don't need to know that -ir does nothing, for example ...
all
>> that does is to raise more questions in my head and it doesn't help me
>> (directly) to find out how to do what I want.
>You seem to want me to somehow divine your situation and intentions, and
>then figure it all out for you. 1) That's more work than I want to do and
> 2)That doesn't solve the underlying issue, which is that you don't quite
know
>how to achieve your goal. This is a classic "give a man a fish and you feed
>him for a day; teach a man to fish and you feed him for a lifetime"
>situation.
>If I can get you understanding what's going on and how ArgyllCMS works,
>then you can figure this stuff out yourself, and I am relieved of a support
>burden.
Well, no. My questions have been quite specific, so I'm not asking you to
divine my requirements.
However, your point is entirely valid. If I could fully understand color
management and how Argyll works then of course I could figure all of this
stuff out for myself. And I am prepared to put effort into it ... I even
downloaded your software and started to have a look, but I just don't have
the time. Books on color management seem either way too basic or way to
complex ... and the ones that MIGHT be better cost a fortune. There is also
a lot of incorrect information out there that muddies the waters further.
The whole subject is very complex and would take a lifetime of study to
understand well. So, is it reasonable to expect a typical user to fully
understand color management and how Argyll implements it? Is this education
a task that you want to (or can) undertake?
>So to answer your specific question: To independently verify a device
>profile, create an independent set of test values, print and measure them
>exactly the same way you did for creating the original profile. Then use
>profcheck on the profile and the check .ti3 file.
Well that is what I've been trying to do :).
1. Create an independent set of test values. How do I create an independent
set of test values that I can specify? I know I can use targen, but I have
little control over the generated colors (or else I don't understand how to
use targen to limit the colors to the ranges that I am interested in). I've
tried to enter values into a .ti1 file but I just get parsing errors. What
would be tremendously useful would be a utility that takes a set of RGB
values and creates a .ti1 file from it, correctly formatted.
2. print and measure them exactly the same way you did for creating the
original profile. That's fine and that is what I've been doing: print with
no color management.
3. Use profcheck. Well, you suggested before to use fakeread and colverify,
which is what I did. Rather than change to profcheck now I would prefer to
make sure that I am using fakeread and colverify correctly (which is why I
listed the commands I'm using). I found that if I did not use '-N' in
colverify that I got large dE errors (40-60), but with '-N' the worst errors
were around 2.5 with an average of around 0.5. I think I understand, from
your previous reply, that '-N' effectively simulates chromatic adaptation
and that I can use it or not, as I wish. My question then is: if I do not
use '-N' and get dE00 errors of around 50, on a paper that has just been
profiled and on a very good and very new printer (Canon iPF6400),and on
very, very good paper (Canson Platine, Baryta and PhotoHiGloss): what is
this telling me? That I have a bad profile? That it's the best that I can
expect? What? Then if I do use '-N' and get pretty good results, what does
THAT tell me? That the profile is good? That the software is aligning the
bad values so that the dE values become good? Why does chromatic adaptation
come into it when both profile and verification are being done using the
same instrument? Since I'm using Argyll for both the profiles and the
verification and using the same i1Pro2 (that I have checked and which does
give VERY consistent readings even after recalibration) I would expect the
verification results to have very low dE values (and certainly not up in the
50s).
In the commands below, is the problem cctiff? It is embedding the profile
into iPFTest.tif, so I am effectively printing it with color management.
But if I did not, would fakeread not then give a color-managed simulation
that would be compared by colverify to a non-color-managed scan?
targen -v -d2 -G -f100 iPFTest
copy iPFTest.ti1 iPFRef.ti1
printtarg -v -r -ii1 -a1.0 -T300 -M6 -pA4 iPFTest
cctiff -v -ia -e profile.icc iPFTest.tif iPFTestO.tif
move /Y iPFTestO.tif iPFTest.tif
fakeread -v -Ia -l profile.icc iPFRef
Pause Print iPFTest.tif using no color management.
chartread -v2 -l iPFTest
Pause The test results will be in iPFValidate.txt
colverify -v2 -N -k -s -w -x iPFRef.ti3 iPFTest.ti3 >iPFValidate.txt
Robert
Other related posts:
