[argyllcms] Re: Generating a target from a reference file
- From: Pascal de Bruijn <pmjdebruijn@xxxxxxxxx>
- To: argyllcms@xxxxxxxxxxxxx
- Date: Fri, 27 Mar 2009 18:27:06 +0100
On Fri, Mar 27, 2009 at 6:03 PM, Pascal de Bruijn <pmjdebruijn@xxxxxxxxx> wrote:
> On Fri, Mar 27, 2009 at 5:59 PM, Pascal de Bruijn <pmjdebruijn@xxxxxxxxx>
> wrote:
>> On Thu, Mar 26, 2009 at 9:16 PM, Pascal de Bruijn <pmjdebruijn@xxxxxxxxx>
>> wrote:
>>> On Thu, Mar 26, 2009 at 9:01 PM, Pascal de Bruijn <pmjdebruijn@xxxxxxxxx>
>>> wrote:
>>>> On Thu, Mar 26, 2009 at 6:56 PM, Lars Tore Gustavsen
>>>> <lars.tore@xxxxxxxxxxxxxx> wrote:
>>>>> On Thu, Mar 26, 2009 at 6:01 PM, Pascal de Bruijn <pmjdebruijn@xxxxxxxxx>
>>>>> wrote:
>>>>>> On Thu, Mar 26, 2009 at 2:22 PM, Graeme Gill <graeme@xxxxxxxxxxxxx>
>>>>>> wrote:
>>>>>>
>>>>>>
>>>>>>>> Not exactly, preferably to RGB... Not all image editor can handle Lab.
>>>>>>>>
>>>>>>>> And I'd like to read back the RGB, to generate a newly (edited),
>>>>>>>> reference file...
>>>>>>>
>>>>>>> I thought that's what I outlined last time.
>>>>>>
>>>>>> I'm probably confusing terminology, which in turn is confusing you :)
>>>>>>
>>>>>> The point of the exercise would be to read back the edited image,
>>>>>> as a new similar (but edited) reference file.
>>>>>>
>>>>>> 1. Convert R080505.txt into an image of an IT8 chart, with the patch
>>>>>> colors
>>>>>> generated from the text file.
>>>>>
>>>>>
>>>>> I just had to play with this. Do you mean something like this:
>>>>>
>>>>> sed -n '17,304p' R080505.txt |awk '{print "convert -colorspace LAB
>>>>> -size 118x94 \"xc:LAB("$5"%,"$6"%,"$7"%)\" " $1".png\n"}' > run.sh
>>>>>
>>>>> Then you have to join them. I think there is a hint for that in the
>>>>> url posted above. I have to play with the kids now..
>>>>
>>>> I was already fiddling with your scripts, adapting it. I attached my
>>>> results. And
>>>> scanin does recognize the image.
>>>>
>>>> Now I need to see if I can read back the image as a reference file, with
>>>> (nearly) identical values. But that's something for tomorrow.
>>>
>>> Scanning back isn't that hard. I've included that script as well.
>>>
>>> However, for the conversion from the reference file, to the image, I have
>>> no explicit color space defined. So that's probably why the read back
>>> data it's matching the original data.
>>
>> I'm finding it hard, to get the same values read back...
>>
>> Also, the generated image seems yellowish... Possibly because of a
>> whitepoint mismatch?
>>
>> When converting from XYZ imagemagick seems to output Linear RGB,
>> but in what color space... Anybody have a good guess?
>
> Well isn't this typical... Minutes are mailing the previous post, I keep on
> fiddling a bit, and using ProPhotoRGB (ROMM), without gamma correction,
> completely solves my issues, and the original values are read back.
Correction, it seems when fakeread can't read the color profile
(because it's v4),
it just keeps the values from the ti2 file. This is why things suddenly matched.
So I'd still appreciate any pointers, for getting the original
reference file and the
fakeread file to match closely.
Regards,
Pascal de Bruijn
Other related posts:
