New project: camera color profile calculation

ArtDen

obitus
So I converted 2 files with different Whitebalance (1 cloudy, 1 manual) from CRW to DNG. The JPG look different, the DNG look the same. I use Adobe Lightroom.

DNG files look the same as my program inserts existing jpeg files into dng ones as preview images. When you convert DNG to image with Lightroom, it produces different images as it uses own auto white balance point calculation.

vit40

ArteDen,

Any chances of getting a version of DNG4PS2 with possibility to modify matrices ?

ArtDen

vit40Any chances of getting a version of DNG4PS2 with possibility to modify matrices ?

I planned it many times ago, but still don't find free time for implementation

vit40

Ok, I will wait. Thanks for your efforts

wontolla

Hello ArtDen!

DNG files look the same as my program inserts existing jpeg files into dng ones as preview images.

I always wondered why, and now that you mention it, I'll dare to ask:

What is the reason for this? I mean, why not a preview image based on the DNG itself instead of the JPG file?
That way, with just the preview, people can decide if the DNG is worth processing or to stick with the JPG result.

BTW. Thanks for all your effort and time on this program. You even keep it up to date with CHDK's new features like AutoISO!

GrAnd

wontolla
I mean, why not a preview image based on the DNG itself instead of the JPG file?

I think, because the JPG, which is automatically produced from RAW/DNG, will be exactly the same.
Even the natural Canon's CRW file has a JPG preview the same as standalone JPG.

lukg

Here my S5is pics. I took only the palette picture cause it's raining and cloudy outside 😁 , not very good weather to take colorful photos! i'll take them asap:

http://www.zshare.net/download/103717017fc52496/

Howdy

ArtDen, did you check lukg's pictures? Are they OK for color matrix' calibration?

If they are not adequate, please write, so I (or lukg, whoever is faster 😀 ) could provide you with better pictures.

THANK YOU VERY MUCH FOR YOUR EFFORTS!!!

ps. is your calibration program ready to be published? I would really like to play with it a little bit...

ArtDen

Howdy, I don't check lukg's pictures yet. Wait a little 😀

ryanpg

Hi all. I'm trying to build dng4ps2 on linux. I've got codeblocks installed and I can build the app from source but the button labels are all wrong i.e. they read: "btnOK" and "btnCancel." I have gone through and manually edited the code fixing the button labels, however, now the "calibrate" function simply crashes the app.

Is there any development happening in a VS like cvs, svn, bzr, git etc. that I can track?

Where can I find the "latest and greatest" source?

Is the matrix work relevant to linux - where can I get the source with the matrix work included?

Should I provided my "button label fixes" to someone, somewhere?

Thanks very much!

yujin

Hi ArtDen, here are my S2 pictures. Hope that helps :-

zSHARE - s2 is.zip

lukg

Artden, i took 2 quick pics before going to work! Tell me if you need more or better pictures! (the palette pic is in my prev post).
Thanks

zSHARE - nuova cartella.rar

arkanoid

Here's my S5 monitor palette shot

s5is.zip - 9.12MB

One question for the developers:
exploring the source code I found this matrix for S5:

13124,-5329,-1390,
-3602,11658,1944,
-1612,2863,4885

why it's so different from other cameras matrix?
A720IS for example:
0.640019,-0.220031,-0.096241,
-0.077419,0.639766,0.044009,
0.017965,0.078396,0.231868
I'm talking about number format, not values 😀

thanks

whoever

arkanoid
exploring the source code I found this matrix for S5:

What you have found is not the color transfer matrix of S5 but rather of PowerShot A640, calculated by Dave Coffin (dcraw) and presented in his format.

arkanoid

whoever

arkanoid
exploring the source code I found this matrix for S5:

What you have found is not the color transfer matrix of S5 but rather of PowerShot A640, calculated by Dave Coffin (dcraw) and presented in his format.

ok, I know it's the same of A640 but it's also under S5 brakets.
Just a newbie question.
This means that there's no specific S5 color transformation matrix at the time and is still "work in progress"?

whoever

arkanoid
This means that there's no specific S5 color transformation matrix at the time and is still "work in progress"?

I should guess so, yes. S5 port was released only recently. Maybe DataGhost, the guy who did the port, looked into it, dunno... Otherwise, S5 has the same sensor as A720 and Ixus950/SD850, so try those matrices, although that's "work in progress" too.

vit40

CHDK for S5 is available only several days I think, so is the raw support for it

About format of color matrix, you can multiply ALL matrix numbers with any number you like, and result of color space conversion will be the same, so for example, matrix

13124,-5329,-1390,
-3602,11658,1944,
-1612,2863,4885

and

1.3124,-0.5329,-0.1390,
-0.3602,1.1658,0.1944,
-0.1612,0.2863,0.4885

will give the same result

ArtDen

arkanoid,
you forgot to set focus to infinity.

lukg,
My internet connection is too slow and unstable. I can't download your big files 🙁

lukg

Don't worry Artden! If i can send you in any other way (ie. for resuming) , just tell me!

Anyway, can I ask you what the numbers of the matrix stand for? Are they related to primary RGB colors and relative feature, like saturation, etc. (since it's a 3x3 matrix). Apologize me if the question is really stupid 😁

ArtDen

lukg
Don't worry Artden! If i can send you in any other way (ie. for resuming) , just tell me!

Just upload files with small sizes

lukg
Anyway, can I ask you what the numbers of the matrix stand for? Are they related to primary RGB colors and relative feature, like saturation, etc. (since it's a 3x3 matrix). Apologize me if the question is really stupid 😁

Don't understand what do you mean. DNG supports 2 color matrices for different illumination conditions (or only one). I calculate one matrix for sunny illumination. Matrix transforms values from point of XYZ color space into camera color sensors values. Saturation, brightness etc are also implicitly included in matrix.

« Previous Page Next Page »