New project: camera color profile calculation - page 5 - RAW Shooting and Processing - CHDK Forum supplierdeeply

New project: camera color profile calculation

  • 480 Replies
  • 233667 Views
*

Offline ArtDen

  • ***
  • 175
    • dng4ps2
Re: New project: camera color profile calculation
« Reply #40 on: 28 / January / 2008, 16:22:22 »
Advertisements
rommel_a720is_2.zip First is with the original specs, then -1/3 and then -2/3
The matrix is
Code: [Select]
0.655856, -0.203610, -0.095090,
-0.061367, 0.670395, 0.058562,
0.018814, 0.095240, 0.244112
Not ideal, but...

*

Offline vit40

  • ****
  • 276
Re: New project: camera color profile calculation
« Reply #41 on: 28 / January / 2008, 16:30:40 »
Some changes in conditions. Before tacking shot exposure compensation have to be setted to -1/3 or -2/3.
It looks like too bright colors forces camera to apply stronger tone mapping for Jpeg image. Please take another photo with new condition and reupload it again. Sorry for unnecessary work.

True! Especially blue and yellow, with brightest color reaching level 255, are selectively oversaturated. Blown cyan colors are desaturated, preventing ugly cyan look of blown parts of the sky (like on some other cameras, even some DSLR-s) etc etc. I suppose in-camera color conversion uses lookup tables, so it's not possible to entirely emulate it this way, but doesn't matter. In most cases, I prefer my profile for A620, colors are more natural, although jpeg straight from the camera is sometimes more attractive.
« Last Edit: 28 / January / 2008, 16:33:29 by vit40 »

Re: New project: camera color profile calculation
« Reply #42 on: 28 / January / 2008, 23:11:18 »
I think the proper way of exposing the target would be to spot-meter the third grey patch from the bottom right. It's very similar to 18% grey that is usually used for determining exposure.

Here's an a570is picture of an actual mini-target exposed that way. White-balanced for daylight which is technically not exactly correct.

a570is_2.zip - 9.74MB

Reference for why i think that patch is 18% grey:
http://www.rags-int-inc.com/PhotoTechStuff/MacbethTarget/
« Last Edit: 28 / January / 2008, 23:15:06 by dvornik »

*

Offline ArtDen

  • ***
  • 175
    • dng4ps2
Re: New project: camera color profile calculation
« Reply #43 on: 29 / January / 2008, 03:14:25 »
a570is_2.zip - 9.74MB
Color is still very saturated there. Try to set exposure compensation to -1


*

Offline vit40

  • ****
  • 276
Re: New project: camera color profile calculation
« Reply #44 on: 29 / January / 2008, 03:41:21 »
ArtDen, ther's no need for darker picture than above, as brightest level is barely reaching 210-215

*

Offline ArtDen

  • ***
  • 175
    • dng4ps2
Re: New project: camera color profile calculation
« Reply #45 on: 29 / January / 2008, 03:49:06 »
vit40, may be. But matrix calculated from this image gives some inaccuracy in results.

*

Offline vit40

  • ****
  • 276
Re: New project: camera color profile calculation
« Reply #46 on: 29 / January / 2008, 03:54:03 »
Can you tell more, maybe I could be of some help ?

*

Offline fudgey

  • *****
  • 1705
  • a570is
Re: New project: camera color profile calculation
« Reply #47 on: 29 / January / 2008, 12:50:21 »
I don't have any of the DNG tools (nor Windows), so I figured I'd stick the a570is matrix into dcraw. I couldn't figure out if it could be fed to it on command line, but the matrices in the source looked sort of similar, just multiplied by 1e4 and rounded  to integers.

dcraw detects the A570IS as A620, so to try out the matrix from this thread

Code: [Select]
0,594925 -0,168536 -0,076592
-0,345835 1,029958 0,111052
-0,045052 0,156228 0,349470

I replaced this in dcraw.c

Quote
{ "Canon PowerShot A620", 0, /* DJC */
        { 15265,-6193,-1558,-4125,12116,2010,-888,1639,5220 } },

by this:

Quote
    { "Canon PowerShot A620", 0, /* ArtDen/dvornik A570IS */
        { 5949,-1685,-766,-3458,10300,1111,-450,1562,3495 } },

and tried 'dcraw -W' for a few CRW files comparing the regular dcraw, modified dcraw and the Canon JPG. I suppose I guessed correctly as the images from the modified dcraw are a lot better (=closer to the JPEGs) than from original dcraw with the A620 matrix.

They still are quite noticeably off in color (and brightness as well I suppose), though.


*

Offline ArtDen

  • ***
  • 175
    • dng4ps2
Re: New project: camera color profile calculation
« Reply #48 on: 29 / January / 2008, 13:07:05 »
Can you tell more, maybe I could be of some help ?
I don't have any idea why is this error occurred when program calculates matrix. Someone tries to create same software concurrent with me. May be he can create better algorithm. Anyway I will try to improve my algorithm later.

*

Offline ArtDen

  • ***
  • 175
    • dng4ps2
Re: New project: camera color profile calculation
« Reply #49 on: 29 / January / 2008, 13:13:34 »
I don't have any of the DNG tools (nor Windows), so I figured I'd stick the a570is matrix into dcraw. I couldn't figure out if it could be fed to it on command line, but the matrices in the source looked sort of similar, just multiplied by 1e4 and rounded  to integers.

I uploaded setup of dng4ps2 software with new matrices here: http://dng4ps2.homeftp.org/dng4ps2-0.2.2matrix.exe so you can use other RAW converter (not only dcraw) for developing your photos.

Btw, can you upload one or two images (RAW and Jpeg files) from A570. I need it to check my results

 

Related Topics