Mike Chaney's Tech Corner
December 02, 2020, 10:28:14 AM *
Welcome, Guest. Please login or register.
Did you miss your activation email?

Login with username, password and session length
News: Feb 2013: Qimage Ultimate Challenges... have fun and explore features!
 
   Home   Help Search Login Register  

Professional Photo Printing for Windows
Print with
Qimage and see what you've been missing!
Pages: [1]
  Print  
Author Topic: Needing a Green bias and other SD14 related questions  (Read 10812 times)
Steaphany
Newbie
*
Posts: 5



View Profile Email
« on: June 06, 2009, 12:22:02 PM »

I'm new to the forum and been working my way up the Qimage learning curve.

My dSLR is a Sigma SD14 and I'm trying to achieve a high degree of image accuracy. I have a X-Rite ColorChecker which I have shot RAW in full Sun light with a Sunlight white balance.

Within Qimage, I am trying to create an accurate Selective Color Table. The problem that I'm having is the SD14 recorded the Red and Magenta squares Saturated, 165 0 58 and 194 0 161 respectively - Completely lacking anything in the Green channel. The sRGB target values for the X-Rite ColorChecker are RED: 175 54 60 and Magenta: 187 86 149, not what I'd call saturated colors.

Since the operation of the Selective Color Table is a pure multiplication, How do I add an appropriate Green bias selectively or to the overall image so I can have Green values to work with ?

The other X-Rite ColorChecker primary color squares, Blue, Green, Yellow, and Cyan each have Red, Green, and Blue channel values, so they are no problem to accurately alter with the Selective Color Table.

I know the SD14 tends to be Red sensitive, sacrificing the Green channel in the process. I also know the SD14 characteristically records intense, saturated, colors, but my goal with the Selective Color Table is to achieve a high correlation between the image and the actual scene.

As an aside, Why does the Embedded Image Information screen list the Depth for a RAW X3F file as 8 Bits when the SD14 uses 12 bit Analog to Digital converters ? Does this imply that Qimage lost some data along the way when the file was opened ?
« Last Edit: June 06, 2009, 12:26:47 PM by Steaphany » Logged
Terry-M
The Honourable Metric Mann
Forum Superhero
*****
Posts: 3197



View Profile WWW
« Reply #1 on: June 06, 2009, 01:02:40 PM »

Firstly, about the selective color table.
While in the table, click F1 to get the Help on the subject.
I think you need to enter a value, right click and select Copy entry to same column.
You do not say if your monitor is calibrated, the values you read on the screen will depend on the capabilities of the monitor and its calibration icc profile.

Surely the "correct" way to get correlation between scene and image is to use colour management with calibrated monitor, and printer profiles.

From what I know, especially with highly saturated colours, there is no such thing as perfect correlation from scene to image because the gamut of monitors, printers etc. is far less than then the original scene and the raw image.
See my post here on gamut comparison http://ddisoftware.com/tech/computer-software/colour-gamut-visualisation/

If colour management is a "black art" to you, come back & I can point you to some reading matter.

Qimage works in 8 bits except that raw files are processed in 16 bits (or at least the 12 of your images) so nothing is lost in that sense.
Mike wrote an article about 16 bit printing which has some useful background:

http://ddisoftware.com/tech/articles/december-2006-hype-or-hero-take-2-16-bit-printers/

Terry
« Last Edit: June 06, 2009, 01:14:20 PM by Terry-M » Logged
Steaphany
Newbie
*
Posts: 5



View Profile Email
« Reply #2 on: June 06, 2009, 01:39:23 PM »

Thanks for the quick response Terry.

First of all, I do have my monitor calibrated, which I update about every other month.

I am not trying to set the Selective Color Table values by eye.

My work flow consists of:

Opening the image in Qimage with Create Associative Filter => Full Edit

I then use to cursor to examine each of the X-Rite ColorChecker primaries, recording the values that Qimage lists as "Curr. Color", which I'm assuming are the file values and not what XP sends to the monitor post color management.

I then do a bunch of calculations, dividing the X-Rite ColorChecker sRGB values by the measurements from Qimage to derive the Selective Color Table values.

For example, Qimage reports the "Curr. Color" of the Green square as [ 96, 146, 87 ], The ColorChecker target values are [ 70, 148, 73 ], so the conversion values needed by the Selective Color Table become [ 70 / 96, 148 / 146, 73 / 87 ] or [ 0.729167, 1.013699, 0.839081 ]

Now the problem comes when the reported "Curr. Color" contains a 0, Red being [ 164, 0, 56 ], and the ColorChecker target value is [ 175, 54, 60 ]. Calculating the Selective Color Table value ends up being [ 175 / 164, 54 / 0, 60 / 56 ] leaving the value for the Green channel effectively undefined. Tentatively, I've used the values of [ 1.067073, 1.0, 1.071429 ] knowing that this will leave the Red saturated.

This is a problem where Qimage uses the formula:

< Filtered Pixel Value > = < Selective Color Table term > * < Initial Image Pixel Value >

and not the formula:

< Filtered Pixel Value > = < Selective Color Table term > * ( < Initial Image Pixel Value > + < A Bias Value > )

If I could add a Green bias, lets say 10, across the whole image, I could create a Selective Color Table which would compensate and the math would work for the Red and Magenta to be on target.

Afterwards, I have no problem accepting that display or printing of an image will have the reproduced colors limited by the gamut mapping.
« Last Edit: June 06, 2009, 01:47:22 PM by Steaphany » Logged
admin
Administrator
Forum Superhero
*****
Posts: 3181



View Profile Email
« Reply #3 on: June 06, 2009, 01:52:59 PM »

The X-Rite ColorChecker is meant to be more of an exposure check than a calibration tool.  It isn't appropriate for calibration or profiling because there aren't enough color patches to work with.  You can make a change to the magenta patch and it may only be appropriate for bright/saturated magenta and it might make duller magenta colors look much worse.  You really need an IT8 target at a minimum to get good camera profiles.  Also, are you trying to profile raw photos from the SD14?  It's not possible to profile JPEG's in most cases because the camera modifies the tone curves and color before you ever get to the profiling stage so you end up with a moving target.  You can make decent raw profiles, but I've already made one of those for the SD-14.

All that said, you can still do what you want, or at least come close.  If there is no value in the green channel, your only option is to add some percentage of the red or blue channel to the green channel.  This can be done in the selective color table.  For example, if in the green channel you enter 0.1,0.0,0.0 using commas as shown, 10% of what is in the red channel will be used for the green channel.  That's how you can get a value in the green channel when there is none present.  This is covered in the help if you press F1 while in the sel clr table.

Mike
Logged
Terry-M
The Honourable Metric Mann
Forum Superhero
*****
Posts: 3197



View Profile WWW
« Reply #4 on: June 06, 2009, 01:54:00 PM »

Quote
I then use to cursor to examine each of the X-Rite ColorChecker primaries, recording the values that Qimage lists as "Curr. Color", which I'm assuming are the file values and not what XP sends to the monitor post color management.
I think this is a wrong assumption but someone more expert than me would need to confirm or otherwise. I think that what the cursor reads is the screen value, therefore it depends very much on your monitor profile.

Your approach seems very unconventional and a lot of hard work, that is why I said about colour management. The conversion from one space to another is done by clever algorithms and the rendering intent selected.
If you are using Qimage for raw conversion, are you using a custom camera profile, well worth the few $ and you'll get the best possible correlation of colour.
Terry.
« Last Edit: June 06, 2009, 01:55:38 PM by Terry-M » Logged
Steaphany
Newbie
*
Posts: 5



View Profile Email
« Reply #5 on: June 06, 2009, 06:00:21 PM »

Thanks Mike, sorry for not reading far enough in the manual.

I see I have a bit of math ahead of me, need to pull out the old books on solving simultaneous linear equations.

Mike, can you please clarify ?
Is the Current Color numerical values from the image file, memory resident with filters applied, or from the monitor values post Color Management gamut mapping ?
Logged
admin
Administrator
Forum Superhero
*****
Posts: 3181



View Profile Email
« Reply #6 on: June 06, 2009, 07:09:08 PM »

Thanks Mike, sorry for not reading far enough in the manual.

I see I have a bit of math ahead of me, need to pull out the old books on solving simultaneous linear equations.

Mike, can you please clarify ?
Is the Current Color numerical values from the image file, memory resident with filters applied, or from the monitor values post Color Management gamut mapping ?

It's the current pixel value displayed on screen.  So if you want the numbers without (monitor) color management, you need to turn Mntr ICC off on the main window first.

Mike
Logged
Pages: [1]
  Print  
 
Jump to:  

Powered by MySQL Powered by PHP Powered by SMF 1.1.21 | SMF © 2015, Simple Machines Valid XHTML 1.0! Valid CSS!