PDA

View Full Version : Gamma correction


robertj
14th March 2006, 10:17 AM
Does RawShooter Premium 2006 apply gamma correction? If it does is it constant (hard coded in the program) for all cameras or varies from camera to camera (actually from CCD/CMOS to CCD/CMOS)?

As I understand it, CCD/CMOS records light intensity (luminosity) linearly, while human eye perceives logarithmically (i.e. deforms the signal in order to be able to cope with huge range of intensities, similar holds for hearing and ear) and the output from the CCD/CMOS (although very close to the light of the original scene photographed) it has to be corrected.

The gamma function is: Ic ~ Ir^g, where Ic is a corrected light intensity, Ir recorder light intensity, g gamma (I can't write Greek letter gamma in ASCII), ~ a sign for proportionality and ^ a sign for exponent (i.e. x is proportional to four squared would be written as: x ~ 4^2).

What I do not understand is the following... if our eyes gamma-correct the light from any source, either from an original scene or its photograph, why a (CCD/CMOS's) photo, which will be light-wise close to the original scene, needs a gamma correction (if CCDs/CMOSes are linear)? In other words, if the nature does not correct its light (from natural scenes) to suite our non-linear eyes, why a photo would need to do that?

A reliable source says... Gamma correction. Digital raw captures have linear gamma (gamma 1.0), a very different tonal response from that of either film or the human eye. So the raw converter applies gamma correction to redistribute the tonal information so that it corresponds more closely to the way our eyes see light and shade.

What is do not understand is why one ought to redistribute the tonal information so that it corresponds more closely to the way our eyes see?

R J

MichaelT
15th March 2006, 06:30 AM
We perceive all sight through our eyes. From nature, and photographs. The sensor (a kind of machine) does not react to light the same as our eyes. Therefore the light from nature is rendered differently by the sensor therefore capturing it in an unnatural way. So yes, all image processing must make this correction from linear to non-linear which we call a gamma correction. All digital processing must do this to render a photograph as viable to our eyes.

robertj
15th March 2006, 09:10 AM
Hello Michael,

Thanks for your replay.

I do not want to enter into theoretical debate, but I still do not understand.

We perceive all sight through our eyes. From nature, and photographs. The sensor (a kind of machine) does not react to light the same as our eyes. Therefore the light from nature is rendered differently by the sensor therefore capturing it in an unnatural way.
All clear here, although it is a matter of convention, it think, to call our eye's light capturing natural and CCD/CMOS's unnatural.

So yes, all image processing must make this correction from linear to non-linear which we call a gamma correction. All digital processing must do this to render a photograph as viable to our eyes.
This is for me a tricky bit as I do not understand why the correction is needed, i.e. it should be irrelevant to our eyes if the light is coming from a real scene (which is by default not corrected) or an uncorrected photograph, assuming that the light from the photograph is close (in a sense that our eyes cannot see the difference, i.e. there can be and certainly is a difference in light spectrum) to that of the real scene (I assumed that all is linear, i.e., CCD/CMOS, paper, etc. which may not be true, but CCD/CMOS should be linear and I could have refereed to CCD/CMOS image rather than printed photograph). If we correct the light captured by CCD/CMOS and then again our eyes corrects it, it looks to me like the correction will be applied twice.

Regards,

Robert

MichaelT
16th March 2006, 06:06 AM
The scene is not linera, so your eyes do not do the correction. The sensor IS linear, so we must do the correction to the captured image.

better?

robertj
16th March 2006, 06:25 AM
Hello Michael,

I found the answer, details here (http://www.amateurphotographer.com/forums/showthreaded.php/Cat/0/Number/302616/page/0/vc/1) and here (http://www.amateurphotographer.com/forums/showthreaded.php/Cat/0/Number/302651/page/0/vc/1)...

The key point is that (almost as a rule) output devices have anti-gamma response, so the two gammas cancel, also traditionally film was non-linear so somehow it has ben established as a rule that JPGs/TIFFs/etc. should be gamma corrected by, I assume, some default amount of gamma (do not know if the gamma is written in the file so that it can be read out), this is called, as I hear, implied gamma.

Do you know what is the gamma applied by RawShooter premium 2006 (http://www.pixmantec.com/products/rawshooter_premium.asp)? Thanks.

Regards,

Robert

MichaelT
17th March 2006, 06:21 AM
I beleive that RS uses a 2.2 gamma.

datadump
27th March 2007, 05:16 PM
does Adobe Lightroom apply gamma correction (2.2) automatically? is this configurable?

MichaelT
28th March 2007, 07:59 AM
All "normal" looking images are set in a gamma space generally between 1.8 and 2.2 depending on the color profiles involved with the image, working space and display (or printer). Since LR uses the ProPhoto working space internally, the images are probably in gamma 1.8, but as displayed on your monitor probably at 2.2 depending on the monitor profile. All of this is automatic within any color managed application such as Lightroom. A RAW file originally is linear (gamma 1.0) and must be transformed to appear as a "normal" image, otherwise it will look dark and murky, nothing that the human eye would relate to as a real image.

datadump
29th March 2007, 01:19 AM
slightly off topic but you said LR uses Prophoto internally. But isnt that configurable? ie: preferences

datadump
29th March 2007, 07:26 AM
slightly off topic but you said LR uses Prophoto internally. But isnt that configurable? ie: preferences

ah disregard. that setting has nothing to do with how LR manages colour itself. sorry.

MichaelT
29th March 2007, 09:37 AM
Right. Actually LR uses a modified ProPhoto internally, but that is transparent to us, except that all readouts and histogram is references to ProPhoto and not s or a RGB, and depending on who you talk to this is not the proper approach, since in the end, it will have to me gamut compressed by something somewhere.