Pages: [1]   Go Down

Author Topic: Camera Profiling  (Read 2812 times)

Robert Ardill

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 657
    • Images of Ireland
Camera Profiling
« on: July 02, 2014, 09:17:09 AM »

Andrew Rodney's excellent Youtube video has encouraged me to use camera profiling.  I had thought that there was little point as the light varies so much from hour to hour and season to season; but Andrew makes a compelling case that it's the general shape of the light spectrum that we need to look at, that the daylight spectrum varies little between dawn,noon and dusk or summer, winter; and that what we are doing is essentially calibrating our individual camera sensors to a generalized lighting condition.

I've played around a little with camera profiling and what certainly seems to be true is that I get more color fidelity if I use an ACR profile created using DNG Profiler or ColorChecker Passport rather than the Adobe Standard profile, irrespective of the outside light.  That makes sense to me because the profile I generate is for my camera whereas the Adobe profile is a generic (average?) one.

However, I'm not so convinced about the time of day.  Here is a plot of the spectra at my house yesterday: mid-day cloudy, early evening with a blue sky, and sunset with a slight haze:



It seems to me that the early evening, blue sky has significantly less green and red, whereas the sunset slight haze has a dip in the yellows and a peak in the red, and the mid-day cloudy is overall flatter than the other two (nearer D50).

The thing that's a bit concerning in terms of camera calibration is that there was only about 20 minutes between the Early Evening Blue Sky and the Sunset Slight Haze - and there is a big peak in the reds in the latter case (together with a significant drop in luminance).  I can't see how only adjusting the white balance of an image would compensate for this ... but, hopefully I'm wrong.

Advice welcome!

Robert
Logged
Those who cannot remember the past are condemned to repeat it. - George Santayana

Schewe

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 6066
    • http:www.schewephoto.com
Re: Camera Profiling
« Reply #1 on: July 02, 2014, 10:00:20 AM »

I can't see how only adjusting the white balance of an image would compensate for this ... but, hopefully I'm wrong.

It can and does...camera profiling (or sensor profiling) is profiling your camera's sensor to the way it responds to spectral illumination. "Daylight" which can run from cool to warm based on time of day is still essentially daylight. Compared to the spectral output of tungsten which has very, very little blue and lots and lots of yellow red, all daylights are essentially the same. Yes, not identical, but close enough for profiling. Other spectral output such as tungsten, LED or fluorescents have potentiality very different spectral output–which is why you need special profiles for those conditions.

Sadly, I think X-Rite oversold the need for multiple custom profiles in early marketing and people got confused between the different in sensor profiles and regular white balance adjustments (just my opinion).
Logged

howardm

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1827
Re: Camera Profiling
« Reply #2 on: July 02, 2014, 11:19:01 AM »

So is it more a matter of ratios of the tristimulous (sp?) values that each illuminant displays?

Vladimirovich

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1311
Re: Camera Profiling
« Reply #3 on: July 02, 2014, 11:35:14 AM »

I've played around a little with camera profiling and what certainly seems to be true is that I get more color fidelity if I use an ACR profile created using DNG Profiler or ColorChecker Passport rather than the Adobe Standard profile, irrespective of the outside light.  That makes sense to me because the profile I generate is for my camera whereas the Adobe profile is a generic (average?) one.

1) from Adobe Labs itself (as official as it can be nowadays)

Eric Chan :: ( http://www.luminous-landscape.com/forum/index.php?topic=62539.msg506044#msg506044 ) @ February 22, 2012 :

Quote
It's unlikely that a custom profile / .dcp profile - my note / will help you with one flavor of daylight more than another.  This is because natural daylight has a pretty full and reasonably smooth spectrum whose changes can be well approximated by per-channel scaling (i.e., simple white balance).  You'll see the benefit of additional custom profiles once you start working with light sources that are spectrally quite different (e.g., compact fluorescents).

2) when you use Adobe PE you always use a base profile... unless you take certain steps that will be one of Adobe's own profiles and so what you do really (in case of daylight) you just get rid of some of Adobe's own LUTs inside dcp profile that impose Adobe's (or rather a certain individual) vision about how color shall be rendered (but the main part of the color transform from camera's "color space" /which many do not consider as such because it does not satisfy certain definitions/ to CIEXYZ D50, etc will stay as Adobe did = http://www.adobe.com/content/dam/Adobe/en/products/photoshop/pdfs/dng_spec_1.4.0.0.pdf  and dcptool dump shall show you), that's it... and to do that you really don't need to shoot any target at all, you can just edit profile... now XRite OEM software or QPCard software might be genuinely creating a totally new dcp profile... but remember - you are going with 24/36 patches from un-individually-measured target vs (probably properly calibrated) monochomator with 5/10nm steps... so while you might like the rendering you shall not speak about "precision", so for most people creating their own profiles (under a variety of a daylight) using such targets it is just a way to replace what Adobe impose on them, that's it... no need to beat in your chest about your own profiling, till you will read most of what Eric Chan (and some others) posted here and some other forums
« Last Edit: July 02, 2014, 11:45:34 AM by Vladimirovich »
Logged

Robert Ardill

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 657
    • Images of Ireland
Re: Camera Profiling
« Reply #4 on: July 02, 2014, 03:37:24 PM »

1) from Adobe Labs itself (as official as it can be nowadays)

Eric Chan :: ( http://www.luminous-landscape.com/forum/index.php?topic=62539.msg506044#msg506044 ) @ February 22, 2012 :

My apologies ... I'm new to this forum and should have done a search before posting my question.

Having said that, if I do a profile on a cloudy noon day, for example, and another on a blue-sky noon day and white-balance both ... well there is a noticeable difference in some of the colors (especially the blues).  Presumably that's because the SPD is weaker in the blues on a cloudy day.  I haven't tried this yet, but I would expect to see a shift in the reds if the profile was done at sunset (compared to a noon profile).

BUT ... in general I agree that the differences are not significant enough to worry about.  It might be good to have three profiles: Early Morning, Noon and Late Evening ... but then there's the nuisance of managing the three profiles for what is probably a small and mostly (for landscapes at any rate) insignificant benefit.

Robert
Logged
Those who cannot remember the past are condemned to repeat it. - George Santayana

Tim Lookingbill

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2116
Re: Camera Profiling
« Reply #5 on: July 02, 2014, 05:17:14 PM »

For my particular camera (Pentax K100D DSLR) I ended up going back to the original Adobe supplied ACR 4.4 after going through the process of creating and using a dual and single illuminant profile.

But now I actually try out all four (including Adobe Standard) depending on the level of a combination of the dynamics/contrast and saturation level in the actual scene captured and "expose for highlights" exposure settings.

Adobe Standard does a REALLY good job taming saturation near highlights like in brightly lit yellow flowers with a tweak to white balance. But I've gotten to the level of post processing tool familiarity that I can apply all four and tweak the tools to get identical results or at least quite close to it.

My main hang up with all profiles is how to visually determine the level of yellow orange vs blue in fresh greens lit by the sun low in the sky should be rendered. That and the hue/sat of blue skies are the biggest differences I see between applying all 4 profiles.
Logged

Robert Ardill

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 657
    • Images of Ireland
Re: Camera Profiling
« Reply #6 on: July 04, 2014, 04:01:26 PM »

But now I actually try out all four (including Adobe Standard) depending on the level of a combination of the dynamics/contrast and saturation level in the actual scene captured and "expose for highlights" exposure settings.

My main hang up with all profiles is how to visually determine the level of yellow orange vs blue in fresh greens lit by the sun low in the sky should be rendered. That and the hue/sat of blue skies are the biggest differences I see between applying all 4 profiles.
Qualifying this with the fact that I have very limited experience of using my own profiles, I would tend to agree with you.  Looking at a few images I find that the sky blues look better with my profile. However I also find that the yellows and oranges and greens seem a bit over-saturated.

The Adobe Standard seems to handle highlights a bit better, but I think that's because it tones down the blue, which is where the clipping often happens.  In my case, my camera profile pops the blues and yellows compared to Adobe Standard, so highlight clipping is reduced in both these colors with Adobe Standard.  But then again, a small tweak of the saturation/ lightness in either of these colors levels things off, so I can't see it as being a problem/benefit in either case.

Either way I'm inevitably going to end up making adjustments in these colors, so I would think that consistency is the thing to aim for ... and for that Adobe Standard or own profile would seem equally valid choices.  The advantage of Adobe Standard for me is that it will work on any PC I might use without my having to bring my profile with me (or be forced to use DNG, which I'm still humming and hawing about!) ... the advantage of own profile is that it's nearer the 'truth' (at least when looking at the ColorChecker). On balance I'll probably go with own profile.

Robert
Logged
Those who cannot remember the past are condemned to repeat it. - George Santayana

digitaldog

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 12308
    • http://digitaldog.net/
Re: Camera Profiling
« Reply #7 on: July 04, 2014, 04:03:20 PM »

Looking at a few images I find that the sky blues look better with my profile.
That's the biggest improvement I see with my custom profile (for a 5DMII). Very visually significant improvements compared to Adobe Standard.
Logged
Andrew Rodney
Author “Color Management for Photographers”
http://digitaldog.net/

Tim Lookingbill

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2116
Re: Camera Profiling
« Reply #8 on: July 04, 2014, 05:57:31 PM »


The Adobe Standard seems to handle highlights a bit better, but I think that's because it tones down the blue, which is where the clipping often happens.  In my case, my camera profile pops the blues and yellows compared to Adobe Standard, so highlight clipping is reduced in both these colors with Adobe Standard.  But then again, a small tweak of the saturation/ lightness in either of these colors levels things off, so I can't see it as being a problem/benefit in either case.

Either way I'm inevitably going to end up making adjustments in these colors, so I would think that consistency is the thing to aim for ... and for that Adobe Standard or own profile would seem equally valid choices.  The advantage of Adobe Standard for me is that it will work on any PC I might use without my having to bring my profile with me (or be forced to use DNG, which I'm still humming and hawing about!) ... the advantage of own profile is that it's nearer the 'truth' (at least when looking at the ColorChecker). On balance I'll probably go with own profile.

Robert

Agree with you on relying mostly on the custom profile which fixes most of my well exposed shots of sunlit or any full spectrum lit scenes with plenty of light, but in a pinch when shooting unprepared or "too lazy" to bother with creating a custom profile for oddball spikey spectra low lighting I find Adobe Standard's strange color correcting DNA makeup works in a pinch.

It worked when I got a wild hair one weekend wanting to shoot some macro of popcorn lit by my LED flashlight to see how dramatic I could make it and the Adobe Standard worked better than using the dual illuminant profile in fixing the color.

Now I could've used the custom profile, but I'ld had to apply such extreme white balance adjustments and HSL tweaks that just by my familiarity with what Adobe Standard can do to color matrices of my Raw captures of similarly lit scenes, it's my first go to alternate option before going through the trouble of creating a custom profile for the LED flashlight. See the demo below which I've included an image where Adobe Standard messed up the hues.
« Last Edit: July 04, 2014, 06:13:01 PM by Tim Lookingbill »
Logged
Pages: [1]   Go Up