Pages: [1] 2 3 4   Go Down

Author Topic: Dual Illuminant Profiles  (Read 77149 times)

BobD

  • Newbie
  • *
  • Offline Offline
  • Posts: 42
Dual Illuminant Profiles
« on: May 19, 2010, 09:44:42 am »

Trying to understand how ACR uses a dual-illuminant camera profile?   According to X-rite, “when you select a dual illuminant profile, your Adobe Raw software will use both light source tables to adapt the profile to match the image’s illuminant.

My specific questions are:
1)  How does ACR “adapt” a “Dual illuminant” [D65 and 2850] camera profile?  (e.g. for an image shot on a cloudy day)

2)  What are the benefits (and repeatability*) of using a “Dual illuminant” camera profile?
.........2a) *repeatability - if there is a subtle change in color temp. (sun moving in and out of clouds), will a common object have a "color shift" because of “adapting” dual-illuminant LUTs?

3) Is it better to use a single-illuminant profile that is closest to the actual shooting color temperature?

4) How critical are the color temperatures (D65 and 2850) illuminating the 24 patch Graytag-MacBeth target when creating files for a Dual-Illuminate camera profile?

Thanks
Logged

jpegman

  • Full Member
  • ***
  • Offline Offline
  • Posts: 129
Dual Illuminant Profiles
« Reply #1 on: May 19, 2010, 01:31:24 pm »

I have the same question - apparently, Photoshop ACR started this dual profile methology and it has filtered down to X-Rites Passport Profiler product, with little explanation on when it is recommended or not.

Eric, do you know the history and theory behind this dual point approach and situations where it might be better to use a single point calibration?

Jpegman
Logged

madmanchan

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2115
    • Web
Dual Illuminant Profiles
« Reply #2 on: May 19, 2010, 01:31:28 pm »

1. Interpolates between the two tables based on your white balance setting. Specific method is inverse correlated color temperature. See DNG Specification if you wish to see the details & mathematics.

2. Improved color reproduction over a wider range of scenes. Degree of improvement depends on the camera model. If the scene lighting conditions (e.g., office fluorescent) varies greatly from the two illuminants used to build the profile (e.g., Solux bulb and natural cloudy daylight) then all bets are off. You will get something "ok" but results will be better under the unusual lighting condition by building a profile for that condition.

3. It's not the shooting temperature that matters, but the spectrum of the illumination. This is more complex and not easily measurable. General advice: if you tend to photograph under daylight, not worth building profile for each flavor of daylight, regardless of the color temp. There will be minor variations between ~D50 and ~D75 lighting (roughly 5000 K and 7500 K CCT) but this tends to get eliminated once WB is considered. If you photograph under unusual artifical lighting often, consider building a profile for that lighting.

4. Not too important. X-Rite's software will tag the profile appropriately.
Logged
Eric Chan

BobD

  • Newbie
  • *
  • Offline Offline
  • Posts: 42
Dual Illuminant Profiles
« Reply #3 on: May 19, 2010, 04:08:22 pm »

Quote from: madmanchan
#1. Interpolates between the two tables based on your white balance setting.

#3. It's not the shooting temperature that matters, but the spectrum of the illumination.
Eric thanks for the info... I have a few more questions.  

In #1 when you say ACR “Interpolates between the two tables based on your white balance setting.”...
- What WB setting are you referring to: the WB of the scene at time of shooting; the WB set on the camera when shooting the raw file or the WB set in ACR?  I was under the assumption that the camera's WB setting was a suggestion not part of the raw data but used only as a starting point if the ACR WB default was set to “As Shot”.

In #3 when you say “It's not the shooting temperature that matters, but the spectrum of the illumination."...
- Are you talking in wavelength or in Color Rendering Index?

Thanks Again,
Bob DiNatale
Logged

madmanchan

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2115
    • Web
Dual Illuminant Profiles
« Reply #4 on: May 19, 2010, 04:59:38 pm »

Hi Bob,

For #1: the WB set in ACR.  Initially, this is the "As Shot" WB which is usually the camera's suggestion, as you say (though it could be a fixed WB setting if you chose a manual WB using the in-camera settings).  As an example, suppose you created a profile for 2 illuminants, approximately A (2856 K) and D65 (about 6500 K). If you drag the temperature slider in ACR towards 6500 K, you'll end up using more of the color matrix and color table for the D65 "half" of the profile, and if you drag the temperature slider in ACR towards 2856 K, you'll end up using the color matrix & table for the A "half" of the profile. And if you're somewhere in between, like 5000 K, you'll use a mix of both.

For #3:  The wavelength. There are infinitely many lights that map to the same correlated color temperature (e.g., 5500 K). In practice, for example, you may have a flavor of natural daylight (some mix of sun & clouds) that gives you 5500 K CCT. You can also have fluorescent tubes that will get you about 5500 K, too. But the color rendering produced by the two lights can (and often will) be quite different, for certain materials. This poses a problem if you're trying to use a color profile made for one illuminant (e.g., the daylight one) for images shot under the other (e.g., the fluorescent one). In short, the CCT (or even a temp/tint pair) is not enough to describe the illumination. Having a spectrum of the overall ambient illumination (i.e., relative power at each wavelength) would help a lot, but this information is generally not easily available.
Logged
Eric Chan

Tim Lookingbill

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2436
Dual Illuminant Profiles
« Reply #5 on: May 26, 2010, 12:37:33 pm »

Here's a comparison showing the difference between using a single and dual illuminant table profiling my Pentax K100D using Adobe's DNG Profile Editor Wizard.

•The first image on the left is my eyeballing the actual appearance of the incandescent lights I used.

•The center As Shot WB image is the one used to build the second 2800K table in DNG Profile Editor CCchart Wizard. Its WB was established using the incamera "Incandescent" WB setting whose cast appearance ACR's As Shot interprets as shown.

•The third version on the right is neutralized using ACR's WB eyedropper tool on the gray patch next to white.

As you can see what a dual table appears to do is maintain the spectral reflectance characteristics (hue/saturation/luminosity) of color relationships no matter what the WB and spectral qualities of light used as long as the color wavelengths of the lights used have no abrupt spikes. From what I've observed the embedded algorithms seem to concentrate on correcting for red/green biases between 2800K-6500K influences.

The red patch in the single 6500K illuminant profile is way too orange and desaturated. The dual table version below looks more accurate though a bit over saturated with a slight magenta bias in the red which I believe is the happy medium between the two illuminants. My camera's AWB also tends to deliver a red bias balancing outdoor daylight.

I've also included the original daylight target image used to build the 6500K table. It has the single illuminant table profile applied which shows a correct looking red compared to the incandescent image.


[attachment=22208:DualSing...DifferLL.jpg]

[attachment=22209:SingleTa...ylightLL.jpg]
Logged

BobD

  • Newbie
  • *
  • Offline Offline
  • Posts: 42
Dual Illuminant Profiles
« Reply #6 on: May 26, 2010, 02:50:06 pm »

Quote from: tlooknbill
…comparison showing the difference between using a single and dual illuminant table profiling my Pentax K100D using Adobe's DNG Profile Editor Wizard.

...As you can see what a dual table appears to do is maintain the spectral reflectance characteristics (hue/saturation/luminosity) of color relationships no matter what the WB and spectral qualities of light used as long as the color wavelengths of the lights used have no abrupt spikes. From what I've observed the embedded algorithms seem to concentrate on correcting for red/green biases between 2800K-6500K influences.

...The red patch in the single 6500K illuminant profile is way too orange and desaturated…
tlook... thanks for the effort

Do you think a single illuminant profile has 2 (dual) daylight LUTs or do you think there is a void in the 2nd LUT area?  I would tend to think that it has 2 of the same “single” illuminant LUT. (Eric?)

Eric says: “If you drag the temperature slider in ACR towards 6500 K, you'll end up using more of the color matrix and color table for the D65 "half" of the profile, and if you drag the temperature slider in ACR towards 2856 K”.  Posing the question, “what would happen if you dragged the slider towards the second illuminant area and there was a void or no LUT?

... “(your) camera's AWB also tends to deliver a red bias balancing outdoor daylight.”
If you are shooting in raw, I would think your camera’s “red bias” is a function of the chip’s bias and not the AWB.  Isn't that why we create a profile in the first place… to remove the camera bias?  Maybe your "…slight magenta bias in the red” is more accurate.  I would hope it's more accurate because if “you drag the temperature slider in ACR towards 2856 K” and it creates a “slightly magenta biased” then I would have a concern that the color, although unique to the dual illuminant profile, would not be accurate.

..."…dual table (profiles)… maintain the spectral reflectance characteristics…no matter what the WB and spectral qualities of light used as long as the color wavelengths of the lights used have no abrupt spikes."
Again, it is my understanding that the camera’s WB has no effect on the profile when shooting a raw file.  The quality of light not having any “abrupt spikes” can be determined by the “Color Rendering Index” (CRI) specification of the light source.
Logged

Tim Lookingbill

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2436
Dual Illuminant Profiles
« Reply #7 on: May 26, 2010, 05:31:29 pm »

All I can tell is if I slide the color temp slider on the Daylight CCchart target image with the single illuminant profile from default As Shot 5100K to 7500K the Hue/Saturation/Luminance relationship between warm and cool colors change proportionally whether I use a dual or single table profile.

If I slide the color temp to 4500K with the single table profile decreasing overall luminance and saturation, changing to a dual table afterward brightened certain colors mainly in the pinks and magenta when it didn't do this at 7500K.

Now color behaves differently doing this on the incandescent shot, so the light source illuminating the initial image is going to have far more influence over whether you use a dual or single table profile. But in my experience a dual table works best with scenes lit under incandescent lighting because there is far more improvement in maintaining HSL when neutralizing or adjusting to taste this type of color cast.

If all you shoot is under daylight such as in landscapes then a single table is all that's necessary.
Logged

BobD

  • Newbie
  • *
  • Offline Offline
  • Posts: 42
Dual Illuminant Profiles
« Reply #8 on: June 07, 2010, 01:33:29 pm »

Quote from: madmanchan
...As an example, suppose you created a profile for 2 illuminants, approximately A (2856 K) and D65 (about 6500 K). If you drag the temperature slider in ACR towards 6500 K, you'll end up using more of the color matrix and color table for the D65 "half" of the profile, and if you drag the temperature slider in ACR towards 2856 K, you'll end up using the color matrix & table for the A "half" of the profile. And if you're somewhere in between, like 5000 K, you'll use a mix of both.
Eric, recently I have been creating dual illuminant profiles using Xrite's ColorChecker Passport software.  The dual illuminants are Daylight and Tungsten A.  However, sometimes these illuminates fall into different LUT tables e.g. “Daylight" resides in Illuminant 1 & "Tungsten A" in Illuminant 2 AND sometime visa versa (see inset)  


Previously you mentioned “…dragging the temperature slider towards one or the other (Illuminant) will use more of that color matrix and color table, …if you're somewhere in between, …you'll use a mix of both.”  That being said, here are my questions:
...1 Does it make a difference which Illuminant goes into which look-up table?
.............i.e.  “should the Daylight Illuminant always in Illuminant 1 LUT?”
...2  What happens when there is only a single illuminant profile, what happens when you drag the slider towards the “void” color matrix and color table?
...3 Does the “Adobe Standard” delivered with Lightroom and ACR have “dual illuminants”?
Bob DiNatale
Logged

mac_paolo

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 431
Re: Dual Illuminant Profiles
« Reply #9 on: May 08, 2012, 03:42:09 am »

I restart this topic with a simple question.
Taken into account that the temperature alone means a little as it doesn't specify nothing about the actual spectrum, how do we chose the hotter temperature for the dual illuminant?

Is it D65 or 6500 K?
Is it better a cloudy day or the specific hour of the afternoon when the sun produces 6500 K light?

The colder illuminant is easier to achieve with a tungsten bulb :)
Logged

Tim Lookingbill

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2436
Re: Dual Illuminant Profiles
« Reply #10 on: May 10, 2012, 01:16:19 am »

I restart this topic with a simple question.
Taken into account that the temperature alone means a little as it doesn't specify nothing about the actual spectrum, how do we chose the hotter temperature for the dual illuminant?

Is it D65 or 6500 K?
Is it better a cloudy day or the specific hour of the afternoon when the sun produces 6500 K light?

The colder illuminant is easier to achieve with a tungsten bulb :)

I think you're making this too complicated.

If you note in the image samples I posted above I established D65/6500K illuminant table from the CC chart lit by noon day sun (close enough to establish R=G=B neutrality). The important thing is you need to use a light source with the fullest color spectrum and you can't get any closer than direct sunlight. All lights including natural and artificial (and overcast) will impose various levels of color errors within a certain range of memory colors with sunlight showing the least color shifts over a wider range of colors compared to other light sources.

The tungsten lit shot was to show the difference between what happens to certain colors (in particular the red patch) neutralizing a tungsten lit scene with a single verses a dual illuminant profile.

Did you notice the slightly less saturated orangish red patch of the single D65/6500K illuminant profile on the top far right compared to the proper red patch appearance using the dual illuminant profile below it after applying the same color temp edit to both?

So the issue isn't really about establishing exact appearance of color temperature, which can become an interpretive experience editing an image due to the adaptive nature of human vision, but it's more about coming up with two different types of profiles that influence color appearance when neutralizing, editing or adjusting ACR/LR's Color Temp sliders using the gray reference shot under completely different illuminants for the best possible color balance appearance.

This is the true goal behind camera profiles, not accuracy to some color temp appearance associated with a Kelvin number. If you stared at any one of those color patches for at least a minute, I can guarantee the appearance of the gray patch used as neutral reference will noticeably change hue even if it read R=G=B. So much for accuracy to some color temp appearance. Accuracy to a balanced appearance? Yes.

Read up on the subject of "color constancy" and what it does to the perception of color balance associated with color temperature.
« Last Edit: May 10, 2012, 01:18:34 am by tlooknbill »
Logged

digitaldog

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 20630
  • Andrew Rodney
    • http://www.digitaldog.net/
Re: Dual Illuminant Profiles
« Reply #11 on: May 10, 2012, 12:35:39 pm »

Is it D65 or 6500 K?

If you want to get real picky...

6500K is a range of colors. D65 was developed using actual color measurements but a lot of em (622) over the course of time in differing locations around the world with differing instruments. So the likelihood you’re exactly getting D65 is questionable.
Logged
http://www.digitaldog.net/
Author "Color Management for Photographers".

Tim Lookingbill

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2436
Re: Dual Illuminant Profiles
« Reply #12 on: May 10, 2012, 03:10:57 pm »

Examine the various appearances of so called 5000K and 6500K on this site's review of desktop/print viewing stations:

http://www.northlight-images.co.uk/reviews/lighting/viewing-pdv3.html
Logged

mac_paolo

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 431
Re: Dual Illuminant Profiles
« Reply #13 on: May 10, 2012, 04:07:52 pm »

So why explicitly ask for a 6500 K white when a plain common daylight is most probably the best choice?
6500 K is quite uncommon for daylight here and there.
Logged

Tim Lookingbill

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2436
Re: Dual Illuminant Profiles
« Reply #14 on: May 11, 2012, 06:10:08 am »

So why explicitly ask for a 6500 K white when a plain common daylight is most probably the best choice?
6500 K is quite uncommon for daylight here and there.

How do you know what color is 6500K daylight? As Andrew pointed out the actual appearance of a spectrally flat white/gray target lit by sunlight can vary depending what area of the globe one resides.

It's just a number that color management software algorithms reference against to compare to their core connection space, Lab D50, which again is an artificially mathematical construct of how humans perceive color under a full spectrum light source. They need the number for mathematical calculation for controlling color appearance assuming a neutral looking viewing display environment.

The appearance of 6500K white on a display can be compared to white formless overcast clouds backlit by noon sun, but again a display is not a full spectrum light source. That particular hue of 6500K display white is only mimicking full spectrum light by tweaking hue/saturation color tables of tagged images (color constancy optical effect) that gives the impression the person is viewing a daylight environment. It's still artificial but it manages to convince us otherwise due to the adaptive nature of human vision.
Logged

mac_paolo

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 431
Re: Dual Illuminant Profiles
« Reply #15 on: May 11, 2012, 07:04:46 am »

How do you know what color is 6500K daylight? As Andrew pointed out the actual appearance of a spectrally flat white/gray target lit by sunlight can vary depending what area of the globe one resides.
I just shoot a WhiBal under direct sunlight, which should mean full spectrum, and measure it back into Lightroom.
I'll try to explain better my point: as for the most part of the globe the direct sunlight is always way below 6500 K, why did Adobe/others explicitly ask for 6500 K instead of 5000/5500 K?

I mean, I know the difference is not that huge, it's more just curiosity. :)
Logged

Tim Lookingbill

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2436
Re: Dual Illuminant Profiles
« Reply #16 on: May 11, 2012, 07:47:08 am »

You'ld have to ask Adobe. Like I said it's just a number used for mathematical calculation for controlling color appearance in the artificially neutral looking editing environment of the display.

Below is an image I shot of some food items lit by a regular 100 watt GE soft white tungsten light bulb as an example of how tweaks to WB hue in relation to color table hue/sat tweaks can change the perception of the appearance of Kelvin based defined daylight.

It's not just about color cast. It's about mimicking the spectral reflectance effects of a full spectrum light source along with the color cast/color temp appearance and what it does to our perception of color balance and why focusing on the definition of a Kelvin/D number is pretty much pointless.

The three images are pretty much self explanatory with the last one pushing neutrals to an unnatural hue that compliment overall appearance of the red tomato, the fresh green lettuce leaf and the toasty warm hue of the cracker. It required more than just using a dual illuminant profile. The first one is VERY ACCURATE to the scene but looks butt ugly.

The last one is the most freshest and desirable looking but very inaccurate and way off any natural appearance of daylight whether Kelvin or D65/50. The hue/sat tweaks were applied in ACR's Hue/Sat panel.
« Last Edit: May 11, 2012, 07:52:32 am by tlooknbill »
Logged

digitaldog

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 20630
  • Andrew Rodney
    • http://www.digitaldog.net/
Re: Dual Illuminant Profiles
« Reply #17 on: May 11, 2012, 09:38:12 am »

I'll try to explain better my point: as for the most part of the globe the direct sunlight is always way below 6500 K, why did Adobe/others explicitly ask for 6500 K instead of 5000/5500 K?

The numbers are kind of meaningless in this context once you understand a boat load of colors correlate to any single kelvin value.

See: http://www.ppmag.com/reviews/200512_rodneycm.pdf

See the lines of correlated color temp that run magenta to green? Any color on that line could be specified as 6500K.
Logged
http://www.digitaldog.net/
Author "Color Management for Photographers".

bjanes

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 3387
Re: Dual Illuminant Profiles
« Reply #18 on: May 11, 2012, 10:24:34 am »

I just shoot a WhiBal under direct sunlight, which should mean full spectrum, and measure it back into Lightroom.
I'll try to explain better my point: as for the most part of the globe the direct sunlight is always way below 6500 K, why did Adobe/others explicitly ask for 6500 K instead of 5000/5500 K?

There is an important difference between sunlight and daylight, and this is well explained in an article on the Handprint web site (scroll down to the section Variations in Natural Light). Sunlight comes directly from the sun and can be visualized by a shaft of sunlight entering a darkened room through a window or skylight. The aperture of the window obstructs skylight (light diffused by the sky and having a bluish color). Daylight is the combination of sunlight and skylight and is what you would measure by taking a WhiBal reading. Daylight sometimes takes into account light reflected from foliage or buildings.

Noon sunlight is approximately 5500K and noon daylight about 6500K, but varies with latitude and atmospheric conditions.

Regards,

Bill
Logged

julianv

  • Jr. Member
  • **
  • Offline Offline
  • Posts: 55
Re: Dual Illuminant Profiles
« Reply #19 on: May 13, 2012, 07:08:21 am »

There is an important difference between sunlight and daylight, and this is well explained in an article on the Handprint web site (scroll down to the section Variations in Natural Light). Sunlight comes directly from the sun and can be visualized by a shaft of sunlight entering a darkened room through a window or skylight. The aperture of the window obstructs skylight (light diffused by the sky and having a bluish color). Daylight is the combination of sunlight and skylight and is what you would measure by taking a WhiBal reading. Daylight sometimes takes into account light reflected from foliage or buildings.

Noon sunlight is approximately 5500K and noon daylight about 6500K, but varies with latitude and atmospheric conditions.

I read the relevant text on the Handprint web site, and I'm afraid that I am still a bit unclear on the distinctions used when referring to daylight, sunlight, and skylight.  I do understand that daytime SPD measurements vary, depending on how much light is received directly from the sun, how much is received via Rayleigh scattering (from a blue sky), how much is filtered through clouds, dust, and pollution, etc.

The part that I don't understand is the advice given for photographing the ColorChecker.  Certainly, the ideal method is to shoot it in the same light that falls on the subject, and to build a profile specific for the scene conditions.  But if I want to build a general purpose camera profile for daytime outdoor conditions, I need to shoot under sunlight.  And that's where I have been reading conflicting advice.  Some sources advise shooting in direct noon sunlight.  Others recommend "open shade", but do not explain what that means.  I have also read comments that the exact conditions do not matter much.

I interpreted "open shade" as follows.  I shot the CC in a location that was shielded from direct sunlight, and only received light from a cloudless mid-day sky.  I tried to minimize exposure to reflected light from other sources, such as foliage.  Is this a proper way to shoot the target for a general purpose single-illuminant profile?

I used the CC Passport software to build custom "open shade" profiles for a D700 and D800.  I processed raw files with ACR and PS CS5, viewing on a calibrated NEC PA271W.  On the D700, the results from using the profile were mixed.  In many cases my shots came out with reds, blues and purples that appear over-saturated.  On the D800 the custom profile gave better results - almost indistinguishable from the Adobe Standard profile for that camera.

Here's an interesting observation: if I load the "open shade" shot of the CC that I made with the D800 and measure the white balance of the lightest gray patch (with the WB eyedropper tool in ACR), I get about 6700K.  If I measure the white patch I get exactly 6500K (and 245/245/245).
Logged
Pages: [1] 2 3 4   Go Up