Color Casts for Some Commonly Remarked on Lenses

[charlie]

Actually, I posted the light in original post, good LED lights. And I used standard auto WB.

Auto white balance could be compensating for the color cast of any given lens which could skew your test. Since you are using a 5600K LED light, set your white balance in camera 5600k instead of auto. It is best to lock down white balance for tests, especially when dealing with color.

In case you are not familiar, adjusting white balance K temperature with live view on is a quick and easy way to see and set WB in ambient light situations.

[Michael Erlewine]

In case you are not familiar, adjusting white balance K temperature with live view on is a quick and easy way to see and set WB in ambient light situations.

I have set WB, but not in LiveView. How is the easy way to do that?

[Andy]

Michael,
you might be interested in the DoE Factsheet on LED color characteristics
http://apps1.eere.energy.gov/buildings/publications/pdfs/ssl/led-color-characteristics-factsheet.pdf

Thanks for sharing your pictures and thoughts. Honestly, I am not yet sure what the objective is, but this might be completely my error. The current setup has imho too many variables in force, to be able to conclude specific lens attributes.

A few observations and thoughts:
1) There is no lens with an "official" Nikkor lensid in the setup. Any potential color cast of the lens hits the sensor and the WB engine of the camera, without the camera being able to "know" what the lens particular color rendition is. In my observation, Nikkor lenses which are "known" to the respective Nikon D-SLR (AF, AFS, ...) expose much less color variation, vs. older MF Nikkor lenses (or Zeiss lenses). It is my assumption (no hard evidence supporting this observation), that the lensid not only make the camera aware of some optical lens particularities like distortion, but might very well also provide some info about the color rendition of the lens type. Allowing the CPU (Expeed) engine to include this information in the overall WB process. Lenses unknown to the camera (even by typing the focal length and aperture into the camera), don't uniquely define the lens type to the camera. So, the camera software has to "guess" what the color rendition of the lens in use actually is - leading to more variation.

2) If you keep the camera WB setting on Auto during these test shots, the WB engine will compensate what it think is the most appropriate compensation. My suggestion: Freeze the WB setting to any value you think is the most appropriate and then do the test with all lenses. Then you can use the patterns on the color checker to measure deviations among the lenses under test.

3) I've put a paper by the dep of energy on LED color characteristics. LED is imho a bad light source for color testing.

If you compare the energy density of different light sources, you might appreciate the impact the light source has on the overall color rendition

Daylight and moonlight


A bulb (thats why the blue channel is usually the problem child in white balancing these pictures - lots of noise in the blue channel due to much stronger amplification needed)


High pressure sodium lights are particular mean to WB attempts with RGB sensors - there is just no energy in the other channels


3) The limits of color calibrated monitors. We are probably "verifying" the output of your pictures on our monitors. They could be not calibrated at all, could be calibrated by software, or by simple HW (colorimeter). Or more sophisticated HW (i.e. a spectrophotometer with 32 bands) which allows to compensate for different color rendition of different panel types (IPS, TN, ...). Is the ambient light in the room where the calibrated monitor is being used, calibrated as well? or not? Most likely, we will not see the same same color rendition we could discuss between us.

4) Not sure I understand the value of whitebalancing the existing images. Any potential color rendition difference of the lenses will be overlayed by software based assumptions the respective SW package will apply. But I might have missed the thinking behind this approach.

Thanks again for sharing.

rgds,
Andy

[charlie]

I have set WB, but not in LiveView. How is the easy way to do that?

If you adjust the WB K temperature setting you can see the white balance change in realtime while viewing the LiveView screen.

By easy I meant that is a quick way to eyeball a neutral WB setting before you even take a picture, doesn't work with strobes, of course.

[Andrea B.]

I understand how to make the DNG profile and restart Lightroom. Then go to Camera Calibration, select the profile of the particular lens/camera operation, and look at it. NOW, after that, what do you do with it. If it is just "tweak" it, well that is just more subjective stuff. Can someone explain what to do with the special profile?

I'm not sure where the confusion lies? There's nothing mysterious really. 

For example, my purpose for making corrected colour profiles is to compare the original shot to a copy which has had the colour correction applied in order to to study & learn how a particular camera + lens + light + filter(if any) records colours. And also to attain reasonably accurate documentary colours for botanical work.

Keep in mind that setting an in-camera white balance or using auto WB with a particular camera + lens + light + filter is not always enough to provide correct colours (even though auto WB has improved immensely over time). If certain lenses or filters (polarizers come to mind) create colour casts, I want to know the direction and strength of such a colour cast (strong magenta? light cyan? blue desaturation? and so forth).

So you have to decide what you want from such a study. Does your camera + lens + light combo provide accurate colours without colour correction or not? Do you want to have accurate colours or do you want to keep a warm or cool colour cast for artistic reasons? If you want accurate, documentary colour, then you apply the colour-corrected profiles to any subsequent photos made with the profiled combination.

It is nice to know how a lens records. Also, with all the gazillion factors affecting colour perception, it is nice to at least be able to take one simple step towards accurate colour by creating colour corrected profiles. Production work for accurate colour goes many steps beyond this.

[I am assuming here that Lightroom works somewhat the same way as Photo Ninja with respect to creating these colour profiles. I do not know from my own experience. I hope what I'm saying is relevant to Lightroom.]

Added Later:  Even with colour profiling using CC Passport, I note that some colours remain extremely difficult to reproduce well due to the constraints of colour spaces (and monitor viewing or printing, etc). I don't think I have ever been able to photograph violet or violet-blue flowers with complete accuracy. Some internal filters cut far too much violet. Even using a camera with the internal filtration removed, I still have trouble with violet colours.

[Andrea B.]

Andy, points well made. But colour profiling is all we have for a first step. I stop there and don't worry about ambient light and other factors because my documentary work is not currently being reproduced in printed form to be viewed in specific kinds of light. (oh well......)

[Frank Fremerey]

In addition to Andy's remarks I found an interesting piece to coonsiderate: http://www.redsharknews.com/production/item/2627-redshark-update-sekonic-c700r-and-measuring-colour-temperature-of-practical-lights

[Michael Erlewine]

The light I use for this kind of work is LED with a CRI  of 95. That is supposed to the same an incandescent. Perhaps someone reading this has more information.

[pluton]

The light I use for this kind of work is LED with a CRI  of 95. That is supposed to the same an incandescent. Perhaps someone reading this has more information.

it has been my experience that many LED vendors offer claims for the color "accuracy" of their units that don't test out in actual use. 
It'd be interesting, but not essential, to know the actual make and model of the LED unit involved.
By the way, CRI is a number that defines human perception of color.  It is not a photographic specification.

[Roland Vink]

4) Not sure I understand the value of whitebalancing the existing images. Any potential color rendition difference of the lenses will be overlayed by software based assumptions the respective SW package will apply. But I might have missed the thinking behind this approach.

Good points Andy. I'd just like to expand on this last point since balance is affected by many factors:

1) The color of the subject being photographed - is it yellow, green, blue, warm, cold?

2) The color of ambient light falling on the subject - is it early morning light, mid-day, cloudy, sunny, tungsten lighting, LED, sodium, fluorescent ...?

3) Glass used in the lens (including filters) - some glasses transmit or absorb more strongly in some parts of the spectrum than others, which is why some lenses are warmer than others.

4) Lens coatings (including filters) - color balance of lenses can be affected by the coatings used. Nikon was well known for it's NIC (Nikon Integrated Coating) which integrated the coatings into the overall design, so there was a more consistent color balance across the entire lens lineup.

5) Filter stack in front of the sensor (points 3 and 4 above)

6) Spectral sensitivity of the sensor - Most sensors use Bayor RGRB pixel array, but the RGB filters used in one camera is not guaranteed to be the same as another camera. Eg one green filter may have higher efficiency in the green part of the spectrum, or greater transmission near the blue or red spectra.

7) White balance applied by the camera to the image.

8 ) White balance applied by any editing software.

9) White balance of the final viewing media - whether it is a monitor/profiling, or the paper/inks/dyes/pigments on a print.

10) Ambient light falling on the image (see point 2 above)

11) Color balance of glasses or contacts you wear (see points 3 and 4)

12) Color balance of your eyes - in my case one eye sees distinctly redder than the other!

13) How your brain interprets color, personal preference/bias - do you prefer green over blue, warm over cold...

The point I'm making is that there are so many factors which affect color balance, that I wonder about the the value of this exercise. The images in the original post are all quite acceptable in terms of color (if anything they all look a bit yellow/warm to me). The color casts of some lenses are warmer than others (all other factors being equal), but they are just one small factor in many, and none are "right" or "wrong". If any are not to your preference it is a simple job to adjust the color balance to your liking...

[Frank Fremerey]

The light I use for this kind of work is LED with a CRI  of 95. That is supposed to the same an incandescent. Perhaps someone reading this has more information.

A spectral response curve actually measured by some trustworthy third party is required to judge the quality of light. Look at the link I posted above for example.

Worst light has strong peaks and large holes between the peaks like mecury vapour in the extreme and some monochromatic LEDs conmbined to a "white" source. 3 or 4 peaks do never make quality light.

Converter bulbs utilizing 5 or 6 bands smeared out by heat ... read: switch on 20 minutes before starting to shoot ... might have peaks, but they offer a significant amount of spectral response over the whole spectrum.

This allows very natural colors of skin and food e.g.

BUT: All of these sources are relatively dull. If you can deal with the 54,000 lumen of full spectrum you get from 1,000 Watts of power intake, you are mostly in the non-moving objects/subjects domain.

For models jumping through the studio you need serious studio flash batteries.


https://www.youtube.com/watch?v=vYwcKyBarzo

http://download.p4c.philips.com/lfb/f/fp-928045295081/fp-928045295081_pss_de_de_001.pdf

same goes for Osram Dulux L 954

[Erik Lund]

I agree that most likely the LED lights are a major factor in the issues you see Michael.

I recall the WB issues I saw when I started shooting the Nikon D1 - Yes the camera had a magenta cast in the firmware but the artificial and ambient light made me aware of digital issues.

Back then I used sometimes all of my lights, the images where unusable, I asked around and ended up with Elinchom, they are accurate both in color and power output and there is a huge amount of light modifiers available. I have a mix of battery powered and mono lights, no color issues at all.
I don't like their 'mount' but that's the only drawback.

Andrea - Super passport tutorial!!! Thanks

I use it from time to time,,, fabrics can be so tiresome,,, 

[simsurace]

I agree with the observations regarding the LED light sources. A continuous light source is recommended. You can only measure light transmission in the parts of the spectrum in which you have light.

Michael, you could make a DNG profile with Adobe's DNG profile editor. You can then study and compare the color tables of different lenses to see which colors have a cast (relatively speaking, as you don't get the spectral transmission curve, but some info that relates to the color spaces and the corrections which are done within that space). It is important to lock down WB because the auto-correction is not always predictable. With Auto WB, it may very well be that lens color casts are being exaggerated.

I wonder whether one could in principle estimate the spectral transmission of a given lens from the color profiling data obtained from an image shot under a continuous light source with known spectrum...

[pluton]

I agree that the use of Auto WB is, sometimes, asking for trouble.  I shoot almost everything at daylight WB.  The camera then acts as a reference to the light source, and informs me how the light source deviates from the theoretical perfect case.