White balance effects RAW files

Taz

Active Member
I have been shooting RAW with auto white balance thinking that it made no difference since I could post process the errors out.
This is WRONG

Shoot some pictures under strange lights, mercury vapor or florescent at higher ISO and you will see a big difference. With auto white balance the shadows will have a green cast to them. There is no way to remove the green cast without totally messing up the flesh tones. I have had some success exporting TIFF files and working with them in Photoshop using the gamma correction on the green channel. Nothing works as well as doing a custom white balance before shooting.

The camera does something more that just tag the file with balance info. Now I am a happy man with no more green shadows.
 

tc95

Well-Known Member
Taz....I would have to agree..anything that you want to keep..that is outside the realm of normal shoothing...I always start with a custom WB....

I am sure this post will help out others...too...

Tony C. :z04_cowboy:
 

jesito

Well-Known Member
Thanks for pointing it out!.
I sold my SD14 because that nasty green cast...
If I had known it, maybe I wouldn't have get rid of it.
(I still keep the SD9 and SD10).

Regards,
Jes.
 

Taz

Active Member
I have attached some examples of effects of auto white balance and manual. This is not the worst case since the ISO is only 200. The first shot is auto white balance. The gray card shows to be gray but there is green cast to the wood under the green bench. Not much at this ISO but it is there.


This shot is with custom WB off the gray card. The green cast in the wood and brick steps is gone.


When the light source is uniform in spectrum the effect is not too bad. If the light souce is from a gas discharge or other non hot body source the effect can become quite bad. I have no explanation for the effect.
The images were developed with SPP4 and the color picker was used on the gray card to make things look as close to the same as possible.
 

Taz

Active Member
The pictures were taken using an old Russian Industar 61 l/z lens. It's not the sharpest lens in the world but I really like it.
 

Bustoner

Well-Known Member
Wow Taz
thanks for advice :)
Now I will use it more and more......

I'm really curios to try an Industar
 

Arvo

Well-Known Member
Custom WB and color wheel corrections are not directly comparable. Setting custom WB alters white point value of color conversion process (thus changing entire decoding matrix), color wheel just adds some tint to image.

You can achieve needed corrections in most raw software, allowing to adjust white point - like Silkypix allows to change both color temperature and green-magenta deflection (the latest even differently for dark areas, which is very handy for SD14 green cast). Sure many other applications allow change white point properly too, ArcSoft products are very comfortable for such corrections for example. SPP unfortunately cannot do that.

Using Custom WB is usually the best option.

Industar 61 has small back lens diameter (or rather short distance from sensor to exit pupil, which in practise gives about the same result) and that creates green corners/magenta core (most probably thanks to Sigma IR filter). This can be removed in postprocessing (apply some almost trasparent radial magenta gradient layer) - but that is not very easy; all aperture and focus distance settings need different processing. Your images show these green corners clearly.

If you can find cheap SMC Takumar 55/1.8 or Chinon 55/1.7, I recommend to buy one - both are sharp lens and do not have green corners effect. I've got both, Takumar is not SMC though. There are alot of similar design lens available (another good one is for example Fujinon EBC 55/1.8), I've just no first-hand eperience with them. I've got Industar 61 too (from old Zenith times) and due to the aforementioned problems I don't use it with Sigma. My copy of it is very sharp though.
 

Bustoner

Well-Known Member
Thanks Arvo, very useful
For info I was try serching an image of the contruction elements of the Industar 61, but nothing.
I understand soo, if the back element of a lens is next to the IR/UV protection filter, there's a risk of internal reflection (on corners when fully open), can this be less visible on different shapes of back elements lenses? I mean if the last element ends in Concave shape than like most are Convex, can make difference...of backward reflections?

A "maybe" idea, to resolve this problem with some good lenses, by removing the Sigma IR/UV filter, and find the same filter somewere to screw it in front of the lenses ....will clear out the problem...? :)

Please, Log in or Register to view URLs content!
 

Arvo

Well-Known Member
Green corners are probably *not* a result of internal reflections.

I'm pretty sure that problem is in multicoated IR filter (hot-mirror). MC filters spectral characteristics depend on incidence angle of falling light ray; looking at SD14 IR filter you can see that it reflects magenta, when looked at angle - thus it makes such rays green after filter.
Some additional effect can be attributed to sensor itself - its spectral response does depend on light falling angle too, although such effect is much smaller than filter caused one.

If exit pupil (virtual image of aperture ring) of lens is near sensor (and aperture is small), then rays, falling to border areas of sensor, are slanted and thereby greener. If exit pupil is further away (and aperture is open), then most rays fall almost perpendicular to sensor and no color shift is seen.

Using external UV/IR filter is problematic. If such filter is multicoated (often it is), then corners become much more green that currently - angle of incidence for these rays is bigger.
If external filter is not MC, but some other variety, then green corners are not an issue - but you have hard times to replicate Sigma filter spectral characteristics. On current forum Steaphany has experimented with various filters, look at one one of her threads:

Please, Log in or Register to view URLs content!


What about Industar 61 optic scheme, then there are some sources on the net; for finding information it is better limit searches to .ru domain.

Please, Log in or Register to view URLs content!
 
The green cast picture

Taz,

Would you still have the RAW file for the photo with the green cast? If yes, could you make it available for download?

There is something I would like to test with SPP 4. Otherwise, I will wait to get one myself.

Thanks
Jeff
 
No more red cast

Since this post from Taz, I have started to do custom white balancing before shooting. Previously, I was having a lot of trouble with red cast in the shades on high dynamic photos with my SD14.

With custom white balancing, the problem is not totally gone, but so far, it is much, much better! Since I do custom white balancing with a piece of tissue, maybe this explains the remaining red cast in the shadows.

I'm planning to buy a white balancing lens to see if I can get rid of this annoying red cast. Any suggestion on brands? Or another tool? Thanks.

Jeff
 

ombud

Member
non-continuus color spectrum

Gas discharge lights have a non continuus color spectrum meaning that the
color uniformity is constantly changing emitting more or less green in an instant. Thats why even a custom WB might produce no even balanced pictures among a series under gas discharge light units (e.g.street lights at night)
I use white balance tools like the SPYDERCUBE which helps in several ways:
you get a medium gray 18% (as a reference)
you get a highlight part to measure (brightest parts that still hold details)
you get a lowlight part to measure (darkest parts that still hold details)
you get an absolute dark point (as a reference)
...and you can judge the light units from the chrome ball on top of the cube

just my 2cents...
 
Top