[panoguy]

I was shooting at a karate tournament this weekend with my new K-3 and the DA*50-135 and struggled with the WB the whole time. When chimping the images on the back LCD, I saw green or red color casts coming in from the sides of the image, but not covering the whole frame. These strange color gradients were later verified in both Adobe Camera Raw 8.1 (via DNG) and the Ricoh DCU5 software that came with the K-3.

At first I thought it was all the (very white) uniforms messing with "Multi-Auto WB" so I switched to "Cool White Fluorescent" since it seemed to render the best skin tones, and at least it should be consistent across the whole frame. However, the gradient color casts continued! Afraid there was something wrong with my new camera, I went outside and quickly shot a grey stone wall, and hooray, no gradient color cast.

It dawned on me that the big industrial lights (totally unsure what type) in the big tournament room were probably to blame so I kept shooting, hoping that I could fix it in post. Well, it's as bad as I thought at the time, and I have no idea what kind of light would do this to my shots. There were no spotlights or even sidelights where the people were competing, only a few big lights way up above.

EXAMPLES - here is a kid moving slowly (he's doing a kata) across a very small distance under this lighting. These shots all have the same WB set in ACR, and the same exposure settings (1/250 at f/3.5 on ISO1600).







**This last shot is what I was seeing when I set the WB at the tournament, and now I'm going through 740 shots and manually adjusting the WB in shots that are "keepers" and for the really nice shots, using a gradient in ACR to correct for the off-colors.

I just want to know if this is something anyone has ever seen before under indoor artificial lighting that appears consistently white to the eye.

---

Last edited by panoguy; 11-17-2013 at 08:32 AM.

[lister6520]

If you copy the WB of the one you fixed manually and multiple paste it onto all the others do you still get the wrong colour balance?

I would think that given the same lighting in all the photos then the same WB correction should be applicable to all and not require to manually correct each one individually.

[VisualDarkness]

Sounds like some kind of fluorescent lamps that pulsates with short cycles. Sometimes they show a heavy color cast at different times in the cycles and if you're unlucky, shooting with fast shutter speeds, they aren't even synced in any good way.

[Nass]

Sounds like some kind of fluorescent lamps that pulsates with short cycles. Sometimes they show a heavy color cast at different times in the cycles and if you're unlucky, shooting with fast shutter speeds, they aren't even synced in any good way.

Yeah that's what I was guessing too

[VisualDarkness]

I've faced it when shooting sport events and almost decided cut the lighting cable in frustration. One violet chin and a second green one looks so ugly. Add a green floor to it and it will get messy no matter what you do.

Look at the examples and see if you recognize it: http://johnbdigital.com/lenses/fluorescent/fluorescent_lighting.php

[vonBaloney]

Yeah, fluorescent lights usually when they're aren't enough of them -- if you've got enough tubes going they sort of cancel each other out, but few in number you get weird greens, etc out of sync. Tough to deal with.

But in big gym maybe they have sodium vapor lights or something else like that? Still guessing the same problem -- weird lights on weird cycles, and not enough total lamps to cancel each other out. Can always go B&W...

[elliott]

But in big gym maybe they have sodium vapor lights or something else like that? Still guessing the same problem -- weird lights on weird cycles, and not enough total lamps to cancel each other out. Can always go B&W...

Most sodium vapor lights also cycle at mains frequency, 60hz or 50hz depending on where you are, so the same rules apply. Usually you can get around this by going with a shutter speed of 1/60 (or 1/50) to capture a full cycle, but this is not ideal with a longer lens. You need to capture full cycles, so 1/60 captures one cycle, 1/30 captures 2 cycles, but 1/45 will capture 1-1/2 cycles and throw the colors out.

[panoguy]

I've faced it when shooting sport events and almost decided cut the lighting cable in frustration. One violet chin and a second green one looks so ugly. Add a green floor to it and it will get messy no matter what you do.

Look at the examples and see if you recognize it: Limitations of Fluorescent Lighting - written by Curtis Newport

Aha, thank you! What a great explanation at that link! I've never experienced this before, but then I don't shoot indoor sports much at all. Now I'll try to remember the 1/120 trick, since here we are on 60Hz electricity, but for the sparring there was no way I'd go lower than 1/250 or everyone would look like a blur.

Now that everyone has mentioned shutter speeds, I've found that in all cases, the cast comes from the "longer" side of the frame, meaning the sides in portrait orientation, or the top and bottom in landscape, so it is definitely the shutter moving faster than the light is pulsing...

[VisualDarkness]

Aha, thank you! What a great explanation at that link! I've never experienced this before, but then I don't shoot indoor sports much at all. Now I'll try to remember the 1/120 trick, since here we are on 60Hz electricity, but for the sparring there was no way I'd go lower than 1/250 or everyone would look like a blur.

Now that everyone has mentioned shutter speeds, I've found that in all cases, the cast comes from the "longer" side of the frame, meaning the sides in portrait orientation, or the top and bottom in landscape, so it is definitely the shutter moving faster than the light is pulsing...

Blame the owners of the building.

[txsbluesguy]

Most sodium vapor lights also cycle at mains frequency, 60hz or 50hz depending on where you are, so the same rules apply. Usually you can get around this by going with a shutter speed of 1/60 (or 1/50) to capture a full cycle, but this is not ideal with a longer lens. You need to capture full cycles, so 1/60 captures one cycle, 1/30 captures 2 cycles, but 1/45 will capture 1-1/2 cycles and throw the colors out.

Most "budget"gymnasiums especially high school and community have sodium vapor lights. They have a ballast like fluorescent lights and to the human eye / mind appear to be continuous light with no variance in brightness and color. When photographed or videotaped they will cause strange issues as the cycle of the lights mix with the shutter speed in cycles. I used to tape HS sports and most of the time it was not an issue but when we shortened the shutter speeds in anticipation of trying to get better slowmo things got wonky. When you fast forwarded thru the footage you could see the brightness pulse and the WB cycle.

I say "budget" to separate from the high end college and pro arenas that use lighting designed to be more compatible with televising their games.

[elliott]

Will they allow you to use flash?

[panoguy]

Blame the owners of the building.

Japanese Canadian Cultural Centre - Welcome

I'm betting the lights are made by Sony, their biggest donor. Maybe if I'd used an A99...?

Will they allow you to use flash?

Never during a full-contact competition, although some folks had P&S cameras that did flash. Mostly it was iPhones shooting video...

[lister6520]

I hadn;t understood that you were getting different results in each photos.If the lights are properly designed then 1/360 (or sometimes 1/180 depending on the type of light) should be always OK. (1/300 or 1/150 in Europe).

A fluorescent light normally goes through two entire cycles in one electrical cyclem so with 60 Hz mains you get 120 pulsations per second as the positive and negative cycles are (usually) identical.

In large buildings the electricity supply is usually 3 phase and fluorescent lights are meant to be connected on sequential phases, preferably within the same light fitting if possible. That way at any one time you have lights on different parts of the cycle and provided that the fittings on the different phases are not too far from each other you get almost continuous light output. In industrial buildings there are regulations actually requiring this so as to avoid stroboscopic effects with machinery, which can be dangerous in some cases.

My guess is that in that gym they saved on costs and wired everything to the same phase, which is probably allowable since it wouldn't be classified as an industrial building. Unfortunately I cannot thing of any way to get around the problem, short of sponsoring them a new lighting system.

This is one of those situations where an iris shutter would help. You would still get differently coloured photos but at least the difference would be uniform throughout the frame and easier to correct. Unfortunately as far as I am aware there exists no k-mount lens with built in iris shutter.

[VisualDarkness]

This is one of those situations where an iris shutter would help. You would still get differently coloured photos but at least the difference would be uniform throughout the frame and easier to correct. Unfortunately as far as I am aware there exists no k-mount lens with built in iris shutter.

My Q-lenses got iris shutters and my new Voigtländer Prominent got two!

[maxfield_photo]

I agree with the assessment that it's the cyclical nature of florescent lights that's to blame, and actually I have an easy fix for you in post.

Since we know the gi should be white, simply flip the image to negative momentarily and sample the offending areas (magenta should look green in the negative). Then note the numerical value for the hue, probably somewhere around 90. Now add a photo filter if you're in photoshop with the hue set to 90 (or whatever you found), and tell it to preserve the luminance values. You may have to adjust the strength of the filter. Now, flip the image back to positive, add a black mask to the photo filter layer, and paint in white on the mask anywhere that you notice the magenta cast. You may even find that you can use a gray gradient for the mask since the color casts appear in wide bands that run across the frame. If you're using Lightroom or Aperture, you may be able to achieve the same effect by adding a weak green graduated filter to the image.