[MrPetkus]

Folks,
I've been hesitant to write this post suspecting that I've been doing something amiss but I'm having issues under certain conditions with the K5, namely:

1) At a dance studio (tungsten lighting) where I've recently done some work the color is absolute garbage - usually 1/2 of the image has a dark reddish hue while the rest is somewhat acceptable. Not every shot but sometimes 1/3 or more of the batch is afflicted. I've done grey card white balance adjustments on the scene, everything checks out OK and then 20 shots in the problem starts anew. Sometimes restarting the camera helps.

I can recover the color in Photoshop but it takes work. But to make matters worse, some of the real bad color examples also have unacceptable noise (even @ low ISO) or the focus is somewhere completely unintended.

Here is an example of the same image, straight from the camera and then recovered.

2) Overexposure, sometimes severe when using spot metering with manual lenses. I never had this problem with the same lenses on the K7.

Any thoughts?

Thanks

---

Last edited by MrPetkus; 04-17-2012 at 08:49 PM.

[raz-0]

Is the light in the picture from a source (multiple source) of one color temp? Or is it mixed. Even if it was mixed, that looks pretty odd. Was white balance auto, or did you set it to a fixed value? My memory wants to say sensor overheating, but I can't find the example i was looking for. If you ahd a fixed value and got inconsistent results without the lighting changing, something's not right with the camera.

[kenyee]

I don't think that's tungsten...looks more like sodium vapor. If so, it oscillates in color every 30th of a sec IIRC. Try to set your shutter to multiples of it. Have you looked at the lighting fixtures?

[JeffJS]

I Have found that in some situations, the K5 tends to over-expose by as much as a stop. I'm also mindful that when spot metering, especially if changing from shot to shot, that one must be careful of the metering area or object. Are you Certain this is all Tungten lighting and not some sort of Mixed lighting?

Your two photos look identical to me (maybe with good reason) except the top one is over-saturated.

I don't have any real answers for you, I'm just making notes here. I will note that I always use RAW so I can make adjustments easier. Using the out of camera JPGs has always been an issue for me. What JPG setting are you using?

[Ole]

Assuming that the middle girl's pants should be black, then your fixed version isn't fixed but more off than the original where that girl's pants are pretty close to being black. It is also washed out.

I think this is a case of mixed light - adjust for the children, not the walls.

Here is your original with correction by levels - one click with the grey eye-dropper on the pants.

[MrPetkus]

First off - thanks for all the quick and helpful responses.

raz-0: WB fixed. Tried auto but the same issue arises after a dozen frames or so.

kenyee: The lighting wasn't sodium vapor but I can see why that would be suspect here. If I head back I'll analyze the lighting - before the students come in. At the time I was becoming more and more frustrated by what was appearing on the preview screen I began to lose reason - and time.

JeffJS: All the same lighting. Rows of hanging incandescent lights. The entire opposite wall was a mirror. What I wanted to show with the 1st image was how the top of the image has the reddish cast where the bottom doesn't. The second picture was less a fix than a quick illustration for this post where I made the entire image match the bottom half. Honestly, I'd prefer if the entire image had the reddish tint than just half.

As for spot metering, there were times when I would switch from center weighted to spot, grab a new baseline exposure with the green button (using old manual lens) - I would see the new shutter speed and I would get a wildly overexposed shot. Shut the camera off, turn it back on (same settings) and then proper exposure.

Ole: Whatever I might or might not have done in PP is not really of concern at the moment - I just wanted to differentiate between an image that is 1/2 screwed vs. consistent.
Assuming there was mixed lighting - since all the lights were in rows overhead, I'm trying to imagine which lights and where would be mixed.

I wonder if the massive mirror had some role to play? This has never happened on *any* other image I've taken with the K5. Put it this way, if the whole session was screwed I would accept the fault was behind the camera. But I can't see how restarting the camera would miraculously make the problem disappear for a short while.

Thanks again everyone.

[Miguel]

Especially in challenging mixed lighting situations, I would suggest shooting RAW and then in ACR creating some presets to the corrected white balance. Jpeg has too many limitations.

In Lightroom it took about 45 seconds to produce an acceptable shot to my standards. I found Temp +5 and Tint -25 worked well. I also reduced the Vibrance a little. I like the brightness of the image and don't consider the first one overexposed on my calibrated IPS monitor. Apply some moderate NR and you are good enough.

M

[davidreilly3207]

Having been the lead electrician at a convention and meeting room facility, I believe it would be very unusual for the lighting to be 100% tungsten, which is another word for incandescent. The resulting heat and cost of using the lights would be prohibitive. Also, unless the space is very large and the ceiling is over 40feet, it is unlikely that mercury vapor, metal halide or sodium vapor lights would be used. These are very intense lighting sources. It is more likely that the lighting is either florescent or a mix of fluorescent and incandescent, the florescent for instant on and economy and the incandescent for its dimming capabilities. Fluorescent lighting will change colors dependent on the part of the 60hz cycle you are experiencing at the moment your shutter opens. Any shutter speed faster than 1/60 will likely catch only part of the cycle and show color shifts randomly.
Looking at your picture, It looks like the perimeter is incandescent and the center is flourescent, a typical setup in meeting rooms.

[MrPetkus]

This sounds reasonable. I'm heading there tomorrow AM and will report back with the definitive lighting setup.

Having been the lead electrician at a convention and meeting room facility, I believe it would be very unusual for the lighting to be 100% tungsten, which is another word for incandescent. The resulting heat and cost of using the lights would be prohibitive. Also, unless the space is very large and the ceiling is over 40feet, it is unlikely that mercury vapor, metal halide or sodium vapor lights would be used. These are very intense lighting sources. It is more likely that the lighting is either florescent or a mix of fluorescent and incandescent, the florescent for instant on and economy and the incandescent for its dimming capabilities. Fluorescent lighting will change colors dependent on the part of the 60hz cycle you are experiencing at the moment your shutter opens. Any shutter speed faster than 1/60 will likely catch only part of the cycle and show color shifts randomly.
Looking at your picture, It looks like the perimeter is incandescent and the center is flourescent, a typical setup in meeting rooms.

[MrPetkus]

Thanks mdave13 for pointing me in the right direction.
I went back to the studio this morning before the students arrived and discovered the lights are indeed fluorescent (the fixtures are styled as if they were incandescents). Just a single light source though.

I'm going to do some tests later with different shutter speeds.

[davidreilly3207]

And there another problem arises. Florescent lamps change color with age, and thats assuming all the lamps in the facility were the same color to start with. Florescent lights come in many different temperatures(colors). Most places, you will find, don't pay attention when re-lamping and you wind up with a mix of lamps. Even if they are diligent in replacing like with like, different batches of the same temperature can vary by as much as 200 degrees, very noticeable. Also, if there are any windows, they will add to the confusion. Florescent lighting is probably the toughest shooting scenario. You may want to think about some flash, gelled for correction to florescent. Good luck.

[Wheatfield]

That is definitely from shooting at too high a shutter speed under florescent lights.
Here's a couple of second series I shot of my dog doing a formal recall.
The colour shifting is quite evident.

[WerTicus]

Yeah my K-5 does similar things colour wise, except i dont get squiggles on faces.

My K-x had more accurate and consistent AWB. No biggy, I shoot raw anyway. Not seeing any of the other issues you're having though.

[MrPetkus]

Wheatfield - that is absolutely and positively what I experienced. I did some more shots yesterday to confirm.

I'm glad this forum was helpful - thanks everyone. To close out my original post:

1) The color issues I experienced are due to difficult and irregular fluorescent lighting. In order to eliminate shadows cast by flash, I used high shutter and ISO to capture motion. In essence, my shutter speed was out of sync with the lighting.

From Wikipedia: "Fluorescent lamps using a magnetic mains frequency ballast do not give out a steady light; instead, they flicker at twice the supply frequency. This results in fluctuations not only with light output but color temperature as well,[24] which may pose problems for photography"

2) The temporary improvements I experienced when restarting the camera, changing WB, etc., where mere coincidence and had nothing to do with those actions.

3) The issues with metering seem to be more of an issue with my older manual lenses.

Thanks again.

I shoot ballet, rehearsals, dress and performance a lot. I always shoot raw and never use the camera meter, I measure incident light for an EV and adjust from Sunny 16 as an anchor. Recently, I shot a photo session at higher speeds than normal to capture action. Mixed light, lots of windows and high fixtures I thought were something like sodium but were probably fluorescent. My WhiBal measured 3650.

Lots of my exposures had variations of color, like a bleed that now I can correlate to the window side and fluorescent side. I was using two K20 bodies with a Tamron 28-75 on one and DA* 50-135 on the other. Everything at /200 was mostly okay, 1/250 cost me alot of time in RAW to fix and 1.320 not so bad. Previously, I was never over 200 and shot a lot of 1/180 and 1/160 speeds with nice results.