OT: Sony EX1 White Balance Question

MUTTLEY wrote on 3/5/2010, 8:38 PM
Hoping someone here might know the answer to this one. Ran into a perplexing situation tonight that I'd never noticed before and can't for the life of me understand.

Multiple EX1s, I did a manual white balance and it read and set my camera to 5100k. Instead of manually white balancing each camera with the card it seemed like it would be easy enough to just set all of them manually to 5100k in the picture profile instead of running around to each camera with the card. The result was that setting it to 5100k was totally green even though getting the white balance off of the card set the camera to 5100k and was perfect. Why wouldn't 5100k be the same on either or? Confused on this one.

Tried this on all four EX1s that were there and it was the same thing. White balance off the cards set it to 5100k, setting it to 5100k manually made it green.

Here is a little vid to demonstrate what is happening:
Sony EX1 White Balance Issue Demo
password: sonyforums

- Ray
Underground Planet

Comments

farss wrote on 3/5/2010, 9:46 PM
Somewhere I did read a cautionary note about adjust CT by dialing in a number with the EX cameras. CT is a poor description of light and when you do a manual WB the EX seems to do more than just assume the CT of a black body radiator. It also explains why doing a WB in real daylight can give perfect results but gives extreme numbers for the CT with the EX cameras.

Bob.
Serena wrote on 3/5/2010, 9:58 PM
Ray,
I've no idea why that happened and getting several EX1s to match indicates it isn't a firmware error. However I just tried that on mine and the colour is fine (not green or any other deviation). First I thought you might have set white offset, but it isn't possible to confuse those. Obviously you set the white balance to "preset" (to use the PP set WB CT), so I'm at a loss to figure how you achieved that outcome.

Serena
PeterWright wrote on 3/5/2010, 10:18 PM
Probably not related, but although the "kosher" way to set WB is to point at something white then press the button under the lens, I've found on occasions that just pressing the button anytime produces an acceptable result.
farss wrote on 3/5/2010, 10:46 PM
My theory is that whatever light source lit the card for the camera that was given a proper WB had a bit of green in it. Not an usual situation. So that camera set the gain on the green channel back a bit to compensate. The other cameras didn't as they were only set to match a theoretical light without the green.

Nothing broken, nothing unexpected once you understand how the EX cameras do a WB and they do take their time about it.

Bob.
megabit wrote on 3/5/2010, 11:19 PM
It's a known fact that dialed-in CT values apply to all RGB channels in "preset" way, i.e. cannot make for magenta (or green for that matter) shifts/hues of a real scene. On the other hand, AWB does - so with multicamera shot it's extremely important to either use dialed-in preset for all cameras, or - better still - AWB all of them to the same white (grey) card.

AMD TR 2990WX CPU | MSI X399 CARBON AC | 64GB RAM@XMP2933  | 2x RTX 2080Ti GPU | 4x 3TB WD Black RAID0 media drive | 3x 1TB NVMe RAID0 cache drive | SSD SATA system drive | AX1600i PSU | Decklink 12G Extreme | Samsung UHD reference monitor (calibrated)

Serena wrote on 3/6/2010, 3:56 AM
Not sure that we've read the situation correctly. As I understand the original query, setting the cameras to 5100K resulted in a strong green bias; "totally green". Even if the lighting has somewhat excess green (e.g. commercial fluorescent) I wouldn't expect the result described. I agree that a WB from a card may compensate for the green while indicating 5100K, and the manual setting would not do so. But totally green?
craftech wrote on 3/6/2010, 4:32 AM
Hoping someone here might know the answer to this one. Ran into a perplexing situation

I know you shoot mostly bands Ray. Were you in a club? What kind of lighting did they use?

John
PerroneFord wrote on 3/6/2010, 6:45 AM
These cameras are twice as sensitive to green as any other color. This doesn't surprise me in the least. Why do you think people use Greenscreen for video based chromakey?
MUTTLEY wrote on 3/6/2010, 9:38 AM
Thanks for all the replies, just for clarification I've sloppily thrown together a little vid just so you can see what I was saying. I've also added it to my original post above.

This was shot in a studio with studio lighting.

Sony EX1 White Balance Issue Demo
password: sonyforums

- Ray
Underground Planet
farss wrote on 3/6/2010, 1:56 PM
So you got a green caste.

1) Page 73 of the EX1 manual:
Except when Preset white balance mode is selected, you cannot confirm the results of changing the Preset White settings on the screen.

2) Any filters on these cameras, specifically the Tiffen IR cut filter?

Bob.
Serena wrote on 3/6/2010, 2:59 PM
Ah, "totally" was a bit misleading. Actually neither shot looked, to me, to be correctly WB. What lighting source? I did my little experiment under daylight and wasn't looking for lesser shifts.
MUTTLEY wrote on 3/6/2010, 4:31 PM

No filters used. I believe they were 5100k lights.

The basic confusion for me is that 5100k should be 5100k regardless right? That is the point of it isn't it? So if the camera says it's set to 5100k whether by card or by dialing it in ...

- Ray
Underground Planet
Ros wrote on 3/6/2010, 4:48 PM
From past experience, I believe this is just normal.

If you white balance under neon lights, you may get a reading of 3200K which will be different from a preset 3200K (for balanced tungsten lights).
That is why indoors, I would always do a white balance, unless I would be certain that it is 3200K studio lights.

I own an EX1, but haven't tested for this matter since this is the way I have been working for over 20 years, mind you that viewfinders where black & white and most of the time no color monitors!

Rob

farss wrote on 3/6/2010, 6:07 PM
"The basic confusion for me is that 5100k should be 5100k regardless right?"

Not at all: http://en.wikipedia.org/wiki/Color_temperature

"The color temperature of a light source is the temperature of an ideal black-body radiator that radiates light of comparable hue to that light source."

Modern light sources are not "ideal black-body radiators". When you do a full WB with the EX it adjusts the gain in the R,G,B channels to get white. There's a large range of combinations of R,G,B light all of which give a CT reading of 5100K.
So when you simply dial in 5100K the EX has to assume an ideal light source, it has no way of knowing it isn't ideal.
So what has happened is one camera has adjusted for whatever the light source truly is and the others haven't.

If you go back to the Wiki and look down the page at the CIE 1960 UCS chart and look along the isotherm for 5000K you'll see it covers a range of colors. In fact you can move the line into the green and you still get 5000K. As noted however for delta uv values greater than 0.5 the CT becomes a meaningless figure. The good news is the EX cameras can cope to.... some extent.


Regarding filters, all the EX cameras we rent out are fitted with B&W 486 IR cut filters. Unless you looked you wouldn't know. It wouldn't suprise me if other rental companies are now fitting the Tiffen T1 filters. If the cameras have the T1 filters fitted then they absolutely must be WB'ed off a card or you will get a green caste. Based on Serena's comment that even the one that was WB'ed properly doesn't look quite right makes me very suspicious that this is what has happened.

Bob.
MUTTLEY wrote on 3/7/2010, 9:24 AM

Well that's a lotta info! Thanks Farss, and everyone else even though its not what I was hoping to hear.

And yea, totally understand Serena about the white balance not looking right, but the goal wasn't to get perfect wb but to try to figure out why the two were different if set to 5100k or perceived as 5100k. I still can't say I full understand the whys and wherefores but it seems the upshot is that you risk getting unwanted color casts when trying to dial in the temperature as opposed to using a white balance card. A shame, the hope was that once wb was achieved on the one camera it would be just as simple to go to the other cameras and change them based on the kelvin that that first camera perceived.

Great info all, thanks for looking into it and taking the time to post.

- Ray
Underground Planet
Serena wrote on 3/7/2010, 3:47 PM
The capabilities built into digital cameras (video and still) encourages us to take all that basic stuff for granted. With film you had to have your box of filters and a good colour temperature meter, as well as a good exposure meter. And a tape to measure subject to focal plane for focus. Incandescent light sources are fairly straight forward (continuous spectrum) but discharge sources (such as fluorescents) do not emit a continuous spectrum and can readily fool the eye and cheaper CT meters. And tubes for commercial lighting vary in characteristics between manufacturers. That we can now get a satisfactory WB at the push of a button is pretty amazing. But don't use bright white printer paper --- it fluoresces and your WB can read 7000K under 5300K lighting.