F350/F355 Preset White Balance
 

Hi,

We have two F350s and one F355 and for the same given color temperature, all three cameras show quite a color tone offset between manually dialing a white preset value and actually white balancing. This is especially true when working under tungsten lighting conditions. For exemple, yesterday we were doing a studio shoot and were reading 3000K under the lights. I dialed in 3000K in the White Preset settings and under this setting the pictures showed a marked and unpleasant greenish tone. Switching back to "white balanced" 3000K showed a much more pleasant image with nicely saturated and well balanced skin tones. I have been noticing this for a while and usually recommend my shooters that they always white balance. We also have a PDW-700 but on that camera, whenever you are dialing a white balance value or white balancing, the picture will look very, very close.

Thierry.

Thierry, in my TV news days we often had a chance to shoot side by side with identical models of the same camera delivered at the same time. More often than not color temp, picture, saturation, white clip, etc. varied (sometimes drastically) between cameras direct from the factory. As a result we always had the station engineer do an initial set up on new cameras so we would at least have a consistent look among staff shooters. Our engineer explained that the differences had to do primarily with different engineers doing the pre-ship set up on cameras.

You would think that would be less prevalent today with cameras being more computer than camera but it isn't. I'm with you and a big fan of white balance unless I have two cams set up by Macy to a consistent standard and then I'll trust preset when appropriate. I can match my F330 & Z1 almost perfectly but I've tweaked the settings a little to my preferred look.

Do you use the same lenses? and are your lenses corrected in a lens file?
Do the different camera's have their user matrix settings inactive?

If not, you won't get matching images.

When you perform whitebalance it also corrects the lens specific shading issues but in preset mode is doesn't.

Changing our using lens files does not affect this issue. Basically, for the same given color temperature, let's say for exemple 2900K, the picture tone balance will look quite different whether you white balance or dial in a white preset value. I'll post a couple frame grabs for you to see.

Thanks.

T.

I noticed the same thing with my 350.

My take on the situation is, that dialing in a preset color temp, is REALLY, just rolling the last white balance value to a higher or lower number. The problem is that every auto white balance is going to have it differences depending on the light under which it was taken. These differences get magnified as you adjust the setting manually.

I got caught once, when I white balanced for a shot outside in the shade, and then shot news clips indoors. Rather than white balancing, I just dialed in 3400K, and later, to my horror, all the flesh tones looked sickly green.

Generally, I shoot preset 3200/5600 when possible, and white balance for the rest. I rarely use that feature, perhaps to warm up a cold looking WB, but with caution.

I really miss the dual filter wheel assembly of my old camera, and I think it will be one of the best additions on the new F800.

Your opinion makes a lot ot sense to me and your use of preset and whitebalance is exactly mine. Thanks!

There is probably a way to test whether the adjusted preset is interacting with the auto white or whether it is independent. My SDX-900 cameras 3200 preset does not have a seperate register from the scene files you can use so the preset does change if you let the auto white function reset the registers. Try Autowhiting the camera to a daylight source while checking the charts on a tungsten preset source. Then autowhite in the tungsten light on the chart. If the preset changes based on what the auto white is then you know they interact.
On the other hand the original question was why don't the cameras all look the same when in Preset vs the autowhite. Improvements in electronics notwithstanding it is actually harder to set and keep electronics tuned to a particular standard than it is to have an autowhite circuit which gets rid of most of the differences. Any drift or adjustment to the camera can change the preset in a camera so it may look alright by itself but different than another camera in the same setting. Add in people being able to customize settings and you need to be able to do a global reset switch to get all the parameters to start from the same place which might not give you the result you want either.

If found that if dialing in whitebalance (Kelvin)-Numbers you should always start from daylight or tungsten (find that in menu ALL 44, "D5600" on/off) and then dial in your Kelvin value.

Dont change the other offsets below the Kelvin field - those should sit "straight" at 0 at 3200 and / or 5600.

Following that rule I dare to dial in shifts of lets say 4300/5600/6300 or other values.

Didnt precisely check the results, but at the moment it turns out to be a good way.

ULi
_______________________
EDIT:
Later I found there is still a difference between dialed in K and Manual Balance...

85 or 80
 

Hi,
I´ve been trying to guess what the native color temperature of the 350 ccds is.
(Didn´t find anyhing in the specs)
I think I get more dinamic range in daylight than in tungsten interiors...
could it be that the sensor is 5600(aprox) native and not 3200?? I also think
noise under low light is more noticeable at 3200, could it be that the cam is
boosting the blue channel? Then a 80 would yield better results.

What do you think?
Thanks

Not sure if you can determine the native color temperature of the chips when using a 3 chip system. Most cameras use a clear filter when in Tungsten so you can bet the electronics are biased in that direction. They use color correction on the daylight filter settings because they want the Tungsten performance to be faster where light levels can be low and don't need the ectra sensitivity as much in Daylight settings. A single chip or sensor like the one used in the Red seems to be Daylight balanced. On a 3 CCD system you should know that not all the channels are treated equally in terms of the amount of light they receive from the beam splitter. Traditionally Green which also is where most of the resolution comes from gets the most. Red gets the second and blue gets the least. Since interior low lighting has very little blue color temperature this ratio is done to maximize color rendition in the lower kelvin color temperatures which the human eye is very sensitive to. This often results in a very noisy blue channel which is basically providing color information and not used for resolution for the most part.

Hi Daniel-

It seems to me that a 3-chip design has an inherent design advantage here over a 1-chip design, as altering the analogue-stage gain per-channel (/per-chip) would to some degree alter the native colour temperature. Obviously analogue gain doesn't come without noise, so I'm not suggesting constant per-channel noise over a range of colour temperatures, BUT it does give you the opportunity to somewhat avoid the problems inherent to digital colour correction (including the dynamic range issue of having very red or very blue images, and obviously the rounding errors inherent to large corrections). No doubt this would be quite obvious to camera/ccd engineers, so the question is whether this is technically feasible, and if so, which (if any) cameras make use of it?

Kris Bird

I think the F355 etc is the same as the PDW-700 in so much as the cameras are balanced for daylight optically and then corrected electronically for tungsten etc.

Traditionally cameras were balanced for Tungsten and then added colour correction optical filters to get to daylight. This was done as CC filters absorb light and thus make the camera less sensitive. Normally when shooting outdoors in daylight sensitivity is not an issue while shooting indoors under tungsten light you used to need every bit of sensitivity you could get.

The down side to this approach is that tungsten contains very little blue light so to get a natural picture the blue channel was often running at quite a high level of gain which increases noise in the blue channel and thus overall noise. In addition when you rotated in the CC filters to get to daylight the sensitivity of the camera was reduced, so you did not have constant gain.

With the PDW-700 (and also the F350 I believe) the cameras are essentially balanced for daylight, without the use of any CC filters, which helps reduce noise in the blue channel. Then for tungsten shooting you electronically re balance the camera. By doing this the overall sensitivity of the camera is constant whether shooting at 3.2K or 5.6K and you only get additional blue channel noise while shooting under tungsten. If you are worried by blue channel noise you can always correct from daylight down to tungsten with an optical CC filter (80A) and leave the camera set to daylight, although this will reduce the systems overall sensitivity by around 1 and a half stops.