UM46k - AWB test series with different lenses & settings

In this thread about white balance I wrote that different lenses give different results with the new auto white balance (AWB) function from firmware 4.0.

So yesterday I finally had an hour to perform a series of tests with different lenses and the results are interesting. I have not recorded any footage to show here, just wrote down the settings and results for further input and discussion.

To rule out any IR pollution or filter distortion I used my Dedolight DLED12.1 with absolutely no IR/UV emission. The LED has been measured to have exactly 5550K and a very high CRI/TLCI.

For the white balance I was using a X-rite ColorChecker White Balance chart. The DLED12.1 was the only light source.

With the iPhone video App Mavis - which is able to show the Kelvins of an AWB - I measured with an iPhone 6s 5400K on the chart - which is pretty close to what it should be. The latest iPhone cameras have been very accurate in this regards.

The Canon XF305 did white balance to 5000K and with an attached Heliopan Digital IR/UV filter it showed 4900K.

Here are the results from the UM46k:
ISO settings did not have any influence on the AWB, so I left it on ISO800.
Focal length did not have any practical influence on the AWB, too.
List displays Shutter, Aperture, Kelvin, Tint [scale goes from -50 to +50].
At 180°/T11 the chart's patch sat around middle gray in the UM46k's histogram.

Tokina Cine T3 50-135mm

• 180°/T4.0 ... 5900 -32
• 180°/T5.6 ... 5900 -26
• 180°/T8.0 ... 5850 -24
• 180°/T11 ... 5900 -23 ... middle gray
• 180°/T16 ... 5850 -17
• 45°/T5.6 ... 5950 -30 ... middle gray

Tokina Cine T3 11-16mm

• 180°/T3.0 ... 6750 -46
• 180°/T4.0 ... 6650 -40
• 180°/T5.6 ... 6600 -37
• 180°/T8.0 ... 6500 -38
• 180°/T11 ... 6650 -32 ... middle gray
• 180°/T16 ... 6600 -29
• 45°/T5.6 ... 6700 -35 ... middle gray
• 180°/T5.6 ... 6350 -35 ... + Heliopan Digital IR/UV
• 180°/T11 ... 6450 -29 ... + Heliopan Digital IR/UV ... middle gray
• 45°/T5.6 ... 6450 -33 ... + Heliopan Digital IR/UV ... middle gray

Sigma F1.8 18-35mm

• 180°/F2.8 ... 6500 -42
• 180°/F4.0 ... 6450 -34
• 180°/F5.6 ... 6400 -33
• 180°/F8.0 ... 6500 -29
• 180°/F11 ... 6450 -25 ... middle gray
• 180°/F16 ... 6400 -24
• 45°/F5.6 ... 6450 -28 ... middle gray

Canon F1.8 85mm

• 180°/F2.8 ... 6550 -35
• 180°/F5.6 ... 6400 -30
• 180°/F11 ... 6450 -21 ... middle gray
• 180°/F16 ... 6400 -24
• 45°/F5.6 ... 6450 -23 ... middle gray

Canon F4.5-5.6 100-400mm

• 180°/F5.6 ... 6700 -28
• 180°/F11 ... 6700 -20 ... middle gray
• 180°/F16 ... 6700 -19
• 180°/F22 ... 6750 -18
• 180°/F32 ... 7050 -8

Tamron F2.8 17-50mm LD XR DiII SP (bought in May 2006)

• 180°/F2.8 ... 6150 -35
• 180°/F5.6 ... 6050 -26
• 180°/F11 ... 6100 -19 ... middle gray
• 180°/F2.8 ... 5850 -33 ... + Heliopan Digital IR/UV
• 180°/F5.6 ... 5800 -24 ... + Heliopan Digital IR/UV
• 180°/F11 ... 5900 -17 ... + Heliopan Digital IR/UV ... middle gray

Tamron F2.8 17-50mm LD XR DiII SP (bought in January 2017)

• 180°/F2.8 ... 6400 -39
• 180°/F5.6 ... 6300 -30
• 180°/F11 ... 6350 -24 ... middle gray
• 180°/F2.8 ... 6200 -37 ... + Heliopan Digital IR/UV
• 180°/F5.6 ... 6050 -29 ... + Heliopan Digital IR/UV
• 180°/F11 ... 6050 -21 ... + Heliopan Digital IR/UV ... middle gray

Observations

1. For being dedicated cine lenses, the Tokina zooms do not match well in color temperature and tint
2. The Heliopan Digital IR/UV filter reduces the color temperature by ~ 250K+-50 and raises the tint by ~+2
3. The surprising differences in the results for the 11 years old and the brand new Tamron could be the result of aging coatings, different manufacturing processes, quality fluctuations, a changed formula or a mixture of any or all. It would be interesting to see if this applies to other manufacturer's lenses in a similar way.
4. You cannot rely on the UM46k color temperature presets, as they mean nothing without taking the lens, the sensor and filters into consideration.
5. ISO settings (200-1600) have no influence on the AWB results as several of my repeatedly done tests have shown.
6. Doing AWB you should let the white balance patch's exposure sit in the middle of your histogram, as this will give the most stable results for the tint. How much the tint drifts (from -2 stops up to middle gray: +6 to +9) is dependent a little bit on the lens. A shorter shutter will raise the tint a bit.
7. Kelvin results are quite stable throughout the exposure range and rise only when the patch is very bright (180°/F2.8) or very dark (180°/F32)

OK, so I did some further testing. This time I used another light source, an ARRILITE 750 Plus. That's a tungsten open face unit. According to the Arri Photometrics app the ARRILITE 750 Plus has a luminous output of 3020 lux (18° spot) at the distance of 3.16m I had placed it to the target patch.
It was ~1 stop brighter than the Dedo DLED12.1 so I set the shutter to 90°.

The ARRILITE 750 Plus specs say its HPL750 lamp has a color temperature of 3200K

iPhone Mavis app: 3450K
Canon XF305: 3000K with and w.o. IR/UV filter


Tokina Cine T3 50-135mm

• 90°/T4.0 ... 3100 -15
• 90°/T5.6 ... 3050 -10
• 90°/T8.0 ... 3050 -8
• 90°/T11 ... 3000 -10 ... middle gray
• 90°/T16 ... 3050 -7
• 22.5°/T5.6 ... 3000 -12 ... middle gray

Tokina Cine T3 11-16mm

• 90°/T3.0 ... 3150 -21
• 90°/T4.0 ... 3150 -16
• 90°/T5.6 ... 3150 -12
• 90°/T8.0 ... 3150 -12
• 90°/T11 ... 3150 -10 ... middle gray
• 90°/T16 ... 3150 -7
• 22.5°/T5.6 ... 3150 -12 ... middle gray
• 90°/T5.6 ... 3200 -6 ... + Heliopan Digital IR/UV
• 90°/T11 ... 3200 -4 ... + Heliopan Digital IR/UV ... middle gray
• 22.5°/T5.6 ... 3200 -6 ... + Heliopan Digital IR/UV ... middle gray

Tamron F2.8 17-50mm LD XR DiII SP (bought in May 2006)

• 90°/F2.8 ... 3100 -15
• 90°/F5.6 ... 3100 -6
• 90°/F11 ... 3100 -4 ... middle gray

Tamron F2.8 17-50mm LD XR DiII SP (bought in January 2017)

• 90°/F2.8 ... 3150 -16
• 90°/F5.6 ... 3150 -7
• 90°/F11 ... 3150 -5 ... middle gray
• 90°/F2.8 ... 3150 -13 ... + Heliopan Digital IR/UV
• 90°/F5.6 ... 3150 -5 ... + Heliopan Digital IR/UV
• 90°/F11 ... 3150 -3 ... + Heliopan Digital IR/UV ... middle gray

Observations

1. To be honest, I am a little bit confused by the results. I did not expect the different lenses to perform so similar under tungsten light.
2. The question is, if there is a flaw in my test setup while using the LED daylight. Does the color checker white balance sheet have a bias with the stronger blue of the LED light?
3. Why has the Heliopan Digital IR/UV filter so much more influence on the AWB under the LED light than under tungsten. One would assume that it would be the other way round?
4. Why is there so little difference between the old and new Tamron under tungsten, but a huge difference under LED daylight?
5. My guess is the sensor of the UM46k is daylight balanced. Probably around 5000K. CMOS sensors are by nature more sensitive to red. So if exposed to a very warm source the red response will be very high and the blue response will be very low. To achieve a neutral gray the blue channel needs digital gain, red channel gets negative gain.

Further tests with the ARRILITE 750 Plus corrected to daylight with an 10 years old maybe a bit faded full CTB gel (it should be full CTB but looks more like three quarter CTB).

iPhone Mavis app: 4950K
Canon XF305: 5300K without and 5100K with IR/UV filter

Tokina Cine T3 50-135mm

• 180°/T4.0 ... 5200 -12
• 180°/T5.6 ... 5150 -5
• 180°/T8.0 ... 5100 -4
• 180°/T11 ... 5050 -5 ... middle gray
• 180°/T16 ... 5100 +-0
• 45°/T5.6 ... 5100 -10 ... middle gray

Tamron F2.8 17-50mm LD XR DiII SP (bought in May 2006)

• 180°/F2.8 ... 5200 -11
• 180°/F5.6 ... 5100 +-0
• 180°/F11 ... 5150 +4 ... middle gray
• 180°/F32 ... 5200 +9

Tamron F2.8 17-50mm LD XR DiII SP (bought in January 2017)

• 180°/F2.8 ... 5350 -13
• 180°/F5.6 ... 5300 -2
• 180°/F11 ... 5300 +2 ... middle gray
• 180°/F32 ... 5350 +4
• 180°/F2.8 ... 5400 -6 ... + Heliopan Digital IR/UV
• 180°/F5.6 ... 5300 -5 ... + Heliopan Digital IR/UV
• 180°/F11 ... 5350 +8 ... + Heliopan Digital IR/UV ... middle gray
• 180°/F32 ... 5350 +15 ... + Heliopan Digital IR/UV

Ok, I found one problem with the Dedo DLED12.1:
It is shifting color temperature when the focus of the lamp is changed.
In full flood I get 5600K -19, at medium I get 5900 -20 and at full spot I get 6250K -25 with the UM46k.
The iPhone is less affected, Mavis shows 5300K / 5450K / 5650K.

Further tests with the Dedolight DLED12.1 corrected to tungsten with an 10 years old maybe a bit faded full CTO gel (it should be full CTO but looks more like three quarter CTO).

Interestingly from full flood to full spot the color temperature only changes around 100K.

iPhone Mavis app: 3000K

Tokina Cine T3 50-135mm

• 180°/T4.0 ... 3050 -12
• 180°/T5.6 ... 3050 -9
• 180°/T8.0 ... 3000 -8
• 180°/T11 ... 3000 -8 ... middle gray
• 180°/T16 ... 3050 -5
• 45°/T5.6 ... 3000 -10 ... middle gray

Tamron F2.8 17-50mm LD XR DiII SP (bought in May 2006)

• 180°/F2.8 ... 3050 -13
• 180°/F5.6 ... 3050 -5
• 180°/F11 ... 3050 -2 ... middle gray
• 180°/F32 ... 3050 +2

Tamron F2.8 17-50mm LD XR DiII SP (bought in January 2017)

• 180°/F2.8 ... 3150 -14
• 180°/F5.6 ... 3100 -7
• 180°/F11 ... 3100 -5 ... middle gray
• 180°/F32 ... 3100 0
• 180°/F2.8 ... 3050 -14 ... + Heliopan Digital IR/UV
• 180°/F5.6 ... 3050 -7 ... + Heliopan Digital IR/UV
• 180°/F11 ... 3050 -5 ... + Heliopan Digital IR/UV ... middle gray
• 180°/F32 ... 3050 +-0 ... + Heliopan Digital IR/UV

And finally a test outside with a weak winter midday sun:

iPhone Mavis app: 5300K
Canon XF305: 5700K without and 5600K with IR/UV filter

Tokina Cine T3 50-135mm

• 11.2°/T4.0 ... 5600 -5
• 11.2°/T5.6 ... 5500 +-0
• 11.2°/T8.0 ... 5400 -1
• 11.2°/T11 ... 5450 -6 ... middle gray
• 11.2°/T16 ... 5450 +10
• 11.2°/T32 ... 5500 +10

Tamron F2.8 17-50mm LD XR DiII SP (bought in May 2006)

• 22.5°/F2.8 ... 5600 -8
• 22.5°/F5.6 ... 5550 +6
• 22.5°/F11 ... 5600 +11 ... middle gray
• 22.5°/F16 ... 5600 +13
• 22.5°/F32 ... 5700 +13

Tamron F2.8 17-50mm LD XR DiII SP (bought in January 2017)

• 22.5°/F2.8 ... 5800 -11
• 22.5°/F5.6 ... 5750 +3
• 22.5°/F11 ... 5750 +5 ... middle gray
• 22.5°/F16 ... 5700 +11
• 22.5°/F2.8 ... 5650 -7 ... + Heliopan Digital IR/UV
• 22.5°/F5.6 ... 5600 +6 ... + Heliopan Digital IR/UV
• 22.5°/F11 ... 5650 +10 ... + Heliopan Digital IR/UV ... middle gray
• 22.5°/F16 ... 5600 +13 ... + Heliopan Digital IR/UV

Whew, Robert! I wonder if the SLR Magic Image Enhancer Pro would help even out some of these temperatures and tints. It's not going to hurt and may help as part of it is designed to smooth out the spectral curve to resemble true daylight. Is than an option for you to consider?

As is, it's a daunting task to colour balance all your lenses and lights and focus of the lights. I wasn't aware that a change in focus made for significant changes in temperature. I only expected intensity falling on the subject area to change.

I know most people have a variety of lenses and lights from a variety of manufactures, but to me, these results argue in favour of shooting with a colour-matched set of lenses and lights on a given project. Certainly keep it consistent within a scene.


Sent from my iPad using Tapatalk

Rick, the issue is not so much changing the focus on the light (spot to flood, I do this with tungsten lights with out issues) it is how the LEDs in the Dedolight is reacting with the different angles of coverage, this is one of the issues using LED lights. I have a small LED disk fill light which is not supposed to change coor temp with different settings (it also has a high CRI rating), but it does shift some.

Robert, I think you are correct in that the older lens coatings on the earlier lenses do effect CB somewhat.
I liked the WB on the AF100, in that you can set the color temp in 1-degree increments, while looking at a Vectorscope meter in the monitor, to nail the CB in a given shot. I did use AWB, worked most of the time, but it can bite you, if yiumare not careful. Shot an outdoor scene, mid day and used the AWB preset, what I did not see was the water nearby (full of green plants) reflected a high amount of greeen light into the scene, I did not see at the time. My footage had a heavy green tint!
Cheers