accuracy of colour temp readings

[Dan Hudgins]

Glass?

Glass?

what lens are you using?

Thats a good point, the lens will tint the light and can make the K value off by several hundred degrees.

I have a Russian lens that is like CC15Y+CC08R or something like that.

Many lenses are Yellow or Yellow-Orange tint, so the camera would need to boost the blue gain in metadata.

Another question is what kind of lens did the RED TEAM use when they calibrated the K meter in the camera?

The most neutral "lens" to calibrate the camera with would be a 2mm pin hole, since any coating and some kinds of glass can tint the light.

Also the IR and UV in the light may unbalance the RED/BLUE counts and thereby make the white balance off. Some lenses block more UV, and IR filters may or may not be in use, any of those can make for some drift in the K auto balance...

 

[Irri]

Do they need to use a lens to calibrate it? Doesn't the camera just balance whatever is in the centre of the screen until the R,G and B values are equal?

 

[Dan Hudgins]

?

?

Do they need to use a lens to calibrate it? Doesn't the camera just balance whatever is in the centre of the screen until the R,G and B values are equal?

How do you get something in the centre of the screen without a lens?

The counts from a Bayer sensor are not equal for white balance, the RED ONE shoots off gain balance fixed gain, later the data is processed through un-equal gammas or slopes to adjust the white clip points to be equal. So at no time are the A2D counts equal (for a gray card or white card).

In most Bayer sensors the green clips first then red then blue, otherwise the highlight extension would not be in the menu, it uses the un-clipped data, if the sensors were in balance no highlight extension would be possable.

Better color temp meters have two readouts not just one K value. Light has both a RED/GREEN and GREEN/BLUE ratio, so you need two numbers not just one.

 

[Fredrik Harreschou]

I just want to give an update to my issue:
Reinstalling software (B17) gave me two identical cameras! I am a happy DITpuppy...

 

[Graeme Nattress]

has anyone seen the following issue or is it just my camera?
during some testing we set up a single light source which we measured to be 3300 Kelvin. We tested with 2 other cameras (both Sony) and they both gave a reading of 3300 K. When I tried my Red it gave a reading of 3831 Kelvin. Further tests with 5600 K light source gave the expected readings on the other cameras but 6500 K on the Red. I have noticed that I am consistently 'correcting' my shots from the red by white balance shift in the same direction ie always towards a higher Kelvin figure. Can anyone (especially someone from Red) comment on this please.
thanks

peter

Just saw the thread.... Been deep into code and not tracking Reduser as much.

Ok - the question is: how do you know what the colour temperature of the light is? In my experience, an affordable colour temp meter is as accurate as it is affordable.

So, with that in mind we use a calibrated certified spectroradiometer and multiple light sources for calibration, checking that the camera reads as near as possible to the correct answer. Not only are a number of colour temperatures checked, but different sources - incandescent, fluorescent, LED etc. are used as they all have very different spectral properties.

For instance, my affordable colour temp meter says my tungsten light is 3200k. However, the manufacturer of the light and the spectroradiometer tell me it's actually a 2800k light, and my particular one is 2856k.

Now, the above relates purely to the B30+ colour science, as that was when the advanced kelvin calibration was introduced.

I hope that explains what is going on.

Graeme