I've clipped the original post around a bit - hope I have not changed the meaning anywhere!
> If it handles the reflectance percentage, I'm happy. But it kind
> of throws the whole issue of color mindedness into question.
You appear to assume that the photodetector in normal usage (i.e not an F5) is colour neutral - it is not.
This can be handled (by the manufacturer) by:
a) adding a filter to the detector (which relies upon assumptions about the light spectrum illuminating detector) and then applying (b)
or just
b) by assuming the spectrum is, say, daylight and calibrating the detector response on that assumption.
But if the spectral distribution of the light is not according to the assumptions then the meter reading will be off (in the strict technical sense, not necessarily end user noticeable).
> What I don't get, won't the color meter go crazy when you mount
> a medium yellow, or an amber, or a red, or a warming filter? How
> do you let the meter know that there is a new baseline color
> applied before the lens that it should treat as a null?
We are not white balancing a video here! The colour you have coming in from the lens is the colour you will get on the film. All the photodetectors can tell you is how many milliwatts of light are exciting them. By making them colour sensitive you know how many milliwatts of R G or B are present, and the base exposure can then be determined for each "cell" of the meter with a slightly better result than just assuming the light is, say, "daylight". Having lots of sensor cells is then like a super matrix system. Adding the picture database and lots of sensors also enables you to tweak the exposure a bit more (if the picture is recognized).
> Does anyone have a suggestion for a test that might show the value
> of color metering (or even to demonstrate that it actually
> exists?)
Not easy, you really need a reference source and/or calibrated reference detector (or at least a spectrally neutral detector such as a bolometer) i.e. access to a fairly well equipped scientific optical laboratory.
Alternatively you could try photographing a grey card with different filters applied using two cameras (F5 and non-F5) and then checking the exposure with a densitometer Before doing this you should have calibrated the camera meters (under say "standard daylight") and possibly your processing techniques. Note: this also contains assumptions about the spectral response of the film used - which is also not uniform. You should also ideally use the same roll of film in both camera or at least from the same batch.
---> messy and probably not that reliable
> I am not bothered by this, because I'm still not convinced that a
> light meter should even care about color.
and
> I say this as an F5 user who has yet to see a difference based on
> color metering.
I think this is a slightly different point.
We know (from experience) that the "old" polychromatic photodetectors worked fairly well, so the addition of colour metering is "only" to (a) provide an incremental improvement to the metering and (b) detect and handle the special cases. Therefore you should only see a closer to correct exposure not a radical difference (unless you are in a special case).
Can we really expect anything more?
Andy Shaw
