So I've been thinking a lot lately about how to correctly calibrate my various light meters. The Internet is chock full of people sharing their rules-of-thumb and get-it-in-the-ballpark methods, pretty much
none of which even vaguely resemble what a professional might do. The closest you ever get to that, is recommendations for specific labs that can do calibration. No one ever talks about
how those labs do it.
That being said, I've recently decided to tackle the problem myself. As a reference, I've been working from the
Exposure Metering (by Jeff Conrad) document.
As tools, I've been using the following:
- An incident lux meter with proper NIST-traceable lab calibration (I have an Extech LT300 and a Minolta CL-200A, though they basically read the same under the same test conditions.)
- A Sekonic Exposure Profile Target II, which contains a very large and high quality gray card. I've measured this on my densitometer to have a density of 0.75, which translates to a reflectance of 17.78% (a.k.a. "18% gray")
- A big diffuse light source, which in my case is the sun under an open sky on an overcast day. (I'd love to use an artificial source for this, but I don't have anything suitable. The smaller the light source, the bigger the variations your geometric conditions introduce with lumisphere-style incident meters.)
- A collection of incident and reflective light meters to compare
I've also decided to do this entire exercise in "EV at ISO 100", since its a standard absolute unit most light meters can be set to read in terms of. There's a lot less room for "rounding error" than something as coarse as standard f-stops or shutter speed stops.
The goal is, given a lux reading from a lux meter, to determine:
- What EV an incident-light meter should give at the same spot
- What EV a reflected-light meter should give from my gray card placed at that same spot
That all being said, here's the math I've come up with for incident-light meters:
EV100 = log2((Lux * 100) / C)
(where C=340 for Sekonic lumisphere-style incident meters)
For reflected-light (e.g. spot) meters, it gets a little bit more complicated:
Z = 10(-1 * D)
Ls = (Z * Lux) / Pi
EV100 = log2((Ls * 100) / K)
Where:
- D = measured density of the gray card being used
- Z = reflectance of the gray card (can skip the calculation and set directly, if you already know it and/or don't have a densitometer)
- Lux = the same incident lux reading used in the previous calculation
- K = a constant that's typically 12.5 for Sekonic/Canon/Nikon meters, and 14 for Minolta/Pentax meters
In my latest worked example, for which I've already put together a spreadsheet to simplify, I get the following results:
- Measured lux: 43400
- Sekonic lumisphere incident meters should measure: EV=13.6
- Sekonic spot meters should measure: EV=14.3
- Minolta/Pentax spot meters should measure: EV=14.1
Does this whole process make sense? Is there some critical detail I've completely overlooked? Just throwing this out there to see what others think of this whole attempt at sorting things out.