Re: R: camera calibration (fwd)

[Jan Hollan <jhollan@amper....muni.cz>]

> Is the "-t" parameter the calibration constant ?
> http://en.wikipedia.org/wiki/Light_meter#Exposure_meter_calibration
>
> Or there is any relation between "-t" and the calibration constant?
> I was finding out the calibration constant of my camera = 12.4

Stefano, this is a nice link, I did not know anything on how ISO is
defined, a further link to ``speed'' explains that somehow (did not study
it now).

As I see from the K definition, it should be possible to compute an
average scene luminance as

 L = K / ( time * ISO * (1/f)**2 )

using the values the camera decides for itself when measuring that scene.

I am using a formula

 L = pixel value / ( time  * ISO * (1/f)**2 * -t_divider)

So, if maximum pixel value would be some 4000, and -t56 would be chosen,
than my formula would translate, for just saturated pixels to K of about
70. For pixels at 18 % of saturation, the K would be of about 12 indeed.

The -t divider has not much to do with optimum setting for the image. It
is a constant to arrive at luminances from pixel readings. And has to be
determined from observations. Of course, you can even employ star images
for that, if they are not saturated...

An important hint: don't believe much those settings in EXIF, which have
been reported by a camera after it determined the settings itself. Use
manual settings instead. The reason is that cameras sometimss don't say if
they use a bit another ISO (like 139 instead of 100, for A/D conversion
from electric signal to pixel digital values) to make a nicer image out of
the raw data.

cheers,
 jenik