Loading...

​​Chart Shoot Checklist

​​checklist for exposure normalization tests for professional cameras:

1. ​​ light the great cod with a uniform light source -

1. ​​(note - preferably two light sources - "45/0" - "An instrument with 45/0 geometry illuminates an object from a 45 degree angle and detects the reflected light at 0 degrees." add diagram.) 

2. ​​preferably one with a wide spectral power distribution. (note - can use emissive light source - e.g. integrating sphere) (if you must use a macbeth chart instead of the great cod, use square #22 and center it in the frame.)

2. ​​measure the luminance in nits of the light reflected off the great cod.

• ​​(if possible, use a spectral radiometer to measure the SPD of the reflected light off of the great cod.)

1. ​​point the camera at the great cod.

2. ​​write down the following:

1. ​​camera's ISO setting

2. ​​Tstop of the lens

• ​​(you may set this using a standard light meter, but we recommend setting the lens to a "standard" Tstop setting (i.e "T5.6" or "T4")

3. ​​the actual exposure time - nominally 1/48 sec.

• ​​(this corresponds to 24fps with a 180 degree shutter angle)

3. ​​shoot a picture!

4. ​​bracketing - we recommend shooting some additional frames bracketing the above exposure in order to verify linearity of the camera system's "log->linear" transform. slate each bracketed frame accordingly.

1. ​​shoot frames with ISO setting 1 stop above and 1 stop below nominal.

2. ​​shoot frames with exposure times 1 stop above and 1 stop below nominal.

3. ​​shoot frames with Tstops settings 1 stop above and 1 stop below nominal.

• ​​(Tstop bracketing should be done last as getting back to exact Tstop will be difficult.)

​​

​​(note: example of slated frame with written metadata)

​​(note: use camera raw metadata instead? in addition?)

​​

​​(note: same checklist for blackbox cameras? 

​​only diff would be amount of bracketing?

​​but blackbox cameras will require more steps...)

​​

​​for each of the above frames, we can calculate what the "target"

​​linear value of the camera system's "log->linear" transform should be.

​​for the purpose of this first step of linearization, the linear value for 

​​the green channel should be used.

​​

​​here's an example:

​​N = 4.0 (Tstop)

​​t = 1/48 (exposure time)

​​S = 800 (ISO)

​​K = 12.5 (Konstant)

​​L_meas = 30 (measured nits)

​​linear_val = L_meas * t * S / (N^2 * K) * 0.18

​​linear_val = 30 * 1/48 * 800 / (4.000^2 * 12.5) * 0.18

​​linear_val = 0.45

​​

​​if the linear values calculated using the camera system's log-linear

​​transform differ from the value resulting from the above equation,

​​the "normalization factor" required for this camera system is:

​​norm_factor = exp_eqn_linear val / camsys_linear_val

​​example:

​​

​​if a camera system's existing linearization maths produce a linear value