When creating the 'grey target' you could fill it with RGB118. This is equal to L*49.7, and is the general grey value that the likes of Nikon and Canon use for midtone Ev0. Just don't fill it with Photoshop 50% grey, as this is too bright, and can lead you astray with your highlight Ev range.
I'd recommend using Rawtherapee instead of Lightroom/ Adobe camera raw for this test, as it is possible to ensure you are disabling all highlight reconstruction, view the raw histogram and have a near perfectly neutral raw process, unlike Adobe. You can even disable color transforms in Rawtherapee, noise reduction, variances in demosaicing, hidden processing, ect, so you are just seeing the pure log encoded image of what your sensor is recording.
Interesting exercise; I love its simplicity. One thing I will point out though, displaying a 49%L image on your monitor is completely arbitrary, as it's a function of the brightness of your monitor. And, for the purposes of this exercise, any sufficiently bright, neutral, and evenly illuminated surface/source would work just as well. I would argue that bright white would work even better, as it would shift the useful dynamic range towards faster shutter speeds and less noise, since sensor noise is cumulative. I would also use the camera's own metering to determine the zone 5 exposure then work my way out from there, skipping to say +/-3 before going in 1/3 stop increments to say +/-5 or 6 if you're feeling lucky. That's only 14-20 shots per round.
Thanks, yes, it's simple, and very effective. Though, L*49 IS NOT ARBITRARY in any way...L*49 or L*50 bracket RGB 118 grey, and this value is what your camera meter is programmed with as middle grey by the big two camera manufacturers, at least. Some camera meters used to be calibrated as low as 112RGB. L*49 equates to 116.4 RGB and L*50 is 118.9 RGB. Photoshop will not allow for decimals in either mode, but of course, you can set it to 118 exactly if you wish, as I say in the video. 118 RGB is L*49.7, so your arbitrary choice is 'either or' not 'anything you like'.
