Towards the end of this discussion (
Optical Measurements of Pentax ME Super Shutter Speed - PentaxForums.com , forum member swanlefitte notes
Originally posted by swanlefitte What do these models report as half steps? I notice my ks-2 uses f/4 f/4.5 f/5.6, not f/4.8.
This prompted me to check my K-1 and K-3. They, too, display the (incorrect!) value of 4.5 for the step selected between 4 and 5.6 when in half-step EV mode.
What gives?! Is this just a mis-labeling or do these cameras actually shoot at the incorrect value? Let’s find out (I’ve got nothing much better to do during these days of stay-at-home orders).
So, off to take some photos of my Kodak gray card with my K-1 and the SMC Pentax-DA 50mm 1.8 lens across a range of f-stops. I chose the 1.8 lens so that I could measure over the widest range of stops.
I took frames at half-step and third-step values from f/1.8 to f/22. On the way up and down, the full-step values were shot 2 or 3 times. The repeatability is very good (a testament to my lights and power supply!).
I used RawDigger to extract the actual R,G,B pixel values averaged over a small rectangle in the center of the gray card, similar to what I did when I checked camera linearity a while back (
How Linear Are Some Pentax Camera Sensors? - PentaxForums.com)
Just as with shutter speeds, it is important to note that the f-stop values reported for just about all cameras are not quite correct. The numbers have been rounded for convenience. This article has a comprehensive discussion:
https://www.scantips.com/lights/fstop2.html
Here are the results, plotted with the (expected/hoped for) mathematically correct aperture values (a lot of fourth and sixth roots of 2 involved) - including the correct value (nearly f/4.8 rather than f/4.5) for the half EV step between f/4 and f/5.6
The data are pretty consistent, although not nearly as good as for the shutter speed results referenced above, and with some systematic deviations. The data clearly show some curvature - peaking in the middle, for all three colors. If pixel value versus aperture were exactly linear, the power law fits should have exponents of 2.00 , rather than the somewhat larger values here (typically 2.12 to 2.13). These values suggest that at the smallest aperture (largest f values), not as much light gets to the sensor as it “should.”
All in all, though, I am fairly impressed with the consistency and goodness of the data. The mechanical tolerances in lens aperture at the smallest openings have to be a small fraction of a millimeter. (At f/16, the aperture here is around 3mm (=50mm/16), and a 0.1mm error results in an aperture area (same as exposure) error of almost 7%.)
So, what about the original question: what is going on with our Pentax cameras when used in half-step mode between f/4 and f/5.6?
Here’s the same plot as above, except I have used the exact value of 4.5 for the half-step value. The points (circled) are clearly out of place.
Looks like Pentax has just mis-labeled this value, and it carries over to the EXIF data, too. Maybe in a future round of updates, this could be fixed.
This dichotomy appears here in the Forum article The Fundamental of Exposure (
The Fundamentals of Exposure - Aperture - In-Depth Articles), where on page 5 the half-step values list F4.8, whereas the animated GIF on the same page has the lens stepping through F4.5 !
What value do you see for your camera(s) (including non-Pentax?) for this mystery stop? Does this error go all the way back to the first digital bodies, or even back to the film SLRs?