In a prior blog post I laid out a process for developing a set of luminosity curves that emulated rather closely silver halide film. When working more carefully with the Zone System and trying to apply it more accurately to digital work there are a couple challenges I needed to consider.
One is a legacy challenge that comes from the film era that involves both the "toe" (roll-on region) and "shoulder" (roll-off region) of a film sensitivity curve. To solve this, photographers like Bruce Barnbaum would place important shadow detail at Zone 4, instead of Zone 3 (see the original Zone System description). They would then develop for the highlights, and would do these two things to keep 1EV (1 stop) steps even across the linear portion of the curve. This would help keep information from attenuating at the top and bottom of the EV range.
A digital challenge is introduced by the nature of sensors and the way that data is written. I've come to learn about about it in doing this little research project. What would be Zone 9 at +4EV with some tonal value in the old Zone System is saturated pure white in digital. The Zone System description puts pure white at Zone 10 or +5EV. As confirmation and in practice, whites in my digital work have too often felt attenuated and seldom have the same tonal separation as my film work. Now I'm beginning to understand why.
What I'd like to do with this blog entry is to try and solve the film and digital challenges when working with images from my Sony APS-C and Full Frame cameras.
Here is the idealized step wedge for digital. I made this using RawTherapee. Using the fact that 0EV is defined as Zone 5, 7f(hex)/127(dec), middle, or 18% gray, I took that tone and raised/lowered the exposure value in 1EV steps to build a step wedge from -7EV to +5EV. Here is the linear step wedge result. Note the tonal values at each EV step. These will act as my reference points.
Once I had the values at each Zone/EV step from the idealized 1EV step simulation I could then try and match the output of my Sony sensors to the linear ideal. Using the Zone 5 definition, here is the step wedge I built centered around 0EV, middle gray.
This _exactly_ matches the idealized curve which solves the film-era "toe"/"shoulder" problem. For those situations where detail in +4EV is not required, this does the trick. But... it does not solve the +4EV/Zone 9 digital saturation challenge. To do that, I recalibrated -1EV upward to Zone 5, moved Zone 10 to +4EV, and remeasured the output luminosity curve. Here is that result.
As can be seen, the output curve is shifted to the right for better highlight control, while retaining the slope of the linear ideal. The overall curve is now better balanced between the highlight and shadow regions as well. Usable dynamic range appears to be around 11EV with this early Sony A7 Full Frame sensor.
With this curve I believe I have solved the film-era "toe"/"shoulder" and the digital Zone 9/+4EV saturation challenges. I said the previoius film-emulation solution I worked on was "adequate", but I now have a better one.
Voila! an accurate properly linear Zone System adaptation to Sony sensor digital.
--------- Where am I on my checklist of perceived "needs"? ---------------
Is what I'm doing here in trying to transfer old film-based Zone System knowledge into something practicable for digital...
- Measurable - Yes
- Accurate - Yes
- Repeatable - Yes
- Controllable - Yes
- Clearly
understandable - Yes, but readers can judge otherwise if they like
- Reveals what Sony's in-camera jpg processor is doing - Not yet
- Helps me choose a set of in-camera jpg engine operations and/or settings that match my RAW output - Remains to be seen
- Bigger Bonus: I now have a linear digital match to the old idealize film-based Zone System
No comments:
Post a Comment