From the mailbag:
Nice A7II IBIS test. It would be great to compare to the A7 sans IBIS. A comment and a question:
It looks like the A7II has a stronger AA in one direction (H), similar to the A7 and D6x0.
Assuming you are trying to evaluate the capabilities of the (IBIS) hardware, why introduce unknown/arbitrary software parameters into the equation (LR) instead of simply rendering to TIFF via dcraw -d -4 -w and feeding that TIFF to Imatest?
With this latter approach you would need to properly set white balance in-camera (or via -r in dcraw) but you could then indeed make quantitative MTF50 comparisons to the sensor’s pitch related limit. If on the other hand LR is part of the equation, results become more qualitative because one never knows how much of the MTF50 is simply due to sharpening by the adaptive rendering algorithms.
WRT the comment: bang on.
About the question. What you are recommending is indeed the high road, and, if I ever tried to do absolute testing of lens performance, I’d certainly consider taking it. However, I’m interested in posts like yesterday’s only in relative performance, and I don’t see how using Lightroom’s default settings with their mild sharpening would affect relative rankings.
Using Lr has some things to recommend it.
First, it’s fast and easy, letting me pick images, form sets, label them with keywords, crop to save disk space, write out with descriptive file names, etc in an efficient manner. I will stipulate that I’m lazy. Well, not lazy exactly; I hate to do work that seems unnecessary. Let me also stipulate that I am prone to error, and that I hate to make errors. The Lightroom workflow that I’ve worked out with its multiple crosschecks, helps me keep from making too many errors and having to redo too much work.
Lr is also indicative of the tool that a typical photographer might use to develop raw files, and their default settings appear to me to be well thought out.
But thanks for the suggestion.