Auto bracketing question

[garry23]

@A1ex

I'm playing around with a few ideas and trying to understanding the 'limits' of the ML auto bracketing.

From my experiments it seems that the shadow end gets evaluated at an equivalent of setting the ETTR shadow SNR to 4Ev.

Is this a reasonable assumption?

In other words, using ETTR to do my bracketing in a script I could get an extra 2Ev movement, i.e. By using 6Ev in the shadow SNR.

Cheers

Garry

[a1ex]

Auto bracketing was implemented before the raw image buffers, so it meters from the YUV image buffer (where the JPEG is created from).

From hdr_check_for_under_or_over_exposure, it allows 2.5% of pixels below 20, and 0.15% above 235. Of course, these levels depend on the picture style.

So they are not exactly equivalent, but probably pretty close.

[garry23]

Great. Thanks for the info Alex.

[garry23]

@A1ex

I thought you might be interested in the conclusions of my 'experiments', which I have built into my latest landscape bracketing script https://gist.github.com/pigeonhill/6620121ebe1548eb1bdaa6bea8598fac

The user now has various exposure bracketing options, built around using the ML ETTR functionality:

1. using the current shadow exposure with the code estimating the ETTR for the highlights, with 1Ev or 2Ev steps.
2. getting the code to fully calculate the exposure brackets based on ETTR for the hightlights and ETTR with a 6Ev SNR for the shadows; once again with 1Ev or 2Ev steps.

As the script is for landscape bracketing, I'm not sure how robust it would be in all use cases. But I'm happy with it