Auto Tone doesn't process bracketed photos to the same final exposure

[Ellarsee]

I'm currently at LrC 13.1, this has been going on for several versions, but might be getting worse.
When shooting I tend to autobracket (usually two stops) just to make sure I get a frame that doesn't blowout.  I would expect that autotone would add two stops to the exposure of the under exposed over what it does for the nominal, and subtract two stops from the over exposed frames.  However the under exposed frames of a bracketed set always come out at least one step under, and the over exposed always come out one step over.
Why does automatically correcting the exposure of otherwise identical frames not give me (apart from blowouts etc.) nearly identical final results?

[JohanElzenga]

I assume that is done on purpose. If you clearly underexposed the images, then Lightroom takes this into account. Apparently you wanted a darker image. Have you tried 'Match Total Exposures'? That should do what you expect.

[JohanElzenga]

Come to think of it, I usually do not see that at all. Quite the opposite in fact. If I shot a series of normal exposures and also a series of HDR brackets, and then used Auto Tone on all these photos, then I often have to check the metadata in order to find the HDR brackets again. All photos look so much the same that I can't find them by simply looking at the photos.

[johnrellis]

[This post contains formatting and embedded images that don't appear in email. View the post in your Web browser.]

I did a quickie experiment with 9 bracketed photos with 0.7 EV increment and observed similar behavior of Auto Settings -- the Auto-adjusted images didn't come close to having similar effective exposures:

The original 9 photos had exposure values ranging from 0.65 to 5.56, while the Auto-adjusted photos had effective exposure values ranging from 1.36 to 3.36.

When you understand how Auto Settings works, there's no reason to expect that it would produce outputs with similar effective exposures. Auto Settings uses a machine-learned algorithm ("AI") that was constructed from a training set of thousands of photos that were adjusted by many "expert" photographers to their particular tastes. There's no particular reason to think that the collective tastes of those photographers would tend to adjust the bracketed photos to have the same effective exposure values. It may be that the photos presented to those photographers were not particularly over- or under-exposed, or it may be that the "expert" photographers tended to preserve the dark appearance of dark photos and the light appearance of light photos.

As Johan indicated, the command Photo > Develop Settings > Match Total Exposures will adjust all the selected photos to have the same effective exposure as the most-selected photo: