No, that’s what it does. Bet the highlights were preserved…
Most metering is about exposing for middle gray somewhere in the scene. In high dynamic range scenes, this’ll put the upper end past the sensor’s saturation point, to various extents. Highlight-weighted schemes insure that the upper end is below saturation, at the expense of pushing the lower data lower. With my particular camera, I’m finding that the highlight preservation is based on the JPEG rendition, which leaves headroom on the table, so to speak, in the raw. So, dialing in some +EV helps recover that headroom. I’m messing with a +1 EV right now, finding in some cases there’s still unused headroom, in others it goes just a little too far.
I’m finding the filmic curve to be useful in pulling up the low parts of the highlight-weighted exposed image. It’s next to impossible to use curve tools to construct a curve that will lift to the extremes somestimes needed, but still leave a little toe at the bottom to keep blacks crisp. The original filmic equation does this nicely. My current batch processing for proofs uses the filmic curve with the Duiker default coefficients, then I go in and adjust the numbers for the particular dynamic range situation. I just finished committing a batch tool for rawproc that lets me take the modified toolchain for one image and apply them to part or all of the rest of the session’s images; used it to good effect for the first time tonight, processing the shots from our family Christmas tree decorating party.
There seems to be no “one curve to rule them all” for highlight-weighted exposing, so this approach probably isn’t for those who despise post-processing…