The HDR gain map export capabilities in LR and ACR are very helpful to share HDR images which still look great on SDR displays (https://gregbenzphotography.com/hdr-images/jpg-hdr-gain-maps-in-adobe-camera-raw/). However, they do not offer the highest quality SDR experience possible due to limited artistic control over the base image (ie the gain map specification itself is fine, but we lack full control over the SDR rendition when exporting the image).
The 7 sliders in the "preview for SDR display" controls in LR / ACR do not enable artists to fully optimize the SDR rendition in a gain map. In many cases, higher quality SDR results can be obtained through other tools - but we still need a way to use a user-generated SDR image in an HDR gain map.
Having full control over both the versions of the image in a gain map provide several benefits:
Superior and more consistent resuls when the HDR is created by upgrading an SDR edit. This is a crucial workflow to help upgrade existing edits to HDR, support workflows which are designed to optimize for print but also support HDR, or avoid limitations of 32-bit Photoshop (such as using liquify and frequency separate for portrait retouching, as they are not available in 32-bits). In all these cases, a perfect and optimal SDR has already been created and a beautiful HDR conversion is easy to create. However, the existing tools result in a degraded SDR after the round trip to HDR and then back through the preview for SDR display. This loss of quality is entirely avoidable by simply using the original SDR edit in the gain map.
Full use of all ACR controls for tone mapping original HDR edits to SDR (ie those available in 32 to 16-bit conversion). This includes local adjustments with sky masks, color-specific corrections, etc. These SDR results are often superior than those available with only the preview for SDR controls.
As HDR support increases in the future, it may be optimal for an artist to encode a gain map built with a base image which is slightly more capable than just SDR. For example, it may be better to encode the image with a base image using 1 stop of headroom (rather than SDR) when exporting an image with 4 stops or more of headroom in the full HDR version. This would require gain map formats beyond JPG, but demonstrates that such user control may become even more useful in time.
Adding an option to specify an SDR source layer in Photoshop's native save dialog or "Export As" panel would of course be nice to have, but is not necessary that Adobe does all that work. A simpler development effort focused only on back end support would still offer significant value (by relying on independent developers). It would be very useful just to add a UXP BatchPlay API which allows 3rd-party developers to specify a layer to use as the base image when exporting an HDR gain map.
The interface and most input validation / edge case management could be left to developers. This would be fairly straight-forward as the base/SDR layer would already have the same document dimensions and color space as the full HDR image. Transparency in the base layer should be ignored. Invalid gain scenarios (such as requesting a bright HDR result from a dark SDR base pixel) could simply clip to the maximum gain to manage edge cases that cannot be encoded.
This request is of course predicated on either Photoshop adding support to save HDR gain maps or enabling an API for developers to request such an export through ACR. Native support in PS is probably ideal for several reasons, including performance. Such support would also add significant value by enabling batch exporting of HDR images for the web (through scripts or actions leveraging a native PS save capability).
Gain maps are already a tremendous achievement, and providing the artist full control over both versions of the image would help fully realize the vision to optimize these images for display on any monitor.