Originally Posted by spacediver
Thanks for the clarification. I don't mean to be stubborn, but why is the inverse camera gamma not relevant?
Because the image the viewer sees is the product. The process used to produce that image is interesting, but sort of irrelevant. Once the image is stored, it's now a screen-referenced image. We already know we're not trying to recreate the exact lightness curve of the actual scene, which is why BT.709's gamma isn't an inverse of the CRT gamma. So the actual light values of the original scene are no longer relevant; only the light values that we expect to see on the display.
If we were trying to recreate the actual scene viewed by the camera, we'd make sure the encoding and decoding curves were exact inverses of each other. We didn't, because we realized that having a positive end-to-end gamma looks better. It's kind of a poor man's appearance model.
The important curve is the display curve. The camera curve needs to be built with an understanding of the display curve, but not the other way around. In a sense you can think of the camera curve as the inverse of the display curve, plus some "special sauce" that fixes and adjusts some things for aesthetic or technical reasons.
For content generated originally in a computer, the camera curve is completely irrelevant. That content was generated, adjusted, tweaked, etc. on a computer screen, and you just want it to look right on the destination screen. The aesthetic tweaks built into BT.709 do not need to apply. The content was generated in the same environment that it's going to be viewed. So in that case, the inverse of the display curve is the relevant curve, not the camera curve. There really is no realm where the inverse of the camera curve is useful.
Ultimately the SMPTE engineers didn't make a spec for display gamma because it wasn't thought necessary. CRTs just were what they were; no gamma spec was considered important. The camera, on the other hand, needed a bunch of circuitry to make the image look good on the display, so that needed to be standardized.
Now, we're in a world where content comes from lots of different places besides a flat-lit studio, and displays have all kinds of different inherent electro-optical transfer functions, so a display standard is needed. And it makes sense for that standard to be based on what a CRT did, because that was the de-facto standard for years.