Comment by neilv

> It isn't just a simple crop; Pro Res Zoom uses Tensor G5 and an all-new generative imaging model to intelligently recover and refine intricate details.

Unclear what "intelligently recover and refine" means here, and I'd like to make a request of people who might also be unclear yet influenced by this idea...

If you build features that sound like this into the image-capturing stage (from the user perspective), and your target users include serious photographers who care about authenticity, please be sure to make any kind of "generative"/inferencing of image details be optional.

And fall back to nth previous generation of sensor processing, autofocusing, and anti-shake technology -- where the compute still influences the images, but errs on the side of missing/fuzzy detail rather than fabrication.

Imagine the priorities of photojournalists as a category of user -- not only users who want appealing snapshots, selfies, or professional editorial fashion/lifestyle shoots.

There's still a place for consciously fabricating/enhancing/fixing in interactive post-processing, when you you choose to do it, and it's clear to you what is being done.