I'm slightly confused about the difference between the “enhanced” image that is saved at the end of an observation plan (or a sequence of captures), and the one you can obtain by applying Deep Sky Stack afterward (I've read the tutorial, but I'm still unsure).

I mean: even though Seestar says “Enhancing”... isn’t what it’s actually doing during stargazing real-time stacking, or am I wrong?

Sure, in Deep Sky Stack I can choose which frames to include, but are the ones made available to me actually the same ones that were used for the enhanced image?

Can someone help me understand?

Attached: the image with 107 minutes is the result of "Deep Sky Stack" having not used a few subframes that in my opinion were bad enough.
The image with 119 minutes is what was left after the complete plan execution, let's say the "default" enhanced image.
Both have been cropped, AI denoised and saturated, all in Seestar App, no external processing. To my eyes, surprisingly enough, the Deep Sky Stack is much worse...