Keeping pick/reject separate from star ratings is generally more flexible.

Flags and ratings often serve different purposes:

• reject can mean “delete this” during the first cull
• pick can mean “selected by client” or “needs further retouching”
• stars can represent quality, favorites, publication status, or relative ranking

Because those meanings are different, a photo can legitimately be both a pick and a 3-star image, or even a rejected client choice that still has value for later reuse. If flags were cleared whenever stars were assigned, those workflows would break.

Another practical point: in some DAM systems, star ratings can be written into image metadata, while flags may live only in the catalog/database. That also argues for keeping them distinct.

Your proposed model is simpler if you want one linear scale, and some users may prefer that. But making flags and ratings mutually exclusive would reduce search, filtering, and workflow options for people who use them for different kinds of information.

So the main tradeoff is:

• single combined system: simpler, less ambiguity
• separate systems: more power and workflow flexibility

Most answers favor the current independent approach because users can still choose to treat flags and ratings as linked, but not the other way around.