Is there a scenario where the "blur then sample" technique (for a different image) makes aliasing worse? I'm thinking of a scenario where the original image is of very high frequency and blurring brings down the image frequency, but not enough s.t sampling still encounters aliasing but at a worse rate due to poor luck (sampling rate vs frequency mismatch).

Similarly to Danny's question, I was wondering if scenarios exist when sampling then blurring would actually end up with a better result than pre-filtering, is it viable to switch to using the other technique for a better result, and if so how would one determine when to switch between techniques/where to apply each (threshold for "lots of high frequency", switching for just a small section that has high frequencies, etc.)?

When we blur then sample, how do we know how much to blur? I.e., when filtering out the high frequencies in the frequency domain, how do we know how much to filter?