If it happened in the worlds most used client OS 20 years ago, it's clearly not impossible it can happen again. Not that much has happened with computers since then.
I remember being like you, believing no virus could come from an image or other data. I have been proven wrong enough times since then. We keep assuming things as programmers and sometimes we get it wrong and then there is a new wulnerability.
using "filename" within the "Content-Disposition" header, you could theoretically trick a user into downloading a non-image file despite the url containing lisa.jpg
I think certain browsers have security limits on the file-extensions you download, which may include when image->"save as" is used.
Don't forget that you can literally concatenate jpegs and zipfiles [header at start of jpeg, but at end of zipfile], so the valid jpeg can also be a valid zipfile.
Combine that with something like Safari's insistence at automatically exploding zipfiles on download, and you got yourself a party.