Another way is to use a content extractor.

So

• the user will select the rectangular region to be extracted
• Extractor will extract
  • Josh's profile picture
  • "Josh"
  • The tick mark SVG
  • "@jpshycodes"
  • "Bro is coding ..." - the comment text
  • Information that a frame is to be reproduced
  • Ryan Els's profile picture
  • "Ryan Els"
  • The tick mark SVG again, but this time it will be deduplicated
  • "@RyanEls4"
  • "12h"
  • "Rate my ..." - text for the comment inside the frame
  • The jpeg picture inside the frame (yes it's a jpeg and not a PNG. IDK why. But look at it)

Then it would convert it into a reproducible package which can then also match your colour scheme for background colour etc.

Now just need to make such an extractor
And a corresponding format in Lemmy to display it

Sure it will be more work, but this way we can retain more fidelity with less space taken.

It will not be usable as proof, because the thing can be just written, but it's not like pictures work as a very good proof nowadays, anyway.