I know there's a desire on PPCG for challenges to be flexible in terms of what they allow for input and output. However, I think there is an argument to be made that ascii-art is a special case, and should be stricter about the permitted output formats.
As a case in point, there was a recent ascii-art challenge that involved outputting dominoes. In the challenge description, the sample output looked something like this:
o|o o | o o |o o
The top answer, at the time of writing, was returning output like this:
[[[False, False, True], [False, 0, False], [True, False, False]], [[True, False, True], [False, 1, False], [True, False, True]]]
I find it hard to believe that anyone looking at that could honestly describe it as ASCII art! The second most upvoted answer is a little better, but not by much.
Now I don't mean to criticise these answers, because the challenge made it quite clear that that sort of output was considered acceptable. But what I'm proposing, is that maybe in the future we should be discouraging challenge authors from being so lenient in what they accept, assuming the question is tagged as ascii-art.
As for what exactly should be permitted, at the very least I'd expect the output to include the actual characters of the artwork that was specified in the challenge. Outputting to stdout, displayed on screen, or returning a string (typically newline-separated) all seem OK to me. Returning an array of strings, or an array of characters, feels perhaps like it's getting too lenient again.
But that's the point of this question: what is the consensus on output formats that should be allowed for ascii-art?