The medium supports images so I’m not sure this critique is valid.
I do agree with the overall point though that images are a suboptimal way of conveying text.
They shouldn't have to do OCR, a description of the image should be provided, there is even an attribute for that on the web, it's called "alt", or use picture/caption. Twitter and its shitty UI has no excuse.
Given the amount of hacks from twitlonger, to unrolling threads, to audio I feel it is simply annoying at this point