The result here seems a bit like what I imagine "Audio Description" for visual media would read like as a transcript. Maybe it would work better if the model was trained on novels, rather than a more general corpus?
What is the point of this? To see if AI can weave a story based on a sequence of stills? If that’s the case I would love to see what it could do with a series of security cameras as it follows a person around a highly surveilled city like London or Shenzen.
0_____0|1 year ago
dyauspitr|1 year ago
mossTechnician|1 year ago
https://www.wired.com/story/nanowrimo-organizers-classist-an...