So since Nano Banana 2 was released, I've been experimenting with what new capabilities it has and what they can do in the way of WAM.
The main improvement seems to be that it has a better idea of cause and effect than earlier tools. When I asked it for someone who's slipped down a muddy riverbank, it added a plausible slide mark where she'd fallen. Screencap sequences are another strong point; it can maintain continuity quite well if you ask for a sequence of events, and generate subtitles if you ask it to. This also works for comics - with those, once you have one page you can feed it back in with a simple prompt "Draw the next page" or "Draw the previous page" and sometimes you'll get sense. If you actually want a coherent plot you'll have to give it more guidance, of course.
NB2's fairly good at adding gunge to an existing image. I had a comment on one of my previous images that the poster wished the clean woman had got the gunge too. Well, now she can
Another thing I've used it quite a bit for is pretending something is from a real show, by having NB2 generate a 'behind the scenes' article about it. So generate the image, then feed it in with the prompt "Page from TV listings magazine. Interview with the actress who plays Marcie" and there you go. Or pretend it's something that happened by turning the photo into a newspaper article, or a TV news bulletin. Not getting enough comments? NB can turn anything into a social media post.
It seems to be able to get gunge into a hazmat suit that someone's wearing over their clothes, which previous tools had trouble with. It's no better at getting people correctly into stocks / pillories, though.
Love you, too
