This week I’ve been reading The Once and Future King by T. H. White, but I’m not done yet.
Ran the second session of my Old School Essentials Dolmenwood game this weekend. The final player, who hadn’t been able to make it to sessions 0 or 1, joined the group and finished their character: a Magic-User. I was very much looking forward to this (Dolmenwood readers will know why), so that’s already shaping up nicely as a potential source of adventure.
This session they managed to resolve the whole “party member rotting while alive” issue they brought upon themselves last session, in the process making a little money, doing a little exploration, and saving the life of an injured elf, who then dropped some lore on them. Then just before the end of the session they saw a mysterious little figure put up a “help wanted” sign on the notice board of the inn they’re staying at.
We’re only two sessions in, so it’s impossible to judge how things will pan out, but so far I feel that Dolmenwood is making it really easy for me to run. There’s so much going on in the setting that I can just give them a rumour about something, and then on their way to investigate that they’ll stumble upon multiple other things to get involved with as well. And yet the setting isn’t so dense that it’s hard to run at the table, for example each hex has a single page description; and nor is it so dense that I can’t make things up and slot in my own adventure sites.
It’s definitely a well-written setting, and I look forward to it being complete and the kickstarter for nice printed copies.
It’s pretty cool, but also not very easy to get good results out of. Most stuff is ok, but a lot is bad. I think I’m getting better at it, I’ve learned some tricks: name the art style, like “digital art” or “oil painting”; write a short sentence rather than a list of keywords; but don’t write too long of a sentence or that’ll confuse it… but even so my success rate isn’t great.
For example, the stag made of blue smoke in the Twitter thread was a total accident, but a really cool one! But I’ve not been able to come up with a prompt to get it to generate anything else in that style.
Finer details like faces are kind of wonky too, especially where something overlaps a face, it doesn’t quite know how to combine them. I wonder how much of that is a limitation of the model, and how much is because the amount of image generation time per prompt is pretty short.
In any case, I’m looking forward to seeing where AI art goes next.