Ok, I've rewritten nostr:npub102ca8pnhy26vh2akepgr4vleyew65nuzug5val3sycslfe0wruwqzezyfp . it will web backed on https://sinkin.ai/ (SaaS - stable diffusions based models)
Paid users will be able to specify a model (listed on sinkin).
As far as I can see, the images generated are not as good as midjourney, but with a bit of practice and a good model, you can get decent results.
nostr:note1k2a0dzg0feu6qv8fh69e0u3fmklks47r2yzlhrq0u7sdqq2apelskt3u27
lower quality is worthwhile tradeoff for no censorship...
does sinkin allow base image? er... maybe impossible to do w bot prompt
after a look on sinkin, it's possible but I need to implement it on my bot. it's not easy as midjourney.
I mean, for add input image, you will need to add a valid image url. On my side, I need a lot of work to implement it.
+ Even if the bot does not censor, the images created are hosted by me (on inosta.cc). Images can be taken down because of law enforcement. You have to download every image.
