Open erinacio opened 1 year ago
I'd say it's better to implement by HTTP API call instead of inference locally (that makes Jotaro a GPU-required program).
Users may run a container to provide Stable Diffusion service if they're running GPU instances, or make use of serverless GPU services like https://www.banana.dev/ .
Describe your suggested feature
I hope new Jotaro can generate new waifu for me according to my description, instead of randomly generated waifu that sometimes don't look great.