By using ausbit's bot, you are losing the ability to influence the generation to your liking. Stable Diffusion has many tools that are hard to implement in a Discord bot. Like ControlNET etc. For example, using controlNET you can extract posture information from another image and use it to guide the diffusion process to give your new image that posture.
You are viewing a single comment's thread from:
I think there will be an open source version or public version of ControlNET coming out soon that will be implemented into the bot, if i'm not mistaken. Last night, after the post was published, we merged three models into one and the results so far have been astounding. The advancements of all the options and abilities is pretty amazing. Thanks for the info!