Support for all of the following for the NovelAI cog
Global: Generation cooldown starting from when the request is submitted to NAI, to control rate better, as users together currently can cause excessive generations. Default is disabled, i.e. 0
Global: DM Allowed to enable or disable generation via DMs entirely. Default is allowed.
Global: Max image size in MBs to control the size of files we load into memory, as due to addition of vibe transfer / reference images, this could become problematic. Default of 25.
User: Ability to set default model, default reference image (vibe transfer) strength (default 0.6) and info extracted (default 1.0), default furry base prompts.
Support for the new Furry v3 model, including it's base prompts
Check image byte sizes before reading, rejecting if too large
Vibe transfer support for up to three images for txt2img and img2img, testing showed this can go up to 10 images at least, but three was reasonable given the limitations of the slash commands. AFAIK, have to define each input in advanced, not aware of multivalue or dynamic inputs.
Made Euler Ancestral the default sampler, as a majority of the time, it is better than Euler.
@hollowstrawberry
Support for all of the following for the NovelAI cog