Archetype-M: Aria - v0.33 is out!

I've been eyeing this one for quite a while, I thought to wait until the platform incorporates something for JSON prompting first, but alas. This is a really capable model for multiple characters on screen. The problem, however, remains in that the control is very limited to JSON prompting.

Of course, you can still use Danbooru tags and Natural Language. I find them to be incredibly effective.

All that I hope is, since there's a model that could use JSON prompting that is not paywalled like Ideogram, for the platform to incorporate the JSON Prompt Builder for MOSS.

I think I'll try out various resolutions as well, I don't think the mochibeans cost change for whatever reason.

I hope this can be one of your options to generate images on this platform!