Yesterday I fed the high-res version of Greenie into DrawThings, and got it to spit out a variety of different versions of him.

-=-=-

Once he was vaguely presentable in a newer form that looked vaguely Greenie-like, I popped the image into ProCreate on the iPad, and drew over it to make a new new high-res version of Greenie, then fed that back into DrawThings, in the hope that it could better understand what exactly Greenie is supposed to look like.


My newest drawing of Greenie, with more rounded eyes, a hairier bald spot, and stupidly stumpy hands.



On previous attempts, it would fail at this step, not knowing WTF it was trying to redraw, but now it seems to understand that he's meant to be a person.
I think that's progress, maybe!?!

Today I'll be drawing a few more frames of the New High-res Greenie in different poses, and then shoving those into, first Textual Inversion, and if not then Dreambooth in the hopes that the resulting output is at least vaguely Greenie like.

Ugh..
All this work, just for a silly picture!
Is it really worth it?!!

... We'll see.