Hypernetworks

[JohnTigue]

It is essentially a style transfer tool in the SD space. It can train models based on some input images:

• Altrepreneur: HYPERNETWORK: Train Stable Diffusion With Your Own Images For FREE!
  • Requires an 8GB VRAM card
• 20 images training

[JohnTigue]

A reddit snippet:

Q: if there’s a way to train the character it self, not the style so I can create the cartoon character in other styles? I’ve tried to create it like an object and face. It really didn’t work.

A: Train hypernetwork or embedding on your character. Generate enough images with that character. Img2img those images into the new style you want Train hypernetwork/embedding on the resulting images. Voila, you have a trained textual inversion or hypernetwork with that character in the style you want!

[JohnTigue]

Kasucast #9 - Stable Diffusion style technique comparison: Hypernetwork vs. Textual Inversion. He comes down on the side of hypernetworks, but also says both together give "very consistent results."