2.00
1.75
1.50
1.00
0.75
0.50
0.25
Fullscreen
AnimationStable Diffusion
Rendered via text-to-image in stable diffusion.

This was rather complicated to produce. The basic steps:

  1. Crop and tag art from the Wetware series

  2. Train hypernetwork

  3. Render frames using hypernetwork and multiple control nets, sourced from a video of Victoria Macreatis

  4. Basic temporal smoothing filter in VirtualDub to reduce flicker

  5. Use segmentation map to mask out the subject

  6. Multiple passes of RIFE to smooth out flickering (render in-between frames, delete original frames, re-render in-between frames, repeat)

  7. Filter the background heavily, since it was all flickering nonsense

  8. Recomposite

Victoria Macreatis CatwalkReference Video
Become a Patron
Comments

Tiger_Driver91 2023-08-07 15:16:434 ♡
JIGGLE PHYSICS :L
Your Quiet Neighbor 2023-08-07 16:32:401 ♡
Whoa!!

Amazing animation that puts the "zing!" in amazing!!!
tygh 2023-08-07 16:49:333 ♡
Perhaps very complicated, but apparently still possible.

It's nice to see at least one artist being able to make good use of AI without giving up on their own skills.
LoveOfO 2023-09-06 14:45:573 ♡
I wish I was built like this T_T
BB-Kiwi 2023-09-30 16:25:392 ♡
Impressive
jlv61560 2023-10-02 00:24:482 ♡
The reference video is pretty freaking hot too.
Posting...