This LoRA creates an afterimage effect, making parts look like they are rapidly moving back and forth between two positions.
In v0.0.1, you can expect the effect to work reasonably well with basic positions, but it is currently a bit finicky and highly experimental. Due to gaps in the training data, the effect rarely works for fellatio or handjobs (these are included in the dataset, but with so few patterns that they are barely reproducible). It also completely fails to activate from a side angle (from side).
I plan to address and fix these issues in the next version.
Description
head_v0.0.0 (Initial Release) Trained on anima_baseV10. The dataset consists of 20 images total: 10 POV-style illustrations featuring up-and-down head movement (pov), and 10 side-view illustrations featuring left-and-right head movement (from side).
Important Notes for the Initial Release:
Side-View Angle (
from side): The afterimage effect for side-view oral sex will not stabilize unless your latent/generation resolution is set to a landscape (horizontal) aspect ratio. This is because most of the training images for this angle were horizontal.POV Angle (
pov): Conversely, the POV training images were mostly portrait (vertical), so you will need to set your latent/generation resolution to a vertical aspect ratio for the effect to work properly.Model Compatibility: On models other than
anima-base-v0, you will need to boost the weights of the trigger words to make it work. For example, try using something like(2way-afterimage, afterimage, motion blur:2).
dim16/alpha8, unet 1e-05(lion8bit/cosine_with_restarts[2]), 20images / 1000 steps