Recommendations (2.0 - Anima)
Two versions for Anima Base v1.0:
v2.0 - Anima: a glowing crystal on a chain. Trigger:
hypnosis crystal. Weight: 0.8 (0.8–1.0)v2.0 - Anima (Pendulum): a hypnotic pendulum (a coin on a string). Trigger:
hypnosis pendulum. Weight: 0.8 (0.6–1.0)
Settings: 30 steps, CFG 4, er_sde / simple, 1024×1024. Useful tags and limitations are in each version's notes.
Recommendations (1.0)
Weight: 0.6 (0.4-1)
Recommendations (0.1-beta)
Lora for generating a crystal on a rope. Works a little strange, but can be used in img2img
Weight: 0.6 (0.4-0.8). Lora can influence the style, so if possible, turn it off at 40% of the generation. This can be done using a plugin https://github.com/cheald/sd-webui-loractl
Size: (1024x1024)
Description
Anima Base v1.0 version of the hypnotic pendulum (a coin/disc on a string), trained on a 58-image dataset. For the crystal, use the v2.0 - Anima version.
Trigger: hypnosis pendulum. On its own it already gave a pendulum (2 of 2 tests); action tags make it reliable.
Weight: 0.8–1.0. 0.6 still works (2 of 2); lower wasn't tested.
Useful tags (from the training captions):
solo, holding pendulum, pendulum in front of face— she holds it herself (8 of 8 at weights 0.6–1.0)pendulum other, dangling pendulum— hangs from above; a hand usually shows at the frame edgeadd
disembodied handfor a clear hypnotist's handswinging pendulum, pendulum in front of face— swings right before her eyesdazed expression, hypnotized, half-lidded eyes— the trance look
Tested with: 30 steps, CFG 4, er_sde / simple, 1024×1024.
Negative: worst quality, low quality, score_1, score_2, score_3, blurry, jpeg artifacts, sepia (add nsfw, nude for SFW).
Notes: indoor scenes, full body and style presets work. It sometimes adds small floating "?" or swirl marks picked up from the dataset. Two characters weren't tested.
DiT-only LoRA (no text encoder weights), dim 32 / alpha 32, 1500 steps.






