This LoRA is designed for creation of situations, where a character shakes her or his hair. It works with liquids (water, blood, honey, etc) and without them.
Prompt examples:
1girl, solo, very long hair, hair swish, floating hair, wet hair, splashing, arched back, head back, from side, ocean1girl, solo, long hair, hair swish, floating hair, big hair, (head tilt:1.3), translucent hair, looking at viewer, parkActivation tags:
hair swish, floating hairLoRA works good enough with a strength of 1.0. If there is no effect, please, increase the strength of the activation tags.
If anime/cartoon/furry pictures are too realistic for you, then put "realistic" in negatives.
Changing the type of liquid is kinda hackish. You should put to prompt something like this:
blood, (red water:1.3)Hair type tags:
wet hair
big hair – windblown hair
translucent hair – the hair is illuminated from behind by sunlight
Hair length tags:
long hair
very long hair
absurdly long hair
These ones may not work well:
very short hair
short hair
medium hair
Character pose tags:
head back
head tilt
arched back
leaning back
leaning forward
partially submerged
wading
afloat – places the character waist-deep in water. This tag seems to be broken in Pony
Other useful tags:
splashing - adds a stream of water
motion blur
You can control the camera with default tags:
straight-on
from side
from behind
from above
from below
perspective
etc.
as well as their combinations: "from side, from above, perspective".
Of course, "looking at" do work too. Like these ones:
looking at viewer
looking up
looking to the side
etc.
Description
Trained on NoobAI Epsilon-pred v1.1.
FAQ
Comments (7)
You used Epilson 1.0 or pred1.1 for Training?
Seems like it works better than Illustrious 1.0 even on None NoobAi models.
Trained on Epsilon-pred 1.1. v1.0 is a version of LoRA itself.
Yes, for some mysterious reason it works great with Illustrious models, including 0.1. Better than a version trained on Illustrious 0.1. Oh, I just realized that Illustrious now has a version 1.0 too. It's all so confusing...
@PalmtopTiger yeah too many versions sometimes crate confusing. They now have 1.1
2.0 and 2.1 of Illustrious will be released few months later I guess
@PalmtopTiger btw make LoRAs on Bun Drop, Hair Brushing, Hair Oiling and Towel Dry too.
Those were big trends once a time on Instagram. Didn't thought it was possible to do it properly on Illustrious.
@TatsuyaAI I'm sorry, but I can't. My main job also computer related. I'm not young myself, and I can't afford to sit in front of a computer for longer than required.
This LoRA has not come easy. A week of selecting photos, removing watermarks, cropping and tagging. ~250 photos, tagged mostly manually as the autotaggers miss a lot of detail. That's probably why LoRA works so well.
Maybe I'll do 1-2 more LoRAs of other themes in the future if I'm in the mood.
Thanks for your interest in my work. I appreciate that.
@PalmtopTiger oh it's ok. I thought you're in your 20s or 25.
All of the images were selected realisitic?
But thanks for making this one ♥️
@TatsuyaAI Mostly photos, yes. About 20 are drawings and anime screenshots.