Mimic styling of artist "1888" or "1.123"
For Anima, trained locally withing ComfyUI using https://civarchive.com/articles/29356/training-anima-in-comfyui
The model is surprisingly weight responsive (in a good way) !
with 0.8 you'll get the styling with sharp line bold lines, and you'll get the whole characters vibe at 1.0 !
For the illustrious model:
Trained locally using "VETE" base training checkpoint
The artist like to mix sexy moment with mundane scene of violence and destruction.
About V1A dataset and training:
Has low amount of violence and destruction, but can still make proper explosion and concrets shards.
Balanced amount of characters, it can draw male and female properly.
Many characters (witches) are in, but not properly tagged yet.
Work fine with popular ill models (wai v15 v16, plantmilk walnut, VETE base lora training chktp)
For some reason use only low weight (0.6 or 0.7) if you using hassakuXL illustrious v34 otherwise the picture will be bad :( .
Some vehicle are in, but don't expect anything good.
good effect start at 0.8.
Enjoy :)
Some pics at perfectly SFW but civit analyse tools is acting like a bitch ...
Description
Lora trained in about 2048 steps with comfyui anima trainer
Same dataset as previously