Our favorite big titty yandere :)
Used kl-f8-anime2 vae for AOM3 model.
Description
READ BEFORE USE:
I should have separated this into 8 different LoRAs for 7 skins and 1 general LoRA where the data didn't belong to any skin, but I didn't.
As a result, this LoRA although convenient to use, will not be nearly as accurate if I chose to do a separate LoRA for each skin. This is only good for general concepts, do not expect to get accurate results like St. Louis luxurious wheels.
I might retrain for individual skins for way better accuracy if this gets popular enough :)
Note: since a lot of her images are her with something on her hair, you'll often get hair ornament like things. You can use a non-latent upscaler to try to avoid this, prompting in negative doesn't help.
Keys (to put as first tag):
taihou - Taihou in general (if you wanna see her in Christmas outfit etc.)
taihoudefault - Default
taihoumuse - Muse
taihouwedding - Wedding
taihouforbid - Forbidden Fruit
taihouphoenix - Phoenix's Spring Song
taihourq - Enraptured Companion (Race Queen)
taihourqsmall - supposed to give her the small bikini verison of Race Queen skin, very hard to get it working, but you can try :/
taihousweet - Sweet Time After School
FAQ
Comments (10)
I have added you on Discord, as I have some questions! Thank you!
I'm having trouble reproducing the examples. When using the stable-diffusion-webui, the model is AOM3 correct? Then how is the kl-f8-anime2.ckpt used? Also why is the lora not part of the prompt? I can't get it to work without including the lora.
Hm, my response is a month late, but better late than never I guess.
Model was AOM3, kl-f8-anime2 vae should be available on the internet, how is it used? You have to lookup onw how to use a vae for Stable Diffusion. LoRA is not part of the prompt? Then I must have used the old way to trigger LoRAs, which is through additional networks instead of the modern way from prompts. Triggering from prompts vs triggering from additional networks will give different results even if everything else are the same. The trigger words I've listed are still correct and should be used.
Very nice,Love this lora so much~~
Thanks, leave a 5 star maybe ;)
@Bloodsuga shut up and take my 5 star XD
nice robustness!! i search from your lora information and train my lora that gains better one:)
Hope to receive updates.
Details
Files
taihou-epoch5.safetensors
Mirrors
taihouAllSkinsAzur_v1.safetensors
taihou-epoch5.safetensors
taihou-epoch5.safetensors
taihouAllSkinsAzur_v1.safetensors
CH-Taihou_all.safetensors
taihou-epoch5.safetensors
taihouAllSkinsAzur_v1.safetensors
taihou.safetensors
taihou-epoch5.safetensors
taihouAllSkinsAzur_v1.safetensors
taihouAllSkinsAzur_v1.safetensors
taihouAllSkinsAzur_v1.safetensors
taihouAllSkinsAzur_v1.safetensors
Available On (2 platforms)
Same model published on other platforms. May have additional downloads or version variants.



















