Download
1 variant available
bf16 SafeTensor
Kaa_eyes__ANI__v2_2_epoch_10.safetensors
BF16, good balance • 33.14 MB
Verified: 10 days ago
155
1.6K
License:
AnimaThe Anima Model is licensed by CircleStone Labs LLC. Copyright CircleStone Labs LLC. IN NO EVENT SHALL CIRCLESTONE LABS LLC BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH USE OF THIS MODEL.
Built on NVIDIA Cosmos
XL/IL model that is trained on output images model Spiral-Only Eyes / Spiralwash Eyes - Hypnosis Concept LoRA.
Anima notes:
This version is the most successful in my opinion. It also works in text2image, but in my opinion, inpaint has a higher priority
Recommendations (ANI):
Denoising strength: 1 (0.75-0.9, without LLLIte inpaint)
Size: (1024x1024)
Strength: 1
IL UPDATE NOTES:
The model under IL turned out to be extremely situational.
Try Tensor art AI tool with this image2image lora: https://tensor.art/template/870531431695165739
Recommendations (IL):
Denoising strength: 0.7 (you can try values 0.5-0.8)
Size: (1024x1024)
Strength: 1 (0.8-1.3)
Token strength: 1.1 (1-1.3)
XL Notes:
It is recommended to use in inpaint mode manually, or using ADetailer.
Version 3 improved performance of Lora in text2img. But I recommend using hires. Added kee eyes token. If you use Lora, remove tokens like bad eyes, Error eyes from the Negative prompt.
Version 2 works better in text2img. But I recommend using hires
Version 1.0-g is trained on the upper parts of the body of characters from one very famous anime game. To avoid a strong influence on the character, I recommend using ADetailer.
To edit only the eyes, you can install this detector model Eyes detector.
Recommendations
Denoising strength: 0.6 (you can try values 0.5-0.8)
Size: (1024x1024)
Donate | More Content