Download
1 variant available
2000 1 2 3 4 5 6 7 8 90 1 2 3 4 5 6 7 8 90 1 2 3 4 5 6 7 8 9
(26)
Jul 17, 2026
Anima
LoRa trained for an additional 20k steps (60k total) on an updated dataset with more fleshed out captions and some quality tags, then weights were refined in ComfyUI. Should also converge in a few less steps compared to v3.0.
Show more

2080 1 2 3 4 5 6 7 8 90 1 2 3 4 5 6 7 8 90 1 2 3 4 5 6 7 8 9
4990 1 2 3 4 5 6 7 8 90 1 2 3 4 5 6 7 8 90 1 2 3 4 5 6 7 8 9
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
Built using the a modified version of the workflow I used for the Illustrious version of uwumerge, this is the Anima version of uwumerge! The default style is influenced by the Anima version of my uwustyle LoRa.
How to prompt:
To get the intended style, use E621 & booru tags such as anthro, furry, male, dog, solo plus quality tags, and natural English prompts wherever you need more complexity. The training data is mainly E621 tags with a small amount of booru tags and natural English in varying levels of verbosity.
The model is also good with prompts describing backgrounds, positioning of things in the image, and text within quotes, including controlling the font size and color.
Be warned, this model does not know as many species as previous versions of uwumerge, but it should know the most popular species, and as a fallback you can use some of the family tags such as canid, felid, etc.
Since this model is meant to generate cuter styled content, for some prompts you may want to add young to the negative prompt to get more adult-looking characters.
