Sign In

Cranberry cookie (Cookie run)

58

Updated: Aug 2, 2026

character

Download

1 variant available

bf16 SafeTensor

cranberry cookie anima 5.safetensors

BF16, good balance • 87.6 MB

Verified:

Type
LoRA
Stats

66

24

27

Reviews
Published

Aug 2, 2026

Base Model

Anima

Usage Tips
Strength: 1
Hash
AutoV2
DA3F58DA9B
Trigger Words
cranberry cookie, chibi, red hair, solid red eyes, hair bun, head accessory, hair over one eye, red dress, black skirt, puffy sleeves, colored eyelashes, white gloves
default creator card background decoration
Followers - 1669

1.7K

Likes - 15989

16K

Second Birthday Badge

License:

Anima

The 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

ComfyUI_14874_.png

Prompts: cranberry cookie, chibi, red hair, solid red eyes, hair bun, head accessory, hair over one eye, red dress, black skirt, puffy sleeves, colored eyelashes, white gloves

weight: 1

Note: Both versions has a bit of a background issue sadly, I will look more into it later becuase I think I know what the issue is.


A request, I never expected for somebody to ask for a Cookie NPC (and one of my favorites at that) but here we are.
The thing is, there just isn't enough images of her to make a good lora. So let me summarize how I made this lora:

  1. Grab the only 2 usable images from the wiki

  2. Train an anima lora (I think the requester was hoping for an IL lora but trust me on this one)

  3. Generate a few images, this is quite hard because of the limited dataset but if you play with the weight and prompts you can make new poses and even outfits if necesary

  4. Grab the best generated images, put them on your dataset folder, create a note pad file and rename it, then grab the prompts you used and copy them in the note pad.

  5. repeat steps 2 through 4 until you have a working lora

TL;DR: Classic "AI generated dataset" BS

I was shocked at how well Anima was able to make the character only using 2 images and I could use the same method in the future but it was a lot of work.

And if you are wondering about the dataset itself, I made 14 images and they are more or less the ones I used for the preview images (that's why there are a ton of them)