Sign In

Regional Eyes Anima: Umrbium

0

Not recently loaded

Download

1 variant available

bf16 SafeTensor

eyestyle v1_anima.safetensors

BF16, good balance • 87.66 MB

Verified:

Type
LoRA
Stats

39

0

Reviews
Published

Oct 26, 2026

Base Model
Training
Steps: 3,000Epochs: 38
Usage Tips
Strength: 1
Hash
AutoV23FC9DBC804
Trigger Words
eyestyle faceeyestyle neckeyestyle chesteyestyle backeyestyle armeyestyle legeyestyle handeyestyle footeyestyle forehead

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

20261026T092023629884600Z_seq-000001_hires_prompt_sections_seed-1117136081930130_job-ed5fb9262e2d464eb9596b0aef704805_node-29_image-0_00001_.png

Add extra eyes to selected areas of a character’s body, from a single forehead eye to elaborate arrangements across the back, arms or legs. Designed for character art, supernatural designs and unusual fantasy illustrations.

Choose a region using its trigger phrase:

  • Forehead

  • Face

  • Neck

  • Chest

  • Back

  • Arms

  • Legs

  • Hands

  • Feet

Hands and feet are still a work in progress. Placement can vary with pose and framing, and further improvements are planned.

Recommended settings

  • Model: Compatible original 28-block Anima checkpoints.

  • Recommended checkpoint: One Obsession V4 Anima works well.

  • LoRA strength: Start at 1.0; try 0.75 for a lighter effect.

  • Sampler: Euler

  • Scheduler: Simple

  • Steps: 30

  • CFG: 4

  • Resolution: 832 × 1248 is a tested portrait starting point.

For Anima Turbo v1.1, the tested settings are 10 steps, Euler/simple, CFG 1. Use the settings appropriate to your checkpoint.

This is a model-only LoRA; no CLIP strength is needed.

Trigger phrases

eyestyle forehead
eyestyle face
eyestyle neck
eyestyle chest
eyestyle back
eyestyle arm
eyestyle leg
eyestyle hand
eyestyle foot

Use the phrase for the region you want. There is no master trigger, and you do not need to include every region.

Prompting tips

Both tags and natural-language descriptions can work. Keep the intended region visible and describe the placement clearly. For example:

eyestyle neck, additional eyes embedded only in the skin of the neck, between the jawline and collarbones, chest covered by a buttoned blouse

Start with one region at a time. Close and medium views generally make the effect easier to see. If eyes appear outside the intended area, try another seed, lower the strength slightly, or explicitly describe which areas should remain unchanged.

Regional placement is not guaranteed: eyes can occasionally spread onto nearby skin, clothing or objects. Stronger weights can also change the pose or rendering style, so higher strength is not always better.