Sign In

🪓 Claret Flint (克拉蕾) (Zenless Zone Zero) - Anima

0

Download

1 variant available

bf16 SafeTensor

claret_v1-anima-szyrus.safetensors

BF16, good balance • 63.22 MB

Verified:

Type
LoRA
Stats

89

0

Reviews
Published

Sep 15, 2026

Base Model

Anima

Training
Steps: 2,619
Epochs: 3
Usage Tips
Strength: 1
Hash
AutoV2
62DF92E803
Trigger Words
claret flint \(zenless zone zero\)
default creator card background decoration
Followers - 28

28

Likes - 150

150

Downloads - 1084

1.1K

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

🪓 Claret Flint (Zenless Zone Zero) - Anima 🦇

Trained for Anima with full text-encoder training - her design holds together with complex prompts, various angles and outfits that never appeared in the training data.

One trigger. No outfit tags. No toggles. The trigger alone gives her full in-game look: the skull-sphere hair ornaments on red ribbons, red eyes with symbol pupils, the mole under her eye, the black strapless top with its white lace bib, black detached sleeves with white cuffs, the black miniskirt with the red bat belt, white bat-print thighhighs and platform knee boots.

What this version gets right that fanart usually doesn't:

  • The back. The long black-and-red cape-tail behind the skirt and the cross-laced back of the top are correct from behind and from the side - trained on in-game captures, not guessed.

  • The face. Mole under the eye, symbol pupils, the skull ornaments, red nails - all automatic.

  • Her wardrobe is still open. Name any outfit and it takes over at strength 1.0 while her identity stays put.


⚙️ Usage

Trigger: claret flint \(zenless zone zero\)

Strength: 1.0

Base: Anima Base v1.0 / waiANIMA

Steps: 25–30

CFG: 4–5

Sampler: er_sde / simple

No extra tags required. You don't need to describe her to get her.


💡 Prompt tips

Signature shots that work:

  • holding axe, weapon over shoulder, smirk, fang, red theme, black background, rim light, floating hair - the poster look

  • night, city street, neon lights, rain, wet, reflections, red and blue lighting - Sixth Street after dark

  • from behind, looking back, full body, wind, rooftop, sunset - the cape-tail

  • portrait, close-up, smile, fang, glowing eyes, hand on own cheek, red rim light - the grin

Outdoor scenes: some training data is in-game captures, so add 3d, game cg to the negative if a background comes out looking rendered.

Flat backgrounds: on red background a flat two-tone print style can surface; add monochrome, flat color, spot color, limited palette to the negative if you see it.

Other outfits: just name them - t-shirt, shorts works at strength 1.0. If a piece of the default outfit lingers, negate it (detached sleeves, black skirt, thighhighs).

Suggested negative:

3d, game cg, worst quality, low quality, blurry, watermark, signature, text

🎬 Showcase clips

The animated previews are stills from this LoRA animated with Wan 2.2 I2V - the LoRA is not used in the video model, the identity comes entirely from the Anima frame.


📊 Training

  • kohya sd-scripts, networks.lora_anima, text encoder trained (lr 2e-5)

  • 1024px, dim 16 / alpha 4, 2619 steps, 3 epochs

  • 215 curated Danbooru illustrations + 15 in-game captures

  • four versions, one variable each, picked by measured seed grids