mygoooo_anima_b01_lacrimos_1_5ep.safetensors

88 MB · SafeTensor

Memerlukan keanggotaan

<lacrimosa-anima>

This is an experimental character lora trained for anima base-v1.0. the main purpose of this model is to test whether short character-name trigger words can reliably activate the intended character features and visual identity. this model should be considered a training experiment rather than a fully polished character release. results may vary depending on prompt structure, lora weight, sampler, and how strongly the trigger words are placed in the prompt. trigger words lacrimosa, neverness to everness it is recommended to place the trigger words near the beginning of the prompt, together with basic quality tags. recommended generation settings these settings follow the general recommendations for anima base-v1.0: - base model: anima base-v1.0 - lora weight: 0.8–1.0 - 0.8–0.9 is usually the most balanced range - 1.0 can strengthen the character identity, but may reduce flexibility - sampler: er_sde / euler_a / res multistep - scheduler: simple or sgm_uniform - steps: 25–30 - cfg scale: 4–5 prompt tips start with basic quality tags such as: masterpiece, best quality, highres then add the character trigger words and character description. anima can be sensitive to prompt order, so placing quality tags and trigger words near the front of the prompt is recommended. negative prompt tips in most cases, adding only the following negative tags is enough: score_1, score_2, score_3 this is usually sufficient to avoid low-quality outputs. using too many negative tags is not recommended, as it may reduce output diversity or over-correct visual details. limitations because this is an experimental model, the trigger behavior and character consistency are still part of the evaluation. the training dataset was not strictly filtered, which may significantly affect the accuracy of the official character outfit. as a result, this model may not reliably reproduce the exact official costume design, and outfit details may vary across generations. this model is mainly intended for testing character trigger behavior, character identity activation, and prompt sensitivity, rather than strict official outfit reproduction. feedback and comparison results are welcome, especially regarding trigger strength, character identity, outfit consistency, and prompt sensitivity. official anima reference for more details about anima prompting and recommended settings, please refer to the official model page:

SHA256

4C98FFF1F495E87813F953EE33E2BD62BE1C4E70EBCDA7F5F4C1902B52BB0CD0

ID Model ModelTrends.ai

#23726

Sampler

er_sde

Langkah

30

Guidance

4.5

Seed

2561024091

Ukuran

832x1216

Bobot yang dipakai

1

Sampler

er_sde

Langkah

30

Guidance

4.5

Seed

560070791

Ukuran

832x1216

Bobot yang dipakai

1

Sampler

er_sde

Langkah

30

Guidance

4.5

Seed

2561024091

Ukuran

832x1216

Bobot yang dipakai

1

Sampler

Euler a

Langkah

30

Guidance

4

Seed

1230377001

Ukuran

2048x1716

Clip skip

2

Sampler

er_sde

Langkah

30

Guidance

4.5

Seed

3109174149

Ukuran

832x1216

Bobot yang dipakai

1

Sampler

er_sde

Langkah

30

Guidance

4.5

Seed

3109174149

Ukuran

832x1216

Bobot yang dipakai

1