The training and publication of this LoRA were conducted entirely by an autonomous Agent equipped with multiple embedded skills and tools. The complete workflow, including data generation, training configuration, model training, result organization, and automated release, was executed without human supervision. Based on the author’s manual inspection of representative samples, the resulting LoRA demonstrates strong character consistency, stable visual identity, and well-defined character anchors. These observations suggest that the proposed agent-driven workflow is both efficient and technically robust.

Further implementation details are available in the accompanying technical article:
https://civarchive.com/articles/32228/single-image-character-lora-training-via-synthetic-expansion

Kafka / 卡夫卡 multi-outfit LoRA for Arknights on WAI v17 / Illustrious-SDXL. Use `kafka_arknights` with `arknights` around weight `0.55`; recommended range is `0.45-0.70`.

Trigger words
kafka_arknights, arknights, 1girl, elite_1, cardboard_fairy

Recommended Usage

### Default Weight

`0.55`

### Useful Range

`0.45-0.70`

### Motion / Composition First

Use the lower side of the tested range, around `0.45`, when pose flexibility or scene composition matters more than strict outfit locking.

### Avoid as Default

Weights above `0.70` are not recommended as the default because this project showed higher risk of reference-prop leakage and reduced composition flexibility.

Outfit Prompts

精英1 / elite_1

Outfit trigger: elite_1

Full prompt

kafka_arknights, elite_1, 1girl, solo, long_hair, brown_hair, amber_eyes, hair_over_one_eye, blunt_bangs, white_hair_streak, black_hooded_coat, teal_trim, yellow_trim, open_coat, black_crop_top, black_shorts, high_heel_boots, teal_boots, teal_hood, metal_collar, metal_gauntlets, thigh_garter, mechanical_knee_pad

Example with LoRA

<lora:kafka_arknights_strict_imagegen_recommended_e000007:0.55>, kafka_arknights, elite_1, 1girl, solo, long_hair, brown_hair, amber_eyes, hair_over_one_eye, blunt_bangs, white_hair_streak, black_hooded_coat, teal_trim, yellow_trim, open_coat, black_crop_top, black_shorts, high_heel_boots, teal_boots, teal_hood, metal_collar, metal_gauntlets, thigh_garter, mechanical_knee_pad, upper body, looking at viewer, soft smile

纸板仙子 / cardboard_fairy

Outfit trigger: cardboard_fairy

Full prompt

kafka_arknights, cardboard_fairy, 1girl, solo, long_hair, dark_brown_hair, amber_eyes, bangs, side_locks, teal_jacket, long_sleeves, gold_trim, floral_print, red_pleated_skirt, white_high_collar, red_neck_ribbon, white_mid-calf_boots, black_sole, white_beret, red_headband, red_ribbon, gold_hair_ornament, hair_bow, gold_waist_bow, white_pom_pom, gold_boot_bows

Example with LoRA

<lora:kafka_arknights_strict_imagegen_recommended_e000007:0.55>, kafka_arknights, cardboard_fairy, 1girl, solo, long_hair, dark_brown_hair, amber_eyes, bangs, side_locks, teal_jacket, long_sleeves, gold_trim, floral_print, red_pleated_skirt, white_high_collar, red_neck_ribbon, white_mid-calf_boots, black_sole, white_beret, red_headband, red_ribbon, gold_hair_ornament, hair_bow, gold_waist_bow, white_pom_pom, gold_boot_bows, upper body, looking at viewer, soft smile

Negative Prompt

low quality, bad anatomy, blurry, bad proportions, deformed hands, extra limbs, extra fingers, disfigured, low resolution, oversaturated, grainy, poorly drawn face, mutated, unrealistic, text, watermark, low detail, bad lighting, noisy background, bad composition, messy, blurry edges, artifacts, wrong perspective, long neck, bad posture, asymmetrical, pixelated, distorted, out of focus, clipping,
different outfit, alternate outfit, outfit change, costume change, wrong outfit, incorrect outfit, mismatched outfit, inconsistent outfit, redesigned outfit, modified outfit, extra clothes, missing clothes, incomplete outfit,

Recommended Generation Settings

Checkpoint: WAI-Nsfw-Illustrious-17
Base resolution: 1024 x 1024
Final resolution: 2048 x 2048
Sampler: DPM++ 2M
Scheduler: Karras
Steps: 30
CFG scale: 6.5
Hires.fix: enabled
Hires upscaler: RealESRGAN_x4plus_anime_6B
Hires steps: 32
Denoising strength: 0.4
Upscale by: 2
Hires CFG scale: 6
LoRA weight: 0.55

License / Usage
NSFW generation is allowed. Author credit is required: jifengwang.
Generated images may not be sold. Selling this model, merges, or redistributed model files is not allowed.
Use on Civitai and other generation services is allowed. Sharing merges and using different permissions on merges are allowed.