Maya Reyes | Original Character LoRA (SDXL)

Maya Reyes | Original Character LoRA (SDXL)

Version: v1 Type: Character LoRA Base model: SDXL [add the exact checkpoint you trained on, e.g. Mio 2 if applicable] Trigger word: mayareyes_char Recommended weight: 0.7-1.0 (start at 0.8)

About

Maya Reyes is an original character: a 25-year-old barista with a dry, knowing smirk, who moves through a quiet city life of café shifts, late-night walks home, and a very round orange cat named Biscuit. This LoRA is trained to keep her face and design consistent across outfits, poses, and settings, and it works well with otome-style and slice-of-life art.

Character traits (baked in by the LoRA) Adult woman, mid 20s Short black ponytail Dark brown eyes Warm skin tone Slim build Subtle, knowing smirk as a default expression

You don’t need to prompt these. The trigger word carries them. Prompt only what you want to change (outfit, pose, background).

How to use

Basic prompt:

mayareyes_char, 1girl, solo, [outfit], [pose], [expression], [background]

Example prompt:

masterpiece, best quality, mayareyes_char, 1girl, solo, otome game style, upper body, denim jacket, soft smile, looking at viewer, city street at dusk, soft glow, bokeh

Suggested negative prompt:

lowres, bad anatomy, bad hands, extra fingers, missing fingers, extra limbs, deformed, blurry, watermark, text, signature, jpeg artifacts, multiple girls, long hair

Settings I used for previews: [sampler], [steps], CFG [value], 832x1216 / 1024x1024, hires fix on.

Tips Works best at weight 0.7-1.0. Go lower (0.5-0.7) if you’re stacking it with a strong style LoRA. The LoRA does not include a signature outfit. Her barista apron and denim jacket appear in the training data, but they aren’t locked in, so prompt any clothing you want. For otome or shoujo looks, add style tags such as otome game style, shoujo, soft shading, sparkles. Use from side, from behind, or looking back for variety. Close-up front views are the most consistent. Hands and props can still need inpainting, as with most SDXL models. Training details Dataset: [number] images, mix of close-ups, upper body, and full body, in multiple outfits and settings Trainer: [Otome / platform name] Rank / alpha: [value] Steps / epochs: [value] Learning rate: [value] Captions: trigger word plus variable tags (outfit, pose, background); constant traits omitted Limitations May drift on rarely seen angles (extreme low or high angles). Strong style LoRAs or heavy weights can alter her face. Not designed for multi-character scenes. For two characters, use regional prompting or inpainting. Content note

Maya is an original adult character (age 25). Please keep generations consistent with that, and don’t use this LoRA to depict her as a minor.

Credits and usage

Original character created by [your name/handle]. Free to use for personal creative projects. Please credit me if you share your work: [your terms: credit required / non-commercial only / etc.].