Sarah Mae

What this model does
This is an SDXL character LoRA trained to reproduce a consistent female identity across images and short-form video. It focuses on facial structure, proportions, and overall likeness while remaining flexible across lighting, outfits, poses, and environments.

What it’s good at

• Maintaining consistent facial identity

• Realistic skin tones and lighting

• Portraits, lifestyle shots, and social-media-style content

• Image-to-video and motion workflows when paired with compatible video models

• Works well at 1024px+ resolutions

Recommended use
Use this LoRA with SDXL-based checkpoints. Best results are achieved with moderate LoRA strength and well-described prompts. Suitable for realistic photography-style generations, character-driven content, and short-form video experiments.

What it’s not designed for

• Heavy stylization or anime workflows

• Extreme camera distortion or exaggerated expressions

• Low-resolution generations

Usage notes
For best results, use with a realistic SDXL base model. Start with a LoRA strength between 0.5–0.8 and adjust based on the scene and motion intensity.