This is a dedicated LoRA for my Original Character (OC). Unlike action-heavy models, this LoRA is specifically trained and optimized for daily life scenarios, featuring stable standing and sitting postures. It captures the character's unique facial features and signature outfits with high consistency.

​💡 Usage Tips & Troubleshooting

​To get the most out of Aurora_OC, please follow these recommendations:

​Style Compatibility: This model was trained on painterly semi-realistic imagery.

​For a cleaner, flatter anime style, try reducing the weight to 0.5 - 0.7.

​To reduce artifacts or graininess from the painterly textures, add (noise:1.2), (grainy:1.2) to your Negative Prompt.

​Character Control (Twin Warning):

​Due to the training data's characteristics, the model may default to generating a twin-sister duo (one with blonde hair, one with purple/gradient hair) if the character count is not specified.

​Solution: Always include 1girl, solo in your prompt to lock in a single character.

​Preventing "Aurora" Backgrounds:

​Because the character's name is "Aurora," the AI may occasionally generate polar lights in the background.

​Solution: If you don't want northern lights, add (aurora:1.4) to your Negative Prompt.

​💡 Technical Limitations & Best Practices

​To get the most out of Aurora_OC, please keep the following technical constraints in mind:

​Training Foundation: This LoRA was trained using Illustrious XL 0.1.

​Pose Constraints: The model is optimized for simple, static poses.

​Complex Actions: If you attempt to generate complex actions or high-dynamic movements (e.g., riding a vehicle, intense sports), you may encounter anatomical artifacts.

​Solution: To fix pose issues, please reduce the weight to 0.5 - 0.7 or remove the LoRA for the initial pose generation.

​Character Consistency: High weights (0.8 - 1.0) will provide the strongest character likeness but will significantly restrict pose flexibility.