Model Architecture: This is a DiT-based model, recommended for use with ComfyUI or Forge_neo.

⚠️ Usage Limitations

• Generalization: Due to training on a minimal dataset, the model has limited generalization capabilities.

• Style Lock: Strong lighting characteristics; the visual style is prone to "locking" into specific lighting effects.

• Scene Limitations: Stability may decrease in complex compositions or realistic/non-anime subjects.

• Recommended Uses: Portraits, half-body shots, close-ups, and emotional expression works with strong atmospheres.

⚙️ Recommended Settings

• Resolution: 1024 × 1536 | 780 × 1344 | 1280 × 819

• Sampler: Euler SDE

• Scheduler: Simple

• CFG Scale: 4.0

• Sampling Steps: 30

⚠️ Disclaimer: Since the actual use of the model is beyond the author's control, all consequences arising from images generated by the model are the sole responsibility of the image outputter.