I will train a stable diffusion lora model, illustrious or pony

The character name and source, preferred outfits, and whether you want an Illustrious or Pony base. Reference images help but aren’t required for well-documented characters.

I can create LoRA models for roughly 99% of anime characters, including older series. My training workflow is optimized to capture the character’s core design, proportions, and visual identity. In rare cases where reference material is extremely limited, I’ll let you know before proceeding.

If a character has very little available reference material, some generations may not fully match expectations. This is a limitation of the source material rather than the training process, and it’s most common with obscure or rarely depicted characters.

All LoRAs are trained on either Illustrious or Pony. They will work on all Illustrious or Pony Checkpoints, results may vary depending on the checkpoint.

The LoRA is designed to be flexible. You can combine it with other LoRAs, use different outfits, poses, camera angles, and backgrounds, and integrate it into your own workflows. Results depend on prompt quality and base model settings, but the model is not locked to a single look or pose.

Outfits that exist in official material of the character. Fan-made designs may require approval or additional discussion.

Intermediate versions saved during training, useful for testing or personal fine-tuning.

No. I only create single-character LoRA models. Combining separate character LoRAs in one image typically produces better results, higher flexibility, and cleaner character separation than multi-character training.

Yes. The model is intended to work alongside other LoRAs. As with all LoRA combinations, weight balancing and prompt structure will affect results.