Can this training paradigm work with SDXL?

[MartinXM]

Can in-context lora work with SDXL?

[huanglianghua]

Can in-context lora works with SDXL?

I haven't tried that. Sounds worth trying.

The effectiveness of in-context LoRA largely depends on the in-context generation capabilities of the base text-to-image model. FLUX performs exceptionally well in this regard, while SDXL might not be as strong.

If you decide to experiment with this, I’d be interested in hearing about your findings!