Hadn't tested out any of the image models for training loras since Flux 1 dev, so I was curious which one is the current best. Results were pretty interesting

Models tested: Ideogram v4, Flux.1 Dev, Flux.2 Dev, Klein, Krea 2, Z-Image