Training low resolution then switching to high resolution later?
Basically I’m training a LTX Lora and since it is so heavy and resource intensive, I started training at a lower resolution just to make sure it correctly captured the concept. Can I finish the rest of the training run by adding in higher resolution training data (same as before just higher resolution). Will that screw things up? Make any meaningful difference? Or would it just be better to restart with the higher resolution.