as far as I know koyha lora trainer needs to crop or resize dataset images so is locon training the same ?
you don’t need to crop or resize images with kohya trainer, for any training. be sure to have “enable buckets” activated. this groups together images of similar dimensions together.
however, there are some use cases where cropping many be useful, or where making sure your images have dimensions to fall under the same bucket. but those are are not the norm.
now we dont crop image like we used to do. just "enable bucketing" when training.