Different between doing coregistration from eo-learn and arosics?

I was trying to do image co-registration of two images using arosics and eo-learn. What is the difference in the methodology of both library. I tried co-registering MODIS and Landsat-8 using arosics. But it takes a lot of time on whole raster. Any example of co-registering specific bands of two images without downloading them and only download co-registered band of image.

A post was split to a new topic: Shift between PlanetScope and Sentinel-2