I made a site which uses CE Image for adaptive images. It uses the instructions at https://www.aaron-gustafson.com/notebook/adaptive-images-in-expressionengine-with-ce-image/ to do this. You can see this in action if you search "srcset" in the first link's source code.
But I'm wondering, wouldn't the standard image manipulations be just as good? I'm talking about the "Constrain or Crop" part of your Upload Directory, where when you upload an image, EE automatically creates derivatives of it.
Or maybe I'm oversimplifying things?