I often find that I'm converting from layers to images or objects because I can't find a node that will do the thing I want it to otherwise.
-It doesn't seem possible to change the size of an image. There's no transform image node. Resize image will pixellate the image but won't have the same effect as transform layer.
-It doesn't seem possible to process the layer with regard to visual effects. It therefore has to be changed back to an image. For example, imagine you want to change the hue of some copied layers.
-It's mostly easier to work with objects when making shapes and things, even if you want to make 2d shapes, so then I end up converting object to layer.
As I go on to write more complex compositions, I can easily imagine changing back and forth between these three modes many times within the composition. So my question is, does it really matter if I switch modes a lot? Or is it better to keep it to the minimum because of CPU? Should I try and find workarounds to keep things in the same mode or should I just change back and forth casually?