How does "match net size to input size" work?

[noobtoob4lyfe]

I'm using DepthAnthing with the "match net size to input size" checkbox enabled and the depth it generates is very good. I would like like know exactly what net height and width values get use with that option. I'm asking because if I manually set the net width and height to match my input image dimension (1080p) it takes much longer to process than if I use "match net size to input size". How is that "net size" calculated when "match net size to input size" is enabled?