Basically it´s one model with many LOD´s ,LOD 0 would be the highest resolution,LOD 1 the second highest and so on. Depending on the view distance you pick the most appropriate.Things next to you will be in high res and things far away are low res. So it´s more like the Addative approach you described. Most engines can do…
Take a look at an article from Eric´s website: http://www.ericchadwick.com/examples/provost/byf1.html This will clear up why and when lod´s should be used. This is information you should probably always keep in mind.
Some exporters for different engines can automatically generate LOD´s. The exporter for Ogre3D can do that.You can also let Ogre generate them. Depending on what you do, you should do them by hand.But in most cases you can get of with automatic methods. Neox once said that the Atangeo Balancer does a rather good job at…