Adaptive Resolution for Height Fields

The circle marks the camera. The red area has full resolution, the green one half. The grey one is the fixed base field. It has to be fixed, because I fear that otherwise the mountains would wobble, if the camera moves. So I have to find some way to move to the bigger area. The first idea was, to eventually extend the inner border so that the second area (the green one) is correctly aligned to a multiple of two of the pixels. That means, I would have to add the two small blue stripes in this example. The green area is 9 times as big as the red one. Then I would extend the green one with the blue stripes to get to the grey ones.

But that is not exactly an easy way to do it. And I think, it would not really result in a good algorithm. So I came up with the second one. This time the green area is directly adjacent to the red one and is exactly four times as big (I count the red area to it, but I certainly will not create data for that). The blue area is again to adapt to the static grey mesh. This time the blue mesh is not regular, but it seems overall much easier to implement. I might even opt to make the green mesh 16 times as big. But here it is only such a small part, because the inner area is only 4x4 pixels. Were it 16x16, then the green stripes would be 8 pixels thick, giving 4 stripes of squares around the red area.

One thing that is missing on both is, that I will probably need some area, where the resolutions are blended together. If I only extended the small bit in the second one, I might not get a lot from the additional triangles.