Sign in to follow this  
Opwiz

LOD Idea

Recommended Posts

I do not have much knowledge about different ways to implement LOD and this is just one idea for a LOD system I got from watching renderings of mountains. The one place where the level of detail matters most is at the edges where the mountain meets the sky. So if we calculate the angle between the normal of the polygon and the view vector and the greater the angle the more details should be added. Is there any terrain engines that does this?

Share this post


Link to post
Share on other sites
Guest Anonymous Poster
IMHO the angle should be calculated between the ground plane and the triangle in question. This way, the "steeper" the triangle, the more chances it will add detail to the scene. In your case, the angle of flat ground polygon normal and view direction(assuming looking straight and level) will be 90 degrees, so more detail will be added to flat ground surface. However, both cases don't deal with lod over distance. IMHO, such a method could be applied when calculating LOD in some traditional distance based approach. Assuming you calculate some kind of screen-space or geometric error in your LOD approach, this error could be adjusted depending on the angle of the polygon with the ground.

Share this post


Link to post
Share on other sites
I had the idea of using 'fins' (much like those used for realtime fur) on terrains to give an apparenly higher level of detail that was actually being used. You'd use the same diffuse texture but with an alpha tested edge to change a sharp profile into a noisey one.

Could be good for further away LOD levels where the seam wouldn't be visible, but I never got around to trying it.

Share this post


Link to post
Share on other sites
This is what the view-dependent LOD algorithms do. The problem with these is that they use a lot of CPU time for a fairly small reduction of triangle count, as well as having to shuffle data across the bus to the card each frame. This might help if triangles were very expensive, but on today's hardware, they aren't, so most of the time, a view-independent method based on static geometry (say, geo-mipmapping) is usually the better win.

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now

Sign in to follow this