Archived

This topic is now archived and is closed to further replies.

Laurent

Progressive Mesh and Single Strips on Radeon boards

Recommended Posts

Hi all, A few months ago, starting from the SDK samples, I have coded a mesh loading class that uses progressive meshes and converts all subsets of all progressive meshes to single strips. This for speed and also to allow easy usage of vertex shaders on Meshes. This was working very nicely and really fast with DirectX 8.1 on GF256, GF2-4 MX, GF3, GF4, Radeon VE, 7000, and 8500 boards. I have converted the above class to DX9 quite easily, and now it is still working perfect on GF2-4 MX, GF3, GF4, debug or release build, debug or retail runtime. However, on ATI boards and strangely also on the GF256, I have ugly artefacts. Vertex shaders or not, it is the same mess: Meshes look like they were "connected" to each other when they are drawn one after the other. When running in the debugger, there are no errors, no memory leaks. Yes, the length of the strip is checked against the device caps. Before blaming it on the ATI drivers ( I am using Catalyst 3.1 ) I would like to know if they are known restrictions or issues on using single strips with progressive meshes. What kind of thing I could have done wrong ? I will appreciate any comment or suggestions, because I have been pulling my hair on this problem since quite a while... Thanks in advance

Share this post


Link to post
Share on other sites