[OpenGL ES] Using glDrawElements with GL_UNSIGNED_INT

[regnar]

Hello!
I was happily using glDrawElements with GL_UNSIGNED_SHORT index data for my index buffer rendering until I needed to draw terrain mesh larger than 256*256 vertices. But it turned out that OpenGL ES doesn't support GL_UNSIGNED_INT. How do I solve this problem now?
Say, I want to draw a mesh with 1024*1024 vertices. If I sub-divide this into smaller index buffers (like 256*256), it means I will have to create 16(!) index buffers just for that one mesh. Sounds really bad to me.

[Ed Welch]

Originally Posted by regnar

Hello!
I was happily using glDrawElements with GL_UNSIGNED_SHORT index data for my index buffer rendering until I needed to draw terrain mesh larger than 256*256 vertices. But it turned out that OpenGL ES doesn't support GL_UNSIGNED_INT. How do I solve this problem now?
Say, I want to draw a mesh with 1024*1024 vertices. If I sub-divide this into smaller index buffers (like 256*256), it means I will have to create 16(!) index buffers just for that one mesh. Sounds really bad to me.

You won't get an iDevice to render one million vertices. The limit for MBX devices (i.e. iPhone 3G) is around 10,000 polygons for a reasonable frame rate.

[regnar]

Originally Posted by Ed Welch

You won't get an iDevice to render one million vertices. The limit for MBX devices (i.e. iPhone 3G) is around 10,000 polygons for a reasonable frame rate.

Well, I am not going to render million vertices, I just need a way to index large vertexbuffer (larger than 65 536 vertices, becouse that's what ushort allows). Having a large index buffer doesn't mean I am going to be rendering it all at once. Of course I use all kind of space partitioning and LOD techniques.

[warmi]

Originally Posted by regnar

Well, I am not going to render million vertices, I just need a way to index large vertexbuffer (larger than 65 536 vertices, becouse that's what ushort allows). Having a large index buffer doesn't mean I am going to be rendering it all at once. Of course I use all kind of space partitioning and LOD techniques.

Having multiple index buffers is not a big problem ... Like you said, you are not going to send everything with one batch so I don't think it makes a lot of difference.