Original Post
So I get the idea in most cases, you keep a low resolution texture for an object that is say 500 feet away, then maybe at 100 feet you load in a better version, but I don't get what the streaming is. I fully understand something like clip mapping, but I'm trying to keep chunks of my terrain far away from using the high detail texture and just use the lowest texture size it needs.
1024 texture
1.) 500 feet away, just load Mip-Map from 1x1 up to 32x32
2.) 100 feet away, Load the full 1024 texture
So on step 2 do I delete the old texture and upload the 1024 texture (even the 1x1 to 32x32), or am I simply forcing an LOD bias at 500 feet so that only the 32x32 and lower are ever actually bound?
I was looking to improve some terrain and saw this (search for "A quadrant mask texture is then created in paint software") and you will see what specific streaming I am looking to do. Basically terrain that has a bunch of chunks with huge textures. So I'm wondering what I stream and where I stream it to.
Terrain Texturing
Maybe I create a 1024 image in my example but only send glTexImage2D to the levels 1x1 to 32x32 and then fill in the other data later? (This just doesn't make sense because I would be wasted memory I'm not using).
Or do I reserve slots where low res images can come and go that I never delete, I just overwrite, and then later when going to the high res I mark that slot as allowed to overwrite with a new low res texture?
1024 texture
1.) 500 feet away, just load Mip-Map from 1x1 up to 32x32
2.) 100 feet away, Load the full 1024 texture
So on step 2 do I delete the old texture and upload the 1024 texture (even the 1x1 to 32x32), or am I simply forcing an LOD bias at 500 feet so that only the 32x32 and lower are ever actually bound?
I was looking to improve some terrain and saw this (search for "A quadrant mask texture is then created in paint software") and you will see what specific streaming I am looking to do. Basically terrain that has a bunch of chunks with huge textures. So I'm wondering what I stream and where I stream it to.
Terrain Texturing
Maybe I create a 1024 image in my example but only send glTexImage2D to the levels 1x1 to 32x32 and then fill in the other data later? (This just doesn't make sense because I would be wasted memory I'm not using).
Or do I reserve slots where low res images can come and go that I never delete, I just overwrite, and then later when going to the high res I mark that slot as allowed to overwrite with a new low res texture?