Thanks!
I re-read the article after your answers and now it all makes sense
Edit: If anyone's interested, i also found this article to be helpful:
How to generate the elevation maps for GPU based geometry clipmaps?
Ok i have another question.
To update the clipmap, i need a residual texture containing the heights which i obtain from the original compressed heightmap. Right? So does that mean that i have to update a dozen textures (one for each level) by decompressing the heightmap and its mipmaps, and then send the textures to the gpu to be sampled by the clipmap?
If so, how big should those textures be? Same size as clipmap?
To update the clipmap, i need a residual texture containing the heights which i obtain from the original compressed heightmap. Right? So does that mean that i have to update a dozen textures (one for each level) by decompressing the heightmap and its mipmaps, and then send the textures to the gpu to be sampled by the clipmap?
If so, how big should those textures be? Same size as clipmap?
That may be one way to go about it. You may want to test if it is faster to update each level independently or to generate mipmaps after updating only the highest level (I suspect it may be faster to just update everything manually).
They should be the same size as their respective source textures.
L. Spiro
They should be the same size as their respective source textures.
L. Spiro
About the residuals, what is the point of using a filter to predict the finer levels and then using a residual texture to correct the prediction? Why not just read the true data? You're already sampling a texture, might just as well be the heightmap itself.
They were trying to avoid extra texture reads in vertex shaders which were expensive at the time of that document’s writing (2004). These days that is no longer the case and you may decide to just add more fetches.
L. Spiro
L. Spiro
This topic is closed to new replies.
Advertisement
Popular Topics
Advertisement