Here are 8 videos demonstrating GPU-accelerated path rendering with the NV_path_rendering extensions for CUDA-capable GPUs:
You can play with these demos yourself.
Ready-to-run Windows versions of the demos shown are available in the NVprDEMOs.zip distribution:
Make sure you get the latest Release 280 GeForce/Quadro driver from:
See the includes ReadMe.txt within the NVprSDK.zip for details.
There’s love for Linux/FreeBSD/Solaris users too, but you’ll have to build the demos from source code found in the NVIDIA Path Rendering Software Development Kit (SDK) found here:
Of course, Windows users can also build the NVprSDK source too.
This is very interesting. However, a quick question: say that I am not using the compatibility profile of OpenGL (ie: 3.0+) and do not have OpenGL state access to transformation matrices (without EXT_direct_state_access), am I able to transform the paths? (Obviously, because there are no vertex/geometry shaders for the paths.)
Am I stuck using EXT_direct_state_access if I wish to transform the paths and use my own matrices + shader uniforms for other 3D stuff? This isn’t too bad a thing, necessarily.
The NV_path_rendering specification states, “For OpenGL 3.x and up, ARB_compatibility is required.” You can’t get ARB_compatibility in a core context. So if you want to use it, you have to use a compatibility profile.
Thanks! Hehe, and it’s the second line in the dependencies paragraph. Don’t know how I missed that before. Probably tired!