I am capturing an OpenGl texture and displaying it to pwindow using the following pipeline:
jit.gl.node -> jit.gl.camera @capture 1 @antialias 1 -> jit.gl.videoplane @transform_reset 2
The antialiasing looks pretty good and the performance seems good.
I have also tried the following approach:
jit.gl.node -> jit.gl.camera @capture 1 @antialias 0 -> jit.slab @file fxaa.jxs -> jit.gl.videoplane @transform_reset 2
as outlined in this infamous tutorial here:
I see no performance difference between these two methods, however at the moment, I am only rendering a jit.gl.mesh of only like 1000 points to a pwindow that is relatively small.
My question is, given the pipeline of each of these methods, is one more efficient than the other? Is hardware acceleration used by either of these methods (my assumption is that since jit.gl.slab is using a shader then it would be HW accelerated if supported by the graphics card) - but I am not sure what the "antialias 1" flag does on the jit.gl.camera object?
To my eyes, the antialiasing looks a bit cleaner using the first method (@antialis 1 flag on the capture object) - while the FXAA method seems a bit too gaussian - but that is just my opinion.