An assortment of small performance changes. I get maybe 20-25% more FPS on a complex hair, but mileage will vary.
Some fixes with mirrored surfaces and additional cameras - one step closer to working VR
Reworked Phong Tangent - shading should be a bit smoother now
Added voxel-based volumentric self-shadowing. It eats into perf, but usually worth it.
A moderate performance boost from better parallelization of some geometry calculations
Fixes for multiple hairs (Still need a separate instance of the plugin per - sorry)
I've revamped how spread and alignment works. The various spread sliders were removed. There are now four primary controls:
Root and tip radius - the spread at the root/tip
Root and tip skin alignment - how much the root/tip follows the skin
Before, sub-hairs were always rendered radially...
I tore out the Impostor render mode. It's Compute-only now as that seems to be the better long-term bet
Tweaks to the k-buffer that should result in significantly better performance at K=4 (default)
Immediately ate into the perf improvement with some occlusion fixes, but should still be a net...
No, it's not you. This uses some uncommon rendering techniques so making VR work will take some effort. I do have a headset to test with and it's on the list, though.
The Compute mode bypasses Unity's regular rendering pipeline, so there's bound to be some issues.
What GPU are you running? Can you also try version 2 to see if that helps? I reshuffled some things in v3 to free up buffers for better AA.
I managed to increase the size of the k-buffer - allowing for better anti-aliasing. I also added an Analytic Coverage toggle, which makes the AA cheaper.
The end result is better-looking hair for more or less the same performance
Before
After