aWarpSharp (float "depth", int "blurlevel", float "thresh", int "cm")
depth & blurlevel :
The settings you may be familiar with in VDub's WarpSharp filter. They are the only settings you need to tweak to achieve any effect. Blurlevel consumes cpu power, but gives a big boost to the warpsharpening.
default : 16.0 & 2
thresh :
A float value. 1.00 mean 100% (max). It's the bump mapping saturation setting. The default value is recommended for maximum quality. if you tweak this setting, keep in mind it would enhance inegality of warping between edges.
default : 0.5 (50%)
cm : (chroma mode)
cm = 0 will disable chroma filtering.
cm = 1 enables chroma warping with luma bump map (recommended).
cm = 2 enables chroma independant warping & bump map.
default : 2
Advanced settings are undocumented. they aren't needed anyway. (i don't use them, and i think it's better if nobody uses them ^^)
aWarpSharp implements high-quality original warpsharping. "Original" means the algo used differs on several points to other warpsharping filters. The code is fully iSSE optimised with high accuracy. see yourself ^^. You should need about 550 Mhz cpu load to warpsharp 640x480 images at 25 fps. This is a worst case, I achieved about 40 fps PAL full resolution MPEG-2 decoding deinterlacing & warpsharping using MPEG2Dec("dvd.d2v").aDeInt().aWarpSharp() with a 1.4 Ghz cpu.
aWarpSharp is very optmised for my cpu (Athlon XP), but all modern cpus would run this filter at the maximal speed allowed by your hardware ^_^.
I hope you'll enjoy my last filter. it's one of the better piece of code I ever wrote and it was very fun to code.
For all anime fans and all avisynth users who like stuff i coded in the last months.
$Date: 2004/08/13 21:57:25 $