MSharpen for Avisynth 2.5

Here is MSharpen compiled for Avisynth 2.5. It currently supports only RGB and YUY2.

*cough* (Yes, I've caught a cold) ;)

I'm such an idiot. Cough.

nice, nice neuron2, do you have any plans to resume development of msmooth?

You could look at ffdshow's msharpen implementation. I attached the source of the sharpen filters. There is a C and assembler version a little down. Look at my warpsharp port, if you'd like to see how the different variables are mapped. I converted Warpsharp/Xsharpen, but didn't want to release msharpen, since you said you were working on it yourself. Edit: I should not that the attachment only contains a source file.

@sh0dan If you have coded MSharpen for YV12, by all means please send me the code. :)

This concerns the Avisynth 2.0 version so it might be a little off topic, but your msharpen seems to work miracles at reducing edge noise and ringing in Akira :) keep up the good work!

Sorry to bump this old thread, but I have another question concerning where in my script I should use the MSharpen filter. Specifically, my question is whether I should use it before or after convolution3d. I just want to be sure that I don't inadvertantly screw up the quality (even if it isn't noticeable, I just want to be on the safe side). Here's my script: Code:

I don't like MSharpen after FieldDeinterlace because it amplifies residual combing. Better to do it this way: SeparateFields() MSharpen() Weave() FieldDeinterlace() Convolution3d() BicubicResize() I think it's OK before Convolution3d() as long as you use a reasonable threshold for MSharpen(). The areas you want Convolution3d() to clean up are the non-edge areas of MSharpen().

Thanks, I'll try that. Although, after using forced-FILM there aren't very many combed frames, and the ones that are are often in such dark and noisy scenes that after resizing, the frame already looks like it was compressed in mpeg4 :)