AWarpSharp2 for VapourSynth

jackoneill · 6th October 2015, 20:47

As requested, a port of AWarpSharp2: https://github.com/dubhater/vapoursynth-awarpsharp2

The output is not 100% the same as the Avisynth plugin's, due to some differences in the way the margins are processed. If you crop away six pixels from the left and the right, then maybe the output will be identical. Then again maybe not, because I didn't bother writing any SSSE3 code.

The "chroma" parameter is a bit different now. There are also two new parameters, "planes" and "opt". It's all explained in readme.rst.

I merged the "aWarp" and "aWarp4" filters. There was really no reason to have a separate filter for the supersampled thing.

The "aWarpSharp" filter only existed for compatibility with old Avisynth scripts. That's not relevant here, so there is no such filter in this plugin.

BakaProxy · 7th October 2015, 16:08

Dude, I love u.

feve · 8th October 2015, 23:29

I love u.too

mastrboy · 9th October 2015, 17:06

Quote:

Originally Posted by jackoneill

I didn't bother writing any SSSE3 code.

So we can assume this would be slower than the avisynth version?

Oh, and thanks by the way

jackoneill · 9th October 2015, 17:40

Quote:

Originally Posted by mastrboy

So we can assume this would be slower than the avisynth version?

Oh, and thanks by the way

I don't know. Compare them? (Don't forget about the threads.)

Q3CPMA · 9th October 2015, 22:55

Thanks a lot. One of my favourite plugins with BM3D.

ryrynz · 7th June 2016, 13:44

Not sure if entirely relevant to this thread, but since the a 64 bit version has been done, what are the chances of 64 bit version for Avisynth+?

feisty2 · 7th June 2016, 13:47

Quote:

Originally Posted by ryrynz

Not sure if entirely relevant to this thread, but since the a 64 bit version has been done, what are the chances of 64 bit version for Avisynth+?

or maybe its just a perfect time to switch to vaporsynth

ryrynz · 7th June 2016, 14:15

It still can't do real time post processing yet can it? I believe ffdshow raw and Avisynth is still the only option until madVR adds support.

jackoneill · 7th June 2016, 16:44

Quote:

Originally Posted by ryrynz

Not sure if entirely relevant to this thread, but since the a 64 bit version has been done, what are the chances of 64 bit version for Avisynth+?

From me, very low. Maybe ask in the Avisynth+ thread.

~SimpleX~ · 10th June 2016, 08:50

Would it ever support >8bit clips?

jackoneill · 10th June 2016, 10:09

Quote:

Originally Posted by ~SimpleX~

Would it ever support >8bit clips?

Who knows.

jackoneill · 10th June 2016, 19:09

Quote:

Originally Posted by jackoneill

Who knows.

Make that a "yes".

https://github.com/dubhater/vapoursy...eleases/tag/v2

The only notable change is the 16 bit support.

divxmaster · 23rd July 2016, 04:51

I am having real trouble getting awarpsharp to load, all 64bit vs, definitely 64 bit libawarpsharp2.dll in plugins64. says no attribute with name warp exists. I have no trouble with any other plugins, qtgmc, fmtc, bm3d etc. any ideas? cheers Divxmaster

jackoneill · 23rd July 2016, 10:11

Quote:

Originally Posted by divxmaster

I am having real trouble getting awarpsharp to load, all 64bit vs, definitely 64 bit libawarpsharp2.dll in plugins64. says no attribute with name warp exists. I have no trouble with any other plugins, qtgmc, fmtc, bm3d etc. any ideas? cheers Divxmaster

Load it manually so you'll get an error message:

Code:

core.std.LoadPlugin("B:/your/path/here/libawarpsharp2.dll")

divxmaster · 23rd July 2016, 22:42

Ah, requires API 3.4, I'm currently on 3.2.
Thanks for the tip on LoadPlugin showing errors, vs autoload that supresses errors, that's useful to know.
edit: For those that want to know, Vapoursynth R30 or newer needed for API 3.4.

Cheers,
Divxmaster

trip_let · 28th August 2016, 17:41

I think I found a bug. With AWarpSharp2 (did not test others), > 8 bit input, chroma=0, and planes=[1,2] (or 1 or 2, i.e. luma not processed for output), the output is garbled, with half the screen being totally wrong.

btw for bug reports in general do you prefer a notice here or on Github? Thanks, anyway. Really appreciate the work.

jackoneill · 28th August 2016, 20:46

v3 fixes the bug.

Quote:

Originally Posted by trip_let

I think I found a bug. With AWarpSharp2 (did not test others), > 8 bit input, chroma=0, and planes=[1,2] (or 1 or 2, i.e. luma not processed for output), the output is garbled, with half the screen being totally wrong.

btw for bug reports in general do you prefer a notice here or on Github? Thanks, anyway. Really appreciate the work.

I prefer Github, if you have an account there. If not, this is fine too.

jackoneill · 15th June 2018, 10:41

v4 is here.

Code:

   * Add parameter "cplace". Valid values: "mpeg1" and "mpeg2". Default: "mpeg1", same as in previous versions.
   * Make the "depth" parameter an array so the depth can be set independently for each channel.

6th October 2015, 20:47	#1 \| Link
jackoneill unsigned int Join Date: Oct 2012 Location: 🇪🇺 Posts: 760	AWarpSharp2 for VapourSynth As requested, a port of AWarpSharp2: https://github.com/dubhater/vapoursynth-awarpsharp2 The output is not 100% the same as the Avisynth plugin's, due to some differences in the way the margins are processed. If you crop away six pixels from the left and the right, then maybe the output will be identical. Then again maybe not, because I didn't bother writing any SSSE3 code. The "chroma" parameter is a bit different now. There are also two new parameters, "planes" and "opt". It's all explained in readme.rst. I merged the "aWarp" and "aWarp4" filters. There was really no reason to have a separate filter for the supersampled thing. The "aWarpSharp" filter only existed for compatibility with old Avisynth scripts. That's not relevant here, so there is no such filter in this plugin. __________________ Buy me a "coffee" and/or hire me to write code!

23rd July 2016, 22:42	#16 \| Link
divxmaster Registered User Join Date: Mar 2015 Location: New Zealand Posts: 45	Ah, requires API 3.4, I'm currently on 3.2. Thanks for the tip on LoadPlugin showing errors, vs autoload that supresses errors, that's useful to know. edit: For those that want to know, Vapoursynth R30 or newer needed for API 3.4. Cheers, Divxmaster Last edited by divxmaster; 23rd July 2016 at 23:08.

Thread Tools	Search this Thread
Show Printable Version Email this Page	Search this Thread: Advanced Search
Display Modes
Linear Mode Switch to Hybrid Mode Switch to Threaded Mode

7th October 2015, 16:08	#2 \| Link
BakaProxy Registered User Join Date: Jan 2015 Posts: 47	Dude, I love u.

8th October 2015, 23:29	#3 \| Link
feve Registered User Join Date: Jul 2014 Posts: 2	I love u.too

9th October 2015, 22:55	#6 \| Link
Q3CPMA Registered User Join Date: Jun 2015 Posts: 33	Thanks a lot. One of my favourite plugins with BM3D.

7th June 2016, 13:44	#7 \| Link
ryrynz Registered User Join Date: Mar 2009 Posts: 3,684	Not sure if entirely relevant to this thread, but since the a 64 bit version has been done, what are the chances of 64 bit version for Avisynth+?

7th June 2016, 14:15	#9 \| Link
ryrynz Registered User Join Date: Mar 2009 Posts: 3,684	It still can't do real time post processing yet can it? I believe ffdshow raw and Avisynth is still the only option until madVR adds support.

10th June 2016, 08:50	#11 \| Link
~SimpleX~ Registered User Join Date: Feb 2010 Location: Saint-Petersburg, Russia Posts: 38	Would it ever support >8bit clips?

23rd July 2016, 04:51	#14 \| Link
divxmaster Registered User Join Date: Mar 2015 Location: New Zealand Posts: 45	I am having real trouble getting awarpsharp to load, all 64bit vs, definitely 64 bit libawarpsharp2.dll in plugins64. says no attribute with name warp exists. I have no trouble with any other plugins, qtgmc, fmtc, bm3d etc. any ideas? cheers Divxmaster

28th August 2016, 17:41	#17 \| Link
trip_let Registered User Join Date: Sep 2012 Posts: 47	I think I found a bug. With AWarpSharp2 (did not test others), > 8 bit input, chroma=0, and planes=[1,2] (or 1 or 2, i.e. luma not processed for output), the output is garbled, with half the screen being totally wrong. btw for bug reports in general do you prefer a notice here or on Github? Thanks, anyway. Really appreciate the work.