TDeint and TIVTC

[pinterf]

With the great help of real.finder:

New release:
TIVTC 1.0.7 (x86/x64)
Please, still treat it as beta (due to moving all inline asm parts to simd intrinsics, plus new SSE2 parts), but of course feedback is appreciated.

**v1.0.7 (20170427)**
- fix crash in FieldDiff (in new SIMD SSE2 rewrite)

**v1.0.6 (20170421) - pinterf**
- project migrated to VS 2015
- AVS 2.6 interface, no Avisynth 2.5.x support
- some fixes
- x64 port and readability: move all inline asm to simd intrinsics or C
- supports and requires SSE2
- MMX and ISSE is not supported, but kept in the source code for reference
- source code cleanups

**v1.0.5 (2008) - tritical**
- see old readmes

[pinterf]

Tested four versions:
1.0.5 original
1.0.5 vc10 rebuild
1.0.7 with inline asm (my internal test build)
1.0.7 with simd intrinsics (release)
All 32 bits for comparison

The latter two is bit identical which is good news.

Speedwise: all same, vc10 build is slower by 4-5%.

The two 1.0.5 builds gave slightly different result.

1.0.7 build is showing differences to 1.0.5 on frame by frame comparison, they are almost the same, but it general 1.0.7 gave less false combed frames.

Tested on an interlaced anime clip, that I was given once in PM.

[pinterf]

New TIVTC build.
TIVTC 1.08 (beta)

(simd intrinsics rewrite fix)
Thanks to mizumanju for the report.

**v1.0.8 (20170429)**
- Fix: TFM PP=2 and PP=5 (Blend deint)

[downloaded]

When using with QTGMC with presets Medium to Very Fast it just crashes for me.

QTGMC with presets Slow+ works fine.

Regardless QTGMC preset works fine using http://www.mediafire.com/file/i2qtli..._3-13-2010.rar .

[cork_OS]

Originally Posted by downloaded

When using with QTGMC with presets Medium to Very Fast it just crashes for me.

No problems with Xeon E3-1270 V3.

[Groucho2004]

Originally Posted by downloaded

When using with QTGMC with presets Medium to Very Fast it just crashes for me.

Define crash. No error message?

[downloaded]

Preview on Avspmod trows a generic error "%d format: a number is required, not float." but it previews anyway.

Launching x264 on cli just makes process stop responding without any error msg. I'm available to debug it, but i need some help on what to do.

[Groucho2004]

Originally Posted by downloaded

Preview on Avspmod trows a generic error "%d format: a number is required, not float." but it previews anyway.

I think that's an error Python throws so this is probably some Avspmod issue, not sure.

Originally Posted by downloaded

Launching x264 on cli just makes process stop responding without any error msg. I'm available to debug it, but i need some help on what to do.

You should try running your script with AVSMeter to see if that reveals any problems.

[downloaded]

AVSMeter 2.5.7 (x64) - Copyright (c) 2012-2017, Groucho2004
AviSynth+ 0.1 (r2504, MT, x86_64) (0.1.0.0)

Number of frames: 30508
Length (hh:mm:ss.ms): 00:21:12.438
Frame size: 1274 x 712
Framerate: 23.976 (24000/1001)
Colorspace: i420

Frames processed: 501 (1000 - 1500)
FPS (min | max | average): 0.423 | 155004 | 17.31
Memory usage (phys | virt): 1306 | 1326 MiB
Thread count: 22
CPU usage (average): 32%

Time (elapsed): 00:00:28.943

Works fine. And yet it still crashes x264.

[downloaded]

AVSMeter 2.5.7 (x64) - Copyright (c) 2012-2017, Groucho2004
AviSynth+ 0.1 (r2504, MT, x86_64) (0.1.0.0)

Number of frames: 30508
Length (hh:mm:ss.ms): 00:21:12.438
Frame size: 1274 x 712
Framerate: 23.976 (24000/1001)
Colorspace: i420

Frames processed: 501 (1000 - 1500)
FPS (min | max | average): 0.423 | 155004 | 17.31
Memory usage (phys | virt): 1306 | 1326 MiB
Thread count: 22
CPU usage (average): 32%

Time (elapsed): 00:00:28.943

Works fine. And yet it still crashes x264.

[pinterf]

Yes,
AVSMeter -avsinfo
would definitely help. And that would answer some other questions, too, such as Avs version, 32 or 64 bits, etc.
My 1.08 build is SSE2 only, requires it and does not use higher processor features than that.

[downloaded]

AVSMeter 2.5.7 (x64) - Copyright (c) 2012-2017, Groucho2004

VersionString: AviSynth+ 0.1 (r2504, MT, x86_64)
VersionNumber: 2.60
File / Product version: 0.1.0.0 / 0.1.0.0
Interface Version: 6
Multi-threading support: Yes
Avisynth.dll location: C:\xxxxxxxxxxxxxxxx\AviSynth.dll
Avisynth.dll time stamp: 2017-06-03, 21:06:38 (UTC)


[Plugin errors/warnings]
___________________________________________________________________________________________________________________________________________________________

No plugin directory references found in the registry.
Plugin auto-loading disabled.
___________________________________________________________________________________________________________________________________________________________

AVSMeter 2.5.7 (x64) - Copyright (c) 2012-2017, Groucho2004
AviSynth+ 0.1 (r2504, MT, x86_64) (0.1.0.0)

Number of frames: 30508
Length (hh:mm:ss.ms): 00:21:12.438
Frame size: 1274 x 712
Framerate: 23.976 (24000/1001)
Colorspace: i420

Frames processed: 501 (1000 - 1500)
FPS (min | max | average): 0.423 | 155004 | 17.31
Memory usage (phys | virt): 1306 | 1326 MiB
Thread count: 22
CPU usage (average): 32%

Time (elapsed): 00:00:28.943

Works fine, but it still crashes with x264.

[Groucho2004]

@downloaded
Hard to say why it crashes with x264 but it doesn't seem to be a problem in the filter chain since it works with AVSMeter.

As for the missing plugin directory references - How exactly did you install Avisynth+ and how did you update to the latest version?

You should do a clean install as I suggested here.

[pinterf]

I suppose you are using 64 bit x264. Still, since you have no autoloaded plugins, I cannot tell from this avsinfo the DLL versions you are exactly using.
I tried to encode a qtgmc'd clip with similar frame dimensions and frame rate, but no crash at all. x264 x64 version of r2665

[downloaded]

Originally Posted by Groucho2004

@downloaded
Hard to say why it crashes with x264 but it doesn't seem to be a problem in the filter chain since it works with AVSMeter.

As for the missing plugin directory references - How exactly did you install Avisynth+ and how did you update to the latest version?

You should do a clean install as I suggested here.

I didnt install, i never did. I just use the dll directly on same dir as x264 binary and inside avspmod dir.

Originally Posted by pinterf

I suppose you are using 64 bit x264. Still, since you have no autoloaded plugins, I cannot tell from this avsinfo the DLL versions you are exactly using.
I tried to encode a qtgmc'd clip with similar frame dimensions and frame rate, but no crash at all. x264 x64 version of r2665

Yes, i'm using 64bit x264. I use AddAutoloadDir inside avs script instead of load plugins. I forgot a detail, maybe its important. Height its not mod4 so i addborders first and crop after decimate. Script follows:

SetFilterMTMode("DEFAULT_MT_MODE", 2)
AddAutoloadDir("C:/xxxxxxxxxxxxxxxxxxxx")
ffvideosource("C:\xxxxxxxxxxxxxxxxxxx.mkv")
ConvertToYV12()
AddBorders(0, 2, 0, 0)
QTGMC(Preset="Very Fast",EdiThreads=2).selecteven()
tdecimate(cycleR=1,cycle=5)
Crop(2,2,-4,-2)
Prefetch(4)

[Groucho2004]

Originally Posted by downloaded

I use AddAutoloadDir inside avs script instead of load plugins.

Ah, OK.

Originally Posted by downloaded

SetFilterMTMode("DEFAULT_MT_MODE", 2)
AddAutoloadDir("C:/xxxxxxxxxxxxxxxxxxxx")
ffvideosource("C:\xxxxxxxxxxxxxxxxxxx.mkv")
ConvertToYV12()
AddBorders(0, 2, 0, 0)
QTGMC(Preset="Very Fast",EdiThreads=2).selecteven()
tdecimate(cycleR=1,cycle=5)
Crop(2,2,-4,-2)
Prefetch(4)

Have you tried without the Prefetch statement?

[downloaded]

Yes, same exact problem.

[pinterf]

Ok, now I know your script, I will run tests with those parameters

EDIT: successfully crashed after changing x264 version to r2762.
EDIT 2: Exception thrown at 0x000000014001D63F in x264_64.exe: 0xC0000005: Access violation reading location 0x00000000C101AD62.
EDIT 3: latest Komisar build x264.2833.x86_64.exe is working again, please try that one.

[downloaded]

This is the one i was using and crashing:

x264 0.150.2833 df79067
(libswscale 4.7.101)
(libavformat 57.72.101)
(ffmpegsource 2.23.0.0)
built by Komisar on May 26 2017, gcc: 4.9.2 (multilib.generic.Komisar)
x264 configuration: --bit-depth=8 --chroma-format=all
libx264 configuration: --bit-depth=8 --chroma-format=all
x264 license: GPL version 2 or later
libswscale/libavformat/ffmpegsource license: GPL version 2 or later

Tested with this one and same issue:

x264 0.150.2833kMod df79067
(libswscale 4.7.101)
(libavformat 57.72.101)
(ffmpegsource 2.23.0.0)
built by Komisar on May 26 2017, gcc: 4.9.2 (multilib.generic.Komisar)
x264 configuration: --bit-depth=8 --chroma-format=all
libx264 configuration: --bit-depth=8 --chroma-format=all
x264 license: GPL version 2 or later
libswscale/libavformat/ffmpegsource license: GPL version 2 or later

[hello_hello]

If TIVTC is updated again sometime in the future, the version numbers it writes to the output files it creates (2 pass or VFR mode) could do with an update too.

The TFM metrics file starts like this:

#TFM v1.0.4 by tritical
field = top
crc32 = 385d356e
0 h + [256]

The TDecimate output file looks like this:

#TDecimate v1.0.3 by tritical
crc32 = f76ebc82, blockx = 32, blocky = 32, chroma = T
21885 12119 361525

And the timecodes file created by TDecimate:

# timecode format v1
Assume 29.970030
# TDecimate v1.0.3 by tritical
# Mode 5 - Auto-generated mkv timecodes file