QTGMC Deinterlacing Script (v3.32)

[Didée]

Is it definite that Lagarith is decoding to YUY2? Perhaps it decodes to RGB ...

---------------------------------
SetMemoryMax(600)
SetMTMode(5, 6)
AviSource("c:\temp\jj.avi" ,audio=false ) # lagarith lossless yuy2 video
return( info() )
#Converttoyv12()
AssumeTFF()
SetMTMode(2)
QTGMC( Preset="Very Slow", EdiThreads=3)
------------------------------------------

[yup]

@ttatcs
Please list for avisynth/pugin directory. May be problem with masktools 26?
yup.

[Gavino]

Originally Posted by yup

Now for me YUY2 colorspace work. SEt 2.6 build, mt_masktool 25 not 26. Kasandro plugins only with SSE2 support.

Are you saying that mt_masktools-25.dll works with SEt's 2.6 build and mt_masktools-26.dll does not? That seems contradictory to previous understanding.

[yup]

@Gavino.
Before using YUY2 colorspace inside QTGMC no problem I can use both version 26 and 25, but after transfer to YUY2 properly work only 25 version.
yup.

[yup]

@ttatcs
I am using following version

31.12.2010  18:13           966144 mt_masktools-25.dll
24.10.2007  10:41            33280 RemoveGrainSSE2.dll
24.10.2007  10:41            24576 RepairSSE2.dll
11.04.2005  12:16            30720 SSE2Tools.dll
28.02.2008  08:03            11776 VerticalCleanerSSE2.dll

@Gavino when I only replacing to 26 version masktools I get error message
---------------------------
VirtualDub Error
---------------------------
Avisynth open failure:
mt_makediff : unsupported colorspace. masktools only support planar YUV colorspaces (YV12, YV16, YV24)
(QTGMC-3.31.avsi, line 775)
(QTGMC-3.31.avsi, line 387)
(E:\Video\Mother\new\2002\deint2002.avs, line 9)
---------------------------
ОК
---------------------------

yup.

[ttatcs]

Originally Posted by -Vit-

Is there a line number with that error message?

Edit: I've tested a Lagarith YUY2 interlaced AVI and it was fine. So...
- That is mt_masktools25.dll you are using?
- You don't have both masktools25 and masktools26 in your plugins do you?
- Could you upload a small piece of that source...

Thanks Vit , Yup and Didée . Your suggestions are priceless .
I kept both "mt_masktools-25.dll" and "mt_masktools-26.dll
" in the plugin directory .

After I put "mt_masktools-26.dll" out of that directory then the error was gone.

Really appreciate your work .

[Dogway]

Im getting heavy ghosting using QTGMC for deshimmer in an IVTC'ed anime source.
I use this line, is this normal?

QTGMC(Preset="Slower", InputType=1,sharpness=0.0)

[SubJunk]

Try different values for InputType (0-3)

[-Vit-]

Although I know a number of people use it with success, QTGMC is not really designed for anime. QTGMC relies on identifying the motion from frame to frame. With a film source the frame rate is fairly high and movements between frames are usually small; the detail of film means good matches can often be found. Anime will run at a lower rate, may contain duplicate frames, have stylized/artificial movement, and be hard to track if there is a lack of fine detail. So motion matches on anime can be poor, which will lead to ghosting on the temporal steps of the algorithm (and possible blurring / halo strengthening). The same problems can apply to game captures to a lesser extent.

Tweaking parameters might help a little [for example, reduce the temporal radiuses - QTGMC( "Slower", TR0=1,TR1=1,TR2=0) ], but if you're getting heavy ghosting then it may simply be that QTGMC is not going to work for this task.

[nhope]

For anyone who is interested, I did some 1080-60i to 480i DVD test renders and converted them to a ready-to-burn DVD project for comparison. One of them uses the "IResize" script referred to a couple of pages back, and another includes a QTGMC pass as suggested by Didée. See here.

[yup]

To all interested about support YUY2.
For all Avisynth build (2.58, 2.58 MT, IanB 2.6, SEt 2.6) will be properly work only mt_masktools-25***.dll, because mt_masktools-26***.dll need YV16 planar colorspace (not from Interlevead2Planar).
see http://forum.doom9.org/showthread.ph...83#post1497883
If we want use 26 masktools we could work with 3 types colorspace YUY2, hacked Kassandro (Interleaved2Planar) and native YV16 it is ridiculous.
yup.

[Heaud]

I've been getting errors running QTGMC 3.31 when using qtgmc("slow", shutterblur=2, fpsdivisor=2) or any other faster preset. The error displays as "super clip does not contain needed color data". Could this be due to that slight speedup in a previous version?

[-Vit-]

Originally Posted by Heaud

I've been getting errors running QTGMC 3.31 when using qtgmc("slow", shutterblur=2, fpsdivisor=2) or any other faster preset. The error displays as "super clip does not contain needed color data". Could this be due to that slight speedup in a previous version?

Well spotted and the correct reason, thanks. I've updated the first post with a fix (v3.32). There are some other very minor changes in there as well.

[K1CS1K]

Hy All!

I'd like you to check some of my footages and give me some advice.
I use QTGMC mainly for deinterlacing video game materials and I'd be glad if you could help me in some ways.

I've uploaded a .zip file here including the following files:

-3 .avs files (details below)
-source file (150 frames), UtVideo codec (in this case ULY0 due to QTGMC's obvious colorspace requirements)
-a test conversion with my script I normally use (x86 edition, with my PC it required 38m52s)
-a test conversion (x64) with the same specs I use in the .avs for x86 (12m1s but because of the quality loss and huge 'artifact' problem or something, I don't know what to call this deterioration of the frames)
-a test file (x64) with modified specs from my x86 .avs (35m35s ; see 'x64 without Blocksize & Overlap.avs' below)
-last test material (x64) with VirtualDub x86 (23m46s & exact filesize as x86's ; somehow x86 & x64 avisynths were installed at the same time but independently, Windows\System32 contained only 32-bit files for the process)

For all normal 64-bit conversions I used Veedub64 and in every case the output codec was ULY2 (in my opinion it's the fastest and makes the best possible results after the 'almighty' raw uncompressed; after a few identical tests it was waaay much faster than using the uncompressed output format in this case of ). All the 64-bit and 32-bit plugins needed for QTGMC are the latest, I can list them with the version numbers if you would like to. When I first met QTGMC, I read literally all the intructions in the QTGMC.avsi (now we know it's separated into 2 files) and after that I started to use these following avisynth scripts I found the toughest :

x86.avs:

SetMTMode(5,0)

AVISource("test.avi")

QTGMC( Preset="Placebo", NNeurons=4, EdiQual=2, Sharpness=0.25, SubPel=4, Blocksize=32, Overlap=16, SourceMatch=3, MatchPreset="Placebo", MatchPreset2="Slower", MatchEnhance=0.6, Lossless=2, Sigma=3, GrainRestore=0.5, FPSDivisor=2, ChromaMotion=false )

--------------------

x64.avs (the same as the x86 edition above but contains redirection to 64-bit plugins after installing avisynth64):

SetMTMode(5,0)

AVISource("test.avi")

SetWorkingDir("c:\Program Files (x86)\AviSynth 2.5\plugins64")

QTGMC( Preset="Placebo", NNeurons=4, EdiQual=2, Sharpness=0.25, SubPel=4, Blocksize=32, Overlap=16, SourceMatch=3, MatchPreset="Placebo", MatchPreset2="Slower", MatchEnhance=0.6, Lossless=2, Sigma=3, GrainRestore=0.5, FPSDivisor=2, ChromaMotion=false )

--------------------

x64 without Blocksize & Overlap.avs (after removing these specs from the script, the image became as clean as it should be):

SetMTMode(5,0)

AVISource("test.avi")

SetWorkingDir("c:\Program Files (x86)\AviSynth 2.5\plugins64")

QTGMC( Preset="Placebo", NNeurons=4, EdiQual=2, Sharpness=0.25, SubPel=4, SourceMatch=3, MatchPreset="Placebo", MatchPreset2="Slower", MatchEnhance=0.6, Lossless=2, Sigma=3, GrainRestore=0.5, FPSDivisor=2, ChromaMotion=false )

--------------------

Usually for normal length gameplay videos (4-5 mins) the average conversion speed is 0,25fps with my PC & the .avs files above.

Please help me in improving my script and make the best out of the source material.

Thanks in advance,
K1CS1K

Our YouTube channel

[-Vit-]

That source is not interlaced. It's progressive that has been weaved strangely. One field of each frame exactly matches the other field of the next frame. A capture issue maybe? Anyway, no deinterlacing is needed:

AviSource( "YourPathTo\test_150_frames.avi" )
ComplementParity() # Might not always need this line
SeparateFields().Trim(1,0).Weave()

The source is sampled strangely horizontally, as though every 7th and 8th column have been swapped. That is either my decoder (I rarely use Ut), or it might be your emulator, capture process, encoder or something... You would need to fix the problem component or write a custom script if the problem is at your end.

You might prefer the look of your encodes because the QTGMC temporal processing smooths things a little. However, QTGMC is not the right tool for this job. [Those QTGMC settings were overkill for any source. MatchPreset="Placebo" is never worth it.]

[nhope]

-Vit-, I re-labelled the preset tables "3.32". You can all sleep soundly again now.

[egrimisu]

maybe it's out topic but shall i compress with qtgmc using selecteven if i want to deinterlace a pal dv source. I want to transfer it to dvd but i'm not sure if dvd players are compatible with pal 50fps progresive. Hope you cand help.

[Boulder]

I think you might want to experiment with the single-rate output options in QTGMC instead of using QTGMC+SelectEven. You could get a more film-like appearance to the result.

[egrimisu]

so basicly that means that i can't use 50fps in a pal dvd. i haven't found any single-rate option in qtgmc, plese guide me.

[Boulder]

You cannot use 50fps streams for DVD. See the "Shutter Speed Motion Blur / Frame Rate" section in the QTGMC documentation, that's where the magic is.