QTGMC good results

[Terka]

Hi, have a new machine gtx 1660 and Ryzen 3900x.
Would like to postprocess my DV recordings to get very good quality output. can you please recommend witch Avisynth version to use and which settings?

[Boulder]

Latest Avisynth+ and QTGMC + utility script by real.finder. Then just try QTGMC("very slow", tr2=x) where x is 0-2. I personally use tr2=0 since I do denoising outside QTGMC.

[shakaraba]

Hi Terka,

I have made a QTGMC Helper to make testing easier. See attached file.

I recommend you to follow Boulder Recommendation of installing latest versions and get QTGMC working first.

To use the helper, you must put in the same folder the video "input.avi". Then you can open the .avs file in VirtualDub2 to see the results. You must edit the .avs file to change parameters and then press F2 in VirtualDub to refresh changes.

To use the helper you will need:

# 0. Avisynth+ MT r2772 or higher - https://github.com/pinterf/AviSynthPlus/releases [AviSynthPlus-MT-r2772-with-vc_redist.exe]
# 1. QTGMC v3.364s - http://avisynth.nl/index.php/QTGMC
# 2. UnsharpMask - http://avisynth.nl/index.php/WarpSharp/UnsharpMask
# 3. vsMSmooth v1.0.0 - http://avisynth.nl/index.php/VsMSmooth
# 4. GradFun2DBmod v1.5 - http://avisynth.nl/index.php/GradFun2DBmod

To install Avisynth+ and all the filter you can follow this video (not mine): https://www.youtube.com/watch?v=C4PyyQoz6eo

Some tips:
0. Installing Avisynth+ is easy.
1. It is very important to follow the FFT instructions of QTGMC.
Place the dll and the copy of the renamed dll to:
32 bit: %SystemRoot%\SysWOW64\
64bit: %SystemRoot%\System32\

2. This will sharp the video like the Unsharpmask from VirtualDub2.
3. This will hide compression and dot grain without loosing detail.
4. If you denoise too much you will need this deband filter.

With the helper you can test and compare multiple QTGMC settings and see what you like best.

I have a Ryzen 2600 with GTX1070Ti and get around 15 fps for an interlaced DV video (720x576), I use mode 2 that is for Heavy denoising.

The QTGMC command for this mode is:

QTGMC( SLMode=2, Sigma=15, ThSAD1=1280, ThSAD2=512, ThSCD1=220, ThSCD2=105, TR2=6, NoisePreset="Slower", Preset="placebo", NoiseProcess=1, NoiseRestore=0.0, GrainRestore=0.0, InputType=0)

Increase Sigma to add more denoising.
If you have a progressive video then use InputType=1.

If you want to Denoise outside QTGMC (NeatVideo or Topaz Video Enhance AI) then you can use mode 0 for best quality.

The QTGMC call is:
QTGMC( Sharpness=0, TR2=1, Preset="placebo", NoiseProcess=0)

You can change the preset if you want to trade quality for speed.

Compare modes:
https://i.ibb.co/XbgY1hv/modes.png (https://ibb.co/rmSdYC1)

Compare Sigma to test Denoiser strength
https://i.ibb.co/5Mjq1ZH/sigma.png (https://ibb.co/p0KC1qF)