Selur's Little Message Board

Hello everyone, Smile

Thank you very much for creating such a great tool, and for maintaining it over the years. It is greatly appreciated.
I'm writing here in this forum to ask for your help with a couple of problems that I can't seem to solve.

problem 1:
When I use the 'def' option in QTGMC, it doesn't work. I get an error message that I don't know how to resolve or what causes it. (The process stops when I start it and does not process anything.)
[Image: gqBua6w.png]

Here is the message that appears at the bottom of the Vapoursynth Editor.

Code:
2025-09-21 03:37:24.733

VapourSynth plugins manager: Failed to load vapoursynth library!

Please set up the library search paths in settings.

2025-09-21 03:37:46.578

Failed to evaluate the script:

Python exception: Degrain1: failed to retrieve first frame from super clip. Error message: 

Traceback (most recent call last):

File "src/cython/vapoursynth.pyx", line 3207, in vapoursynth.publicFunction

File "src/cython/vapoursynth.pyx", line 3209, in vapoursynth.publicFunction

File "src/cython/vapoursynth.pyx", line 832, in vapoursynth.FuncData.__call__

File "C:\Program Files\Hybrid\64bit\Vapoursynth\Lib\site-packages\vs_deepdeinterlace\DeF.py", line 131, in process_frame

output_img = inference(arrays, even_values, model, device, fp16)

^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

File "C:\Program Files\Hybrid\64bit\Vapoursynth\Lib\site-packages\vs_deepdeinterlace\DeF.py", line 85, in inference

re_frame = model(input).cpu()

^^^^^^^^^^^^

File "C:\Program Files\Hybrid\64bit\Vapoursynth\Lib\site-packages\torch\nn\modules\module.py", line 1751, in _wrapped_call_impl

return self._call_impl(*args, **kwargs)

^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

File "C:\Program Files\Hybrid\64bit\Vapoursynth\Lib\site-packages\torch\nn\modules\module.

Traceback (most recent call last):

File "src/cython/vapoursynth.pyx", line 3378, in vapoursynth._vpy_evaluate

File "src/cython/vapoursynth.pyx", line 3379, in vapoursynth._vpy_evaluate

File "C:\Temp\Hybrid-temp\tempPreviewVapoursynthFile03_37_24_564.vpy", line 59, in 

clip = qtgmc.QTGMC(Input=clip, Preset="Slower", InputType=0, TFF=True, TR2=1, SourceMatch=3, Lossless=1, MatchPreset="Slower", MatchPreset2="Slower", opencl=True, EdiExt=clipEdiExt) # new fps: 29.97

^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

File "C:\Program Files\Hybrid\64bit\vsscripts\qtgmc.py", line 914, in QTGMC

degrain1 = core.mv.Degrain1(edi, ediSuper, bVec1, fVec1, thsad=ThSAD1, thscd1=ThSCD1, thscd2=ThSCD2) # 0.00 0.33 0.33 0.33 0.00

^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

File "src/cython/vapoursynth.pyx", line 3114, in vapoursynth.Function.__call__

vapoursynth.Error: Degrain1: failed to retrieve first frame from super clip. Error message: 

Traceback (most recent call last):

File "src/cython/vapoursynth.pyx", line 3207, in vapoursynth.publicFunction

File "src/cython/vapoursynth.pyx", line 3209, in vapoursynth.publicFunction

File "src/cython/vapoursynth.pyx", line 832, in vapoursynth.FuncData.__call__

File "C:\Program Files\Hybrid\64bit\Vapoursynth\Lib\site-packages\vs_deepdeinterlace\DeF.py", line 131, in process_frame

output_img = inference(arrays, even_values, model, device, fp16)

^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

File "C:\Program Files\Hybrid\64bit\Vapoursynth\Lib\site-packages\vs_deepdeinterlace\DeF.py", line 85, in inference

re_frame = model(input).cpu()

^^^^^^^^^^^^

File "C:\Program Files\Hybrid\64bit\Vapoursynth\Lib\site-packages\torch\nn\modules\module.py", line 1751, in _wrapped_call_impl

return self._call_impl(*args, **kwargs)

^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

File "C:\Program Files\Hybrid\64bit\Vapoursynth\Lib\site-packages\torch\nn\modules\module.

------------------------------------------------------------------------------------------------------------------
------------------------------------------------------------------------------------------------------------------
------------------------------------------------------------------------------------------------------------------
Problem 2 : I am trying to replicate this flow in hybrid, but I cannot figure out how to generate the VFR times and then insert them into the final container.

4-Step Workflow
1️⃣ TFM + TDecimate → CFR + Timecodes

Generate a progressive (CFR) clip and a timecodes file for VFR.
No AI is applied at this stage.

2️⃣ Apply AI (DEF / vs_deepdeinterlace)

Use the AI model to remove ghosting, shimmer, and improve frame quality.
This is applied on the CFR clip generated from step 1.

3️⃣ Render the AI-enhanced video

Export a clean, progressive MKV/MP4 with all visual improvements.

4️⃣ Reapply VFR using the timecodes

Use the timecodes file to reconstruct the final VFR video, preserving the original timing and rhythm.
Example with

Code:
mkvmerge -o final_output.mkv --timestamps 0:timecodes.txt video_ai.mkv

How can I do the same thing in Hybrid? What is the most reliable method that provides the best quality? And can it be accelerated with a powerful nvidia GPU?.

Thank you very much for your time. I hope someone can help me. Heart

Quote:I am trying to replicate this flow in hybrid, but I cannot figure out how to generate the VFR times and then insert them into the final container.

You can not.
Hybrid does not support deinterlacing VFR content while keeping it vfr.
This is, because:

I think it's total madness to create such content.
the only cases where I have seen such content is when MakeMKV created it, out of DVDs or Blu-rays. But instead of VFR and interlaced the content should have been flagged CFR and telecined.
simple extracting and reapplying the time codes isn't possible, since deinterlacing and IVTC can change the frame count!

=> I suspect your content isn't really interlaced and vfr to begin with. Judging by the name of your input file, your content should not be vfr, since Blu-rays

Quote:"TFM + TDecimate → CFR + Timecodes"

Hybrid doesn't support mode 3 or 5 for TDecimate, also those modes are not meant to be applied on VFR content, but on CFR content.

Side note: those steps seem like something some ai chat bot came up with,...

Quote: Apply AI (DEF / vs_deepdeinterlace)

Use the AI model to remove ghosting, shimmer, and improve frame quality.
This is applied on the CFR clip generated from step 1.

Hybrid also does not support applying DeF, which is for deinterlacing, on progressive content. (after step 1. the content should be progressive)
Also from my experience DeF does not really remove ghosting, it can improve some of the shimmering (but that could be addressed better) and it might improve the quality of the deinterlacing compared to some deinterlacers on some content.

About the crash of the preview: Sadly, I can't reproduce the issue.
using the same QTGMC settings you did:

Code:
# Deinterlacing using QTGMC

clipEdiExt = clip

# adjusting color space from YUV420P8 to RGBS for vsDDD

clipEdiExt = core.resize.Bicubic(clip=clipEdiExt, format=vs.RGBS, matrix_in_s="470bg", range_in_s="limited", range_s="full")

clipEdiExt = vs_deepdeinterlace.DeF(clip=clipEdiExt, tff=False, fp16=True, device="cuda")

# Making sure content is preceived as frame based

clipEdiExt = core.std.SetFrameProps(clip=clipEdiExt, _FieldBased=vs.FIELD_PROGRESSIVE) # progressive

clipEdiExt = core.resize.Bicubic(clip=clipEdiExt, format=vs.YUV420P8, matrix_s="470bg", range_in_s="full", range_s="limited", dither_type="error_diffusion") # additional resize to allow target color sampling

clip = qtgmc.QTGMC(Input=clip, Preset="Slower", InputType=0, TFF=False, TR2=1, SourceMatch=3, Lossless=1, MatchPreset="Slower", MatchPreset2="Slower", opencl=True, EdiExt=clipEdiExt) # new fps: 29.97

# Making sure content is preceived as frame based

clip = core.std.SetFrameProps(clip=clip, _FieldBased=vs.FIELD_PROGRESSIVE) # progressive

clip = clip[::2] # selecting previously even frames

works fine here.
I assume you have a GPU with at least 16GB VRAM and supports FP16, since when I apply Def in 1920x1080 even with FP16 (QTGMC always does this when DeF is used) uses all the 16 GB VRAM of my GeForce RTX 4080 (and takes ages).
(also eats a lot of normal RAM, around ~30GB peeking to around 43GB)

=> If you can share a sample of your source (i.e. by splitting it using mkvtoolnix) I can look at that sample and
a. try to reproduce the problem
b. try to suggest something.

Cu Selur

Ps.: Playing around, I could reproduce the crash here, when my system ran out of vram, so I guess that is the same issue you are having. (to be frank applying DeF on HD content seems simply not really feasible and worth it)

Thank you very much for your response. It is greatly appreciated. Could you please advise me on the best way to deal with this type of content in your program to obtain the best possible result? Thank you for your help.

https://drive.google.com/file/d/1lpj3UnU...drive_link

https://www.mediafire.com/file/gb7v4i5bs...D.mkv/file

--------------------------------------------------------------------------------------------------------
Regarding the DEF problem, here is the information about my video card so you can confirm whether it is compatible and, if so, help me solve that problem as well. Thank you.

[Image: g4G0WMc.png]

Quote:Could you please advise me on the best way to deal with this type of content in your program to obtain the best possible result?

No clue about the 'best possible' way, but here's how I would start.
Looking at the file.
First thing I noticed, when bobbing the content, every frame is duplicated, so the content isn't really interlaced.
Looking at the content when overwriting the scan type to progressive, you see combing.
The content isn't interlaced (at least not all the time), but probably field shifted or telecined (or a mix). Smile

Looking at the fields separated:

so there is already ghosting in some of the fields. Sad

When disabling bobbing, you still see duplicates, so the content probably is telecined.
So either using TFM+TIVTC or QTGMC(Bob)+sRestore(frate=23.9760), neither of these is perfect.

=> For interlaced handling, I would probably use TFM+TIVTC and add Santiag (or Vinverse2) after it.
For denoising&co I would probably use MCDegrainSharp and sharpen additionally a bit with CAS(0.6 or 0.7). But I can't really say whether that is something you might want to do. (Additionally, adding DPIRDeblock(mlrt) at the end might help)
So here's the script I end up with:

Code:
# Imports

import vapoursynth as vs

# getting Vapoursynth core

import logging

import site

import sys

import os

core = vs.core

# Limit frame cache to 48473MB

core.max_cache_size = 48473

# Import scripts folder

scriptPath = 'F:/Hybrid/64bit/vsscripts'

sys.path.insert(0, os.path.abspath(scriptPath))

os.environ["CUDA_MODULE_LOADING"] = "LAZY"

# Force logging to std:err

logging.StreamHandler(sys.stderr)

# loading plugins

core.std.LoadPlugin(path="F:/Hybrid/64bit/vs-mlrt/vstrt.dll")

core.std.LoadPlugin(path="F:/Hybrid/64bit/vsfilters/SharpenFilter/CAS/CAS.dll")

core.std.LoadPlugin(path="F:/Hybrid/64bit/vsfilters/Support/libsangnom.dll")

core.std.LoadPlugin(path="F:/Hybrid/64bit/vsfilters/Support/fmtconv.dll")

core.std.LoadPlugin(path="F:/Hybrid/64bit/vsfilters/Support/EEDI2.dll")

core.std.LoadPlugin(path="F:/Hybrid/64bit/vsfilters/Support/EEDI3m_opencl.dll")

core.std.LoadPlugin(path="F:/Hybrid/64bit/vsfilters/ResizeFilter/nnedi3/NNEDI3CL.dll")

core.std.LoadPlugin(path="F:/Hybrid/64bit/vsfilters/DeinterlaceFilter/TIVTC/libtivtc.dll")

core.std.LoadPlugin(path="F:/Hybrid/64bit/vsfilters/SourceFilter/DGDecNV/DGDecodeNV_AVX2.dll")

# Import scripts

from importlib.machinery import SourceFileLoader

vsmlrt = SourceFileLoader('vsmlrt', 'F:/Hybrid/64bit/vs-mlrt/vsmlrt.py').load_module()

from vsmlrt import Backend

import antiAliasing

import validate

# Source: 'C:\Users\Selur\Desktop\C-15-Genesis - Three Sides Live (1982) [2014 Blu-ray].mkv'

# Current color space: YUV420P8, bit depth: 8, resolution: 1920x1080, frame rate: 29.97fps, scanorder: telecine, yuv luminance scale: limited, matrix: 709, transfer: bt.709, primaries: bt.709, format: AVC

# Loading C:\Users\Selur\Desktop\C-15-Genesis - Three Sides Live (1982) [2014 Blu-ray].mkv using DGSource

clip = core.dgdecodenv.DGSource("J:/tmp/mkv_4baacdb2bacedf5b6d10360e0d08f745_853323747.dgi",fieldop=0)# 29.97 fps, scanorder: telecine

frame = clip.get_frame(0)

# setting color matrix to 709.

clip = core.std.SetFrameProps(clip, _Matrix=vs.MATRIX_BT709)

# setting color transfer (vs.TRANSFER_BT709), if it is not set.

if validate.transferIsInvalid(clip):

  clip = core.std.SetFrameProps(clip=clip, _Transfer=vs.TRANSFER_BT709)

# setting color primaries info (to vs.PRIMARIES_BT709), if it is not set.

if validate.primariesIsInvalid(clip):

  clip = core.std.SetFrameProps(clip=clip, _Primaries=vs.PRIMARIES_BT709)

# setting color range to TV (limited) range.

clip = core.std.SetFrameProps(clip=clip, _ColorRange=vs.RANGE_LIMITED)

# making sure frame rate is set to 29.97fps

clip = core.std.AssumeFPS(clip=clip, fpsnum=30000, fpsden=1001)

# making sure the detected scan type is set (detected: telecine)

clip = core.std.SetFrameProps(clip=clip, _FieldBased=vs.FIELD_TOP) # tff

# Deinterlacing using TIVTC

clip = core.tivtc.TFM(clip=clip)

clip = core.tivtc.TDecimate(clip=clip)# new fps: 23.976

# Making sure content is preceived as frame based

clip = core.std.SetFrameProps(clip=clip, _FieldBased=vs.FIELD_PROGRESSIVE) # progressive

# applying anti aliasing using santiag

clip = antiAliasing.santiag(c=clip, nns=2, qual=2, pscrn=2, opencl=True)

clip = core.std.Crop(clip=clip, left=240, right=240, top=0, bottom=0)# cropping to 1440x1080

# contrast sharpening using CAS

clip = core.cas.CAS(clip=clip, sharpness=0.700)

# adjusting color space from YUV420P8 to RGBH for vsDPIRmlrtDeblock

clip = core.resize.Bicubic(clip=clip, format=vs.RGBH, matrix_in_s="709", range_in_s="limited", range_s="full")

# adjusting deblocking using DPIR (mlrt)

clip = vsmlrt.DPIR(clip = clip, strength=50.000, model=3, backend=Backend.TRT(fp16=True,device_id=0,bf16=False,verbose=True,use_cuda_graph=True,num_streams=3,builder_optimization_level=3,engine_folder="J:/TRT"))

# adjusting output color from: RGBH to YUV420P10 for NVEncModel

clip = core.resize.Bicubic(clip=clip, format=vs.YUV420P10, matrix_s="709", range_in_s="full", range_s="limited", dither_type="error_diffusion") # additional resize to allow target color sampling

# set output frame rate to 23.976fps (progressive)

clip = core.std.AssumeFPS(clip=clip, fpsnum=24000, fpsden=1001)

# output

clip.set_output()

(note, that I changed the filter order compared to the default order)

Alternatively, with QTGMC and sRestore and something like:

Code:
# Imports

import vapoursynth as vs

# getting Vapoursynth core

import logging

import site

import ctypes

import sys

import os

core = vs.core

# Limit frame cache to 48473MB

core.max_cache_size = 48473

# Import scripts folder

scriptPath = 'F:/Hybrid/64bit/vsscripts'

sys.path.insert(0, os.path.abspath(scriptPath))

# Loading Support Files

Dllref = ctypes.windll.LoadLibrary("F:/Hybrid/64bit/vsfilters/Support/libfftw3f-3.dll")

os.environ["CUDA_MODULE_LOADING"] = "LAZY"

# Force logging to std:err

logging.StreamHandler(sys.stderr)

# loading plugins

core.std.LoadPlugin(path="F:/Hybrid/64bit/vs-mlrt/vstrt.dll")

core.std.LoadPlugin(path="F:/Hybrid/64bit/vsfilters/SharpenFilter/CAS/CAS.dll")

core.std.LoadPlugin(path="F:/Hybrid/64bit/vsfilters/Support/libsangnom.dll")

core.std.LoadPlugin(path="F:/Hybrid/64bit/vsfilters/Support/EEDI2.dll")

core.std.LoadPlugin(path="F:/Hybrid/64bit/vsfilters/DenoiseFilter/ZSmooth/zsmooth.dll")

core.std.LoadPlugin(path="F:/Hybrid/64bit/vsfilters/GrainFilter/AddGrain/AddGrain.dll")

core.std.LoadPlugin(path="F:/Hybrid/64bit/vsfilters/DenoiseFilter/NEO_FFT3DFilter/neo-fft3d.dll")

core.std.LoadPlugin(path="F:/Hybrid/64bit/vsfilters/DenoiseFilter/DFTTest/dfttest2_nvrtc.dll")

core.std.LoadPlugin(path="F:/Hybrid/64bit/vsfilters/Support/EEDI3m_opencl.dll")# vsQTGMC

core.std.LoadPlugin(path="F:/Hybrid/64bit/vsfilters/ResizeFilter/nnedi3/NNEDI3CL.dll")

core.std.LoadPlugin(path="F:/Hybrid/64bit/vsfilters/Support/libmvtools.dll")

core.std.LoadPlugin(path="F:/Hybrid/64bit/vsfilters/Support/fmtconv.dll")

core.std.LoadPlugin(path="F:/Hybrid/64bit/vsfilters/DeinterlaceFilter/Bwdif/Bwdif.dll")

core.std.LoadPlugin(path="F:/Hybrid/64bit/vsfilters/Support/akarin.dll")

core.std.LoadPlugin(path="F:/Hybrid/64bit/vsfilters/SourceFilter/DGDecNV/DGDecodeNV_AVX2.dll")

# Import scripts

from importlib.machinery import SourceFileLoader

vsmlrt = SourceFileLoader('vsmlrt', 'F:/Hybrid/64bit/vs-mlrt/vsmlrt.py').load_module()

from vsmlrt import Backend

import srestore

import antiAliasing

import vs_deepdeinterlace

import dfttest2

import qtgmc

import validate

# Source: 'C:\Users\Selur\Desktop\C-15-Genesis - Three Sides Live (1982) [2014 Blu-ray].mkv'

# Current color space: YUV420P8, bit depth: 8, resolution: 1920x1080, frame rate: 29.97fps, scanorder: top field first, yuv luminance scale: limited, matrix: 709, transfer: bt.709, primaries: bt.709, format: AVC

# Loading C:\Users\Selur\Desktop\C-15-Genesis - Three Sides Live (1982) [2014 Blu-ray].mkv using DGSource

clip = core.dgdecodenv.DGSource("J:/tmp/mkv_4baacdb2bacedf5b6d10360e0d08f745_853323747.dgi",fieldop=0)# 29.97 fps, scanorder: top field first

frame = clip.get_frame(0)

# setting color matrix to 709.

clip = core.std.SetFrameProps(clip, _Matrix=vs.MATRIX_BT709)

# setting color transfer (vs.TRANSFER_BT709), if it is not set.

if validate.transferIsInvalid(clip):

  clip = core.std.SetFrameProps(clip=clip, _Transfer=vs.TRANSFER_BT709)

# setting color primaries info (to vs.PRIMARIES_BT709), if it is not set.

if validate.primariesIsInvalid(clip):

  clip = core.std.SetFrameProps(clip=clip, _Primaries=vs.PRIMARIES_BT709)

# setting color range to TV (limited) range.

clip = core.std.SetFrameProps(clip=clip, _ColorRange=vs.RANGE_LIMITED)

# making sure frame rate is set to 29.97fps

clip = core.std.AssumeFPS(clip=clip, fpsnum=30000, fpsden=1001)

# making sure the detected scan type is set (detected: top field first)

clip = core.std.SetFrameProps(clip=clip, _FieldBased=vs.FIELD_TOP) # tff

# Deinterlacing using QTGMC

clipEdiExt = clip

# adjusting color space from YUV420P8 to RGBS for vsDDD

clipEdiExt = core.resize.Bicubic(clip=clipEdiExt, format=vs.RGBS, matrix_in_s="709", range_in_s="limited", range_s="full")

clipEdiExt = vs_deepdeinterlace.DDD(clip=clipEdiExt, tff=True, fp16=True, device="cuda")

# Making sure content is preceived as frame based

clipEdiExt = core.std.SetFrameProps(clip=clipEdiExt, _FieldBased=vs.FIELD_PROGRESSIVE) # progressive

clipEdiExt = core.resize.Bicubic(clip=clipEdiExt, format=vs.YUV420P8, matrix_s="709", range_in_s="full", range_s="limited", dither_type="error_diffusion") # additional resize to allow target color sampling

clip = qtgmc.QTGMC(Input=clip, Preset="Fast", InputType=0, TFF=True, TR2=1, SourceMatch=0, Lossless=0, opencl=True, EdiExt=clipEdiExt, TR0=2, TR1=2, Rep0=1, Rep1=0, Rep2=4, DCT=5, ThSCD1=300, ThSCD2=110, Sbb=0, NoiseProcess=2, GrainRestore=0.0, NoiseRestore=0.4,  StabilizeNoise=False, NoiseTR=0, NoiseDeint="bob") # new fps: 59.94

# Making sure content is preceived as frame based

clip = core.std.SetFrameProps(clip=clip, _FieldBased=vs.FIELD_PROGRESSIVE) # progressive

# applying anti aliasing using santiag

clip = antiAliasing.santiag(c=clip, nns=2, qual=2, pscrn=2, opencl=True)

# adjusting frame count and rate with sRestore

clip = srestore.sRestoreMUVs(source=clip, frate=23.9760)

clip = core.std.Crop(clip=clip, left=240, right=240, top=0, bottom=0)# cropping to 1440x1080

# contrast sharpening using CAS

clip = core.cas.CAS(clip=clip, sharpness=0.700)

# adjusting color space from YUV420P8 to RGBH for vsDPIRmlrtDeblock

clip = core.resize.Bicubic(clip=clip, format=vs.RGBH, matrix_in_s="709", range_in_s="limited", range_s="full")

# adjusting deblocking using DPIR (mlrt)

clip = vsmlrt.DPIR(clip = clip, strength=50.000, model=3, backend=Backend.TRT(fp16=True,device_id=0,bf16=False,verbose=True,use_cuda_graph=True,num_streams=3,builder_optimization_level=3,engine_folder="J:/TRT"))

# adjusting output color from: RGBH to YUV420P10 for NVEncModel

clip = core.resize.Bicubic(clip=clip, format=vs.YUV420P10, matrix_s="709", range_in_s="full", range_s="limited", dither_type="error_diffusion") # additional resize to allow target color sampling

# set output frame rate to 23.976fps (progressive)

clip = core.std.AssumeFPS(clip=clip, fpsnum=24000, fpsden=1001)

# output

clip.set_output()

(not sure whether this is better, but it is even slower; sadly there is no faster sRestore for Vapoursynth)

Whether you want to denoise or sharpen the content more is up to your preference. (applying for example DetailSharpen seems to be too much for me https://imgsli.com/NDE2ODc5)

Cu Selur

Honestly, dealing with these videos is a headache.

I really appreciate your time and help. Heart

papageno

Selur

papageno

Selur

papageno