Selur's Little Message Board

Pages: 1 2 3

In SMDegrain, there are certain combinations of refine motion, block & overlap size, and color format which cause error in VS preview. Are they bugs, or supposed so?

In Hybrid_dev_2026.06.25-152245 (old version of SMDegrain):

YUV410p8, YUV411p8: always error

refine motion on + overlap size 2 + YUV ( != 444 ) : error
Not happening in refine motion off / overlap size != 2 / YUV444&GRAY

In Hybrid_dev_2026.10.02-19438:

YUV410p8, YUV411p8: always error

and in other situation, if YUV ( != 444 ), these combination will cause error:

refine motion 1/2/3 pass + size(block,overlap) = (32,2)
refine motion 2/3/4 pass + size(block,overlap) = (64,4)

sample code & error output: refine motion 1 pass + size(block,overlap) = (32,2)

Code:
# Imports
import sys
import os
import vapoursynth as vs
# using Vapoursynth R79
# getting Vapoursynth core
core = vs.core
# Import scripts folder
scriptPath = 'D:/APPD/Hybrid_dev_20261002-19438/64bit/vsscripts'
sys.path.insert(0, os.path.abspath(scriptPath))
# loading plugins
core.std.LoadPlugin(path="D:/APPD/Hybrid_dev_20261002-19438/64bit/Vapoursynth/Lib/site-packages/vapoursynth/plugins2/akarin/libakarin.dll")
core.std.LoadPlugin(path="D:/APPD/Hybrid_dev_20261002-19438/64bit/Vapoursynth/Lib/site-packages/vapoursynth/plugins2/mvutensils/mvutensils.dll")
core.std.LoadPlugin(path="D:/APPD/Hybrid_dev_20261002-19438/64bit/Vapoursynth/Lib/site-packages/vapoursynth/plugins2/fmtconv.dll")
core.std.LoadPlugin(path="D:/APPD/Hybrid_dev_20261002-19438/64bit/vsfilters/SourceFilter/LSmashSource/LSMASHSource.dll")
# Import scripts
import nnedi3_resample
import smdegrain
import validate
# Source: 'C:\Users\mmddffkk\Desktop\tmp\SAMPLE.mkv'
# clip current meta; color space: YUV420P8, bit depth: 8, resolution: 1920x1080, fps: 23.976, color matrix: 709, color primaries: Unspecific, color transfer: Unspecified, yuv luminance scale: limited, scanorder: progressive, full height: true (Source)
# Loading 'C:\Users\mmddffkk\Desktop\tmp\SAMPLE.mkv' using LWLibavSource
clip = core.lsmas.LWLibavSource(source="C:/Users/mmddffkk/Desktop/tmp/SAMPLE.mkv", format="YUV420P8", stream_index=0, cache=0, prefer_hw=0)
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.
prop_name = '_Range' if core.core_version.release_major >= 74 else '_ColorRange'
clip = core.std.SetFrameProps(clip=clip, **{prop_name: vs.RANGE_LIMITED})
# making sure frame rate is set to 23.976fps
clip = core.std.AssumeFPS(clip=clip, fpsnum=24000, fpsden=1001)
# making sure the detected scan type is set (detected: progressive)
clip = core.std.SetFrameProps(clip=clip, _FieldBased=vs.FIELD_PROGRESSIVE) # scan type: progressive
# removing grain using SMDegrain
clip = smdegrain.SMDegrain(clip, RefineMotion=1, interlaced=False, overlap=2, blksize=32, thSCD1=1600, tools={'mv': 'mvutensils', 'expr': 'akarin', 'nnedi3': 'znedi3', 'dfttest': 'dfttest', 'nlmeans': 'nlm_ispc', 'bm3d': 'bm3d', 'rg': 'zsmooth', 'median': 'zsmooth', 'boxblur': 'std', 'scd': 'std'})
# adjusting output color from YUV420P8 to YUV420P10 for SvtAv1Model
clip = core.resize.Bicubic(clip=clip, format=vs.YUV420P10)
# set output frame rate to 23.976fps (progressive)
clip = core.std.AssumeFPS(clip=clip, fpsnum=24000, fpsden=1001)
# output
clip.set_output()
# clip current meta; color space: YUV420P10, bit depth: 10, resolution: 1920x1080, fps: 23.976, color matrix: 709, color primaries: Unspecific, color transfer: Unspecified, yuv luminance scale: limited, scanorder: progressive, full height: true (Meta)
# script was created by Hybrid 2026.10.02.1

Code:
Failed to evaluate the script:
Python exception: Degrain: Degrain2: Failed to retrieve first frame from super clip. Error message: Recalculate: The chosen block size has no multiple that will process the entire frame without exceeding the super clip padding, derive a new suitable super clip and try again

Traceback (most recent call last):
File "vapoursynth.pyx", line 3832, in vapoursynth._vpy_evaluate
File "vapoursynth.pyx", line 3833, in vapoursynth._vpy_evaluate
File "\\fx705d\ENCTMP\tempPreviewVapoursynthFile02_48_22_671.vpy", line 41, in 
clip = smdegrain.SMDegrain(clip, RefineMotion=1, interlaced=False, overlap=2, blksize=32, thSCD1=1600, tools={'mv': 'mvutensils', 'expr': 'akarin', 'nnedi3': 'znedi3', 'dfttest': 'dfttest', 'nlmeans': 'nlm_ispc', 'bm3d': 'bm3d', 'rg': 'zsmooth', 'median': 'zsmooth', 'boxblur': 'std', 'scd': 'std'})
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "D:\APPD\Hybrid_dev_20261002-19438\64bit\vsscripts\smdegrain.py", line 259, in SMDegrain
output = MV.Degrain(mfilter, super_render, *vectors, thsad=thSAD, thsadc=thSADC, plane=plane, limit=limit, limitc=limitc, thscd1=thSCD1, thscd2=thSCD2,
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "D:\APPD\Hybrid_dev_20261002-19438\64bit\vsscripts\misc.py", line 1410, in Degrain
return self._degrain(clip, super, *vectors, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "D:\APPD\Hybrid_dev_20261002-19438\64bit\vsscripts\misc.py", line 1352, in _degrain
return core.mvu.Degrain(clip, super, vec_list, limit=mvu_limit, **weight_args)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "vapoursynth.pyx", line 3559, in vapoursynth.Function.__call__
vapoursynth.Error: Degrain: Degrain2: Failed to retrieve first frame from super clip. Error message: Recalculate: The chosen block size has no multiple that will process the entire frame without exceeding the super clip padding, derive a new suitable super clip and try again

Code:
YUV410p8,  YUV411p8: always error

I assume you mean: YUV420P8, YUV422P8, since that is what your script uses.
Seems like I need to add some padding when mvutensils is used.
As a workaround you can set SMDegrain->Tools->MV to MV.
From what I can see, this is not a bug, but a limitation of mvutensils which I was not aware of.

=> I'll look at it tomorrow and think of a general workaround. (my guess is that this might also happen in some other filters)

Cu Selur

Replacing the files in Hybrid\64bit\vsscripts with the latest from https://github.com/Selur/VapoursynthScriptsInHybrid should fix the problem.
(will release new dev version with this tomorrow)

Cu Selur

Hybrid_dev_2026.10.04-204526 fixed the issue mentioned above.

There's another thing about thSCD1, by design Hybrid will adjust the value of thSCD1 according to the fixed block size I choose. and I think there are some bugs during adjusting thSCD1.

case 1: choose block size 4, and thSCD1 will be set to 100, and 8->400 & 16->1600, now dial back to block size 8 or 4, thSCD1 won't dial back to 400/100, it stay at 1600.

case 2. Hybrid do not adjust thSCD1 if block size is 32 or above. In this case thSCD1 might be 0 or any value previously adjusted and left in the field of thSCD1. I don't know what value thSCD1 should be when block size >=32, but if the value is 0, the output video look like that it's never filtered by smdegrain, all noise are still there.

Will look at it when I'm back home,...

It's probably not a bug.
The assumtion that a specific blacksize enforces a specific threshold is wrong.
Changing the block size will result in a change of only the minimum threshold.
So unless the value is not below that minimum it will not change.
=> will look at it and do some testing.

Cu Selur

Okay, there were bugs.
Thanks for the report, both cases were real bugs.

What happened: Hybrid didn't actually set thSCD1 from the block size, it only raised the lower limit of the field (4: 100, 8: 400, 16: 1600). So the value went up, but
never back down, and for block sizes 32 and above nothing was adjusted at all.
With thSCD1=0 every block counts as "changed", so every frame is treated as a scene change and SMDegrain filters nothing - that's why your output looked untouched.

While looking into it I also measured that the coupling isn't needed at all: mvtools and mvutensils both normalize thSCD1 to an 8x8 block internally, so 400 means the same for every block size.

Changed for the next version:
- thSCD1 is no longer tied to the block size and is available with any block size (including "auto").
- 0 now means "auto" (400, or 0.35*thSAD+260 with "4.x formulas") and is never passed on to the script; otherwise the range is 1-16320.

Until then: with a fixed block size simply enter thSCD1 by hand (400 is the normal default), and don't leave it at 0.

Cu Selur

Uploaded a new dev with adjusted smdegrain script and ui.
(and tons of other small bugs, I ran into while preparing for Vapoursynth R80+)

Cu Selur

Built-in SMDegrain in Hybrid_dev_2026.10.05-203137 works ok by itself.

But if I install the whole addon (vsmlrt, vsgan, onnx model, and both VapoursynthR79_torch_2026.09.26 & VapoursynthR79_torch_2026.10.03 were tested), SMDegrain will not work at all, job crashed immediately. guess the vstorch addon needs adjusted too?

Hmm, I expected VapoursynthR79_torch_2026 to work, but will look at it when I'm back home.
Does Vapoursynth Preview show an error?

Cu Selur

Pages: 1 2 3

andrewschen

Selur

Selur

andrewschen

Selur

Selur

Selur

Selur

andrewschen

Selur