Deoldify Vapoursynth filter

[Selur]

I noticed some files were still missing from the checkpoints-folder, but that does not explain the R72 vs R68 issue.
=> Uploading new torch add-on. (compressing takes ~6min, uploading takes ~50min)

Cu Selur

[zspeciman]

I did a fresh install with Hybrid_dev_2025.07.19-131522, error in preview remains.

2025-07-19 07:25:04.161
Version mismatch: The VapourSynth Python module version is R72 but the VapourSynth core library is R68. This usually indicates a broken install.
Version mismatch: The VapourSynth Python module version is R72 but the VapourSynth core library is R68. This usually indicates a broken install.
2025-07-19 07:25:08.915
D:\Hybrid\64bit\Vapoursynth\Lib\site-packages\vsdeoldify\fastai\imports\core.py:29: UserWarning: pkg_resources is deprecated as an API. See https://setuptools.pypa.io/en/latest/pkg_resources.html. The pkg_resources package is slated for removal as early as 2025-11-30. Refrain from using this package or pin to Setuptools import pkg_resources

D:\Hybrid\64bit\Vapoursynth\Lib\site-packages\vsdeoldify\fastai\imports\core.py:29: UserWarning: pkg_resources is deprecated as an API. See https://setuptools.pypa.io/en/latest/pkg_resources.html. The pkg_resources package is slated for removal as early as 2025-11-30. Refrain from using this package or pin to Setuptools import pkg_resources

2025-07-19 07:25:23.336
Failed to evaluate the script:
Python exception: Failed to initialize ColorMNet[remote] try ColorMNet[local]

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 "D:\HybridTempPath\tempPreviewVapoursynthFile07_25_03_797.vpy", line 42, in
clip = havc.HAVC_main(clip=clip, VideoTune="veryvivid", ColorFix="none", ColorTune="none", EnableDeepEx=True, DeepExMethod=0, DeepExRefMerge=0, ScFrameDir=None, DeepExModel=0, DeepExEncMode=0, DeepExMaxMemFrames=0, enable_fp16=False)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "D:\Hybrid\64bit\Vapoursynth\Lib\site-packages\vsdeoldify\__init__.py", line 328, in HAVC_main
clip_colored = HAVC_deepex(clip=clip, clip_ref=clip_ref, method=DeepExMethod, render_speed=DeepExPreset,
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "D:\Hybrid\64bit\Vapoursynth\Lib\site-packages\vsdeoldify\__init__.py", line 755, in HAVC_deepex
clip_colored = vs_colormnet(clip, clip_ref, clip_sc, image_size=-1, enable_resize=enable_resize,
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "D:\Hybrid\64bit\Vapoursynth\Lib\site-packages\vsdeoldify\vsslib\vsmodels.py", line 49, in vs_colormnet
return vs_colormnet_remote(clip, clip_ref, clip_sc, image_size, enable_resize, frame_propagate,
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "D:\Hybrid\64bit\Vapoursynth\Lib\site-packages\vsdeoldify\colormnet\__init__.py", line 166, in vs_colormnet_remote
HAVC_LogMessage(MessageType.EXCEPTION, "Failed to initialize ColorMNet[remote] try ColorMNet[local]")
File "D:\Hybrid\64bit\Vapoursynth\Lib\site-packages\vsdeoldify\vsslib\vsutils.py", line 49, in HAVC_LogMessage
raise vs.Error(message_text)
vapoursynth.Error: Failed to initialize ColorMNet[remote] try ColorMNet[local]

[Selur]

Seems like the spatial_correlation_sampler_backend is missing,...
=> I'll create a new package an upload it,.. (check in ~1hour)

[Selur]

new torch add-on is available,..

[zspeciman]

its working now!!!  Thank you so much.

[Dan64]

Hello Selur,

   I was finally able to apply in HAVC some of color adjustment filters that I developed in Vapoursynth described is this post: New AutoColor adjustment filter

   The results are satisfactory as you can see in the picture below

      

   With the previous version (image on the left) the colors were dull with a majority of "rose" color (the color of skin), with the new version (image on the right) both problems has been solved.

   To reduce the impact on GUI I recycled the parameter BlackWhiteTune, which now can have the following values:

:param BlackWhiteTune:      This parameter allows to improve contrast and luminosity of frames colored with HAVC.
                            Allowed values are:
                                    'None' (default)
                                    'Light',
                                    'Medium',
                                    'Strong'
                                    'Light+Custom',
                                    'Medium+Custom',
                                    'Strong+Custom'
                                    'Light+Restore',
                                    'Medium+Restore',
                                    'Strong+Restore'

 
  In the GUI is also necessary to add the following new parameter (a numeric combo-box):

:param BlackWhiteMode:      Method used by BlackWhiteTune to perform colors adjustments.
                             Allowed values are:
                                      0 : Apply Contrast Limited Adaptive Histogram Equalization on Luma
                                      1 : Apply simple Histogram Equalization on all RGB channels
                                      2 : Apply CLAHE on all RGB channels
                                      3 : method=0 and method=1 are merged
                                      4 : Automatic brightness and contrast optimization with ScaleAbs
                                      5 : Auto brigh and contrast optim. with saturation arithmetic (default)

 
I attached the RC1 of new version 5.5.0 and a sample containing the new colorization cases introduced by the new/changed parameters

I hope that you had time to updated the filter and the GUI.

Thanks,
Dan

[Selur]

Thanks for the update.
Working most of the day today, and I'm away with friends this eventing, but I should have some time tomorrow to work on it.

Cu Selur