PHOENiX's compilation of VapourSynth Script's and Functions
pvsfunc (PHOENiX's VapourSynth Functions) is my compilation of VapourSynth Scripts, Functions, and Helpers.
Install VapourSynth first! (this is different to the pypi/pip
vapoursynth python wrapper!)
pip install --user pvsfunc
Done! However, there are further dependencies listed below that you may need to install depending on the classes you intend to use, and your use-case. Don't forget to install them if needed!
Building from source requires Poetry.
poetry install or
poetry build for distribution wheel and source packages.
This project is released under the GNU GENERAL PUBLIC LICENSE Version 3 (GPLv3) license. Please read and agree to the license before use, it can be found in the LICENSE file.
Below is information about the projects included in pvsfunc that are available to use. Don't take it as full documentation as that is being worked on.
Convenience class for working with DGIndex D2V project files (MPEG-1/2 videos). Includes source loading, frame matching, deinterlacing, and more.
from pvsfunc import PD2V from functools import partial from havsfunc import QTGMC clip = PD2V(r"C:\Users\john\Videos\s01e01.d2v", verbose=True).\ ceil().\ deinterlace( kernel=partial(QTGMC, FPSDivisor=2, Preset="Very Slow"), verbose=True ).\ clip # ... any manual changes to clip clip.set_output()
The above example will load a D2V project file located at
C:\Users\john\Videos\s01e01.d2v in Verbose mode.
Verbose mode will display extra information during the PD2V use.
It then runs
ceil() which frame-matches the progressive sections of the video with the interlaced sections by
duplicating the progressive frames (instead of interlacing).
It then deinterlaces the interlaced sections of the video with QTGMC as the kernel. The Kernel must have a
TFF argument to be compatible, but the field order should not be manually set by the user.
Finally, it takes the clip and set's it for VapourSynth output.
- d2vsource (core.d2v) VapourSynth plugin
- DGIndex v1.5.8 or newer
- mkvextract Only required if you plan on providing non demuxed streams (e.g., mp4, mkv)
To install d2vsource it's as simple as
vsrepo install d2vsource on Windows. Other Operating System user's know the
drill, go check your package repository's or compile it yourself.
Make sure DGIndex and mkvextract is available on your environment path and has execution permissions. Note Linux Users: Add to system profile path, not terminal/rc path. DGIndex is Windows-only but is supported if you install Wine.
Convenience class for working with L-SMASH-WORKS LWI project files. Includes source loading and deinterlacing. More features are to be implemented in the future once a Python-based LWI project parser is available.
Refer to PD2Vs example usage as it's very similar to how PLS is used.
- lsmash (core.lsmas) VapourSynth plugin
- mkvmerge Only required if input file has a container-set frame rate that differs to the encoded frame rate
To install lsmash it's as simple as
vsrepo install lsmas on Windows. Other Operating System user's know the drill,
go check your package repository's or compile it yourself.
Make sure mkvmerge is available on your environment path and has execution permissions. Note Linux Users: Add to system profile path, not terminal/rc path.
Lightweight class to apply de-boxing based on an output aspect-ratio. Similar scripts would annoyingly want you to just crop in yourself which is incredibly annoying.
Decimate (delete) frames in a specified pattern using cycle and offsets. This is typically used for Inverse-Telecine purposes.
Decimation may often be done for IVTC purposes to remove constant pattern pulldown frames.
Kernel storage class for any custom Deinterlacing kernels I'm working on or tinkering with that you may want to use. Just know that they are most likely not the type of deinterlacing you may expect, as it's generally a playground for looking into strange tactics of deinterlacing.
VoidWeave is a deinterlacing method involving in-painting with machine-learning.
It takes footage and separates the weaved fields so that each field is full-height with the missing data of the field being 255 RGB green instead of empty/black. The in-painting machine-learning system would then in-paint the missing data wherever it finds the 255 RGB Green pixels. It works quite well as YUV 4:2:0 DVD data doesn't seem to ever reach 255 green though it does get close, but never quite 255.
I tested it on a Live Action DVD and the results were honestly outstanding! The time it took to run the in-painting was about the same as QTGMC on Very Slow, with similar results! I think where this method may really shine is with Cartoon/Animated sources as QTGMC does not do them well.
It all still requires a lot more testing, but it looks like it could be a really nice method! Especially now that I've learned Disney has also been working on it around the same tim, back in 2020 :P
Release history Release notifications | RSS feed
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.