Quick Primer

VapourSynth scripts are all written using Python. This includes lvsfunc. To write a basic VapourSynth script, there are a couple fundamentals you must understand about Python. This page is written to give you a very basic overview, as well as to explain some common terminology.

Before you begin, I highly encourage you read this page. This includes links to a lot of rudimentary Python exercises and learning material.

Basic usage

lvsfunc is a Python module, which means you must import it into your script before it can be accessed. You can import it by doing the following:

import lvsfunc

Simple, right? But writing “lvsfunc” every time is cumbersome, so we recommend you give it an alias.

import lvsfunc as lvf


The rest of the documentation will assume you alias lvsfunc as lvf.

With the module now imported, you can call functions in your script by referencing the module and writing the function name behind it.

import lvsfunc as lvf

clip = lvf.source("PATH/TO/YOUR/VIDEO")

lvsfunc exposes every function in the global scope, but also exposes every sub-module individually. Both of the following calls will call the same function:

import lvsfunc as lvf

descaled_clip_a = lvf.descale(clip)  # global scope
descaled_clip_b = lvf.scale.descale(clip)  # local scope

Calling them from the relevant sub-module is considered good practice, but for convenience, it may be easier to call the function from the global scope instead.

You can also import individual functions from sub-modules.

from lvsfunc.noise import chickendream

grained_clip = chickendream(clip)

This is useful if you only need a single function and don’t want to pollute your auto-completion with all the other lvsfunc functions.

For further information about specific functions, please refer to their individual documentation. You can find them by scouring the “filters” pages to the left, or using the search bar.

Common terminology

The following video, programming, and VapourSynth-related terminology may be used throughout the documentation:


A Python function is simply code that performs a specific task. lvsfunc comes with a lot of functions that are written to handle with specific video artifacting, and can be adjusted to help solve problematic issues with your video.


A kernel in VS lingo often refers to the setting used for a scaling operation. A common example of this is Bicubic. lvsfunc makes extensive use of the vskernels package for most scaling and format-conversion related tasks.

A kernel, in essence, is simply a class. They must be used as such:

from vskernels import Bicubic

kernel = Bicubic().scale()

The “Bicubic” that was imported is simply a preset that defines the base parameters for the scaling operation. It can be further tweaked by changing the values as such:

kernel = Bicubic(b=0, c=1).scale()

As Bicubic is a class, it has methods. These are the following:


A lot of functions accept a Kernel object. All you need to do is simply pass a Kernel you want to use to the function, and it will use the methods as necessary internally.

lvf.a_function(clip, kernel=Bicubic())

If you run into a kernel-related error, you may need to simply pass the class, not an object!

lvf.a_function(clip, kernel=Bicubic)