DeBlock

Revision as of 22:06, 24 October 2022

Description

It performs deblocking on the picture, using the deblocking filter of H264. It's useful to deblock the result of MVCompensate of MVTools plugin, etc.

Requirements

• [x86]: AviSynth+ or AviSynth 2.6
• [x64]: AviSynth+
• Supported color formats: Y8, YV12, YV16, YV24, YV411
  • AviSynth+: all planar Y and YUV formats (8/10/12/14/16bit and float) are supported.

Syntax and Parameters

Deblock (clip, int "quant", int "aOffset", int "bOffset", int "opt")

clip   =

Input clip, width and height must be mod 8.

int  quant = 25

The higher the quant, the stronger the deblocking. Range is from 0 to 60.

int  aOffset = 0

Quant modifier to the blocking detector threshold. Setting it higher means than more edges will deblocked.

int  bOffset = 0

Another quant modifier, for block detecting and for deblocking's strength. There again, the higher, the stronger.

string  planes = "yuv"

Specifies which planes to process between y, u and v.

int  opt = 0

Sets which cpu optimizations to use.

• 0 : Auto-detect.
• 1 : Use C++ code.
• 2 : Use SSE4.2 code.
• 3 : Use AVX2 code.

Only 8-16bit is optimized, not float.

• If quant + aOffset is less than 16, the filter does nothing at all. The same goes for quant + bOffset.

Examples

Deblock with default settings:

MPEG2Source("Blocky.dv2")
Deblock(quant=25, aOffset=0, bOffset=0, planes="yuv")

To deblock the motion compensation (MVTools plugin):

vectors = source.MVAnalyse(isb = false, lambda = 1000)
compensation = source.MVCompensate(vectors, mode = 0)
compensation.Deblock()

Changelog

Version      Date            Developer       Changes

@9802e03     2020/08/30      299792458m      - Added SIMD optimization for 8-16bit
                                             - Added "opt" parameter
v1.3         2018/05/29      MysteryX        - Back-ported code from VapourSynth
                                             - Added high-bit-depth support
                                             - Added Planes parameter to specify which planes to process
                                             - AviSynth+: self-registers as a MT_NICE_FILTER
v0.9         2013/12/03      tp7             - Faster and supports all planar colorspaces
                                             - Compiled with Microsoft Visual Studio C++ 2012.
v1.2         2006/04/23      Fizick          - Moved code into separate project
v1.1         2006/01/09      Manao/Fizick    - Present in old MvTools (up to v1.1)

Archived Downloads

• *299792458m: performance optimized version; builds compiled with Intel C++ 19.2 without AVX2 requirement.

External Links

• GitHub Source code repository (latest version, v1.3 with SIMD optimizations)
• GitHub Source code repository (older version, v1.3, 2018)
• GitHub Source code repository (older version, v0.9, 2013)

Back to External Filters ←