/AviSynth-vsTTempSmooth

A temporal smoothing filter.

Primary LanguageC++GNU General Public License v3.0GPL-3.0

Description

TTempSmooth is a motion adaptive (it only works on stationary parts of the picture), temporal smoothing filter.

This is a port of the VapourSynth plugin TTempSmooth.

Requirements:

  • AviSynth 2.60 / AviSynth+ 3.4 or later

  • Microsoft VisualC++ Redistributable Package 2022 (can be downloaded from here)

Usage:

vsTTempSmooth(clip, int "ythresh", int "uthresh", int "vthresh", int "ymdiff", bool "umdiff" , bool "vmdiff", int "strength", float "scthresh", bool "fp", int "y", int "u", int "v", clip "pfclip", int "opt", int "pmode", int "ythupd", int "uthupd", int "vthupd", int "ypnew", int "upnew", int "vpnew", int "threads")

Parameters:

  • clip
    A clip to process. It must be Y/YUV(A) 8..32-bit format.

  • maxr
    This sets the maximum temporal radius.
    By the way it works TTempSmooth automatically varies the radius used...
    This sets the maximum boundary.
    pmode=0 - must be between 1 and 7.
    pmode=1 - must be between 1 and 128.
    At 1 TTempSmooth will be (at max) including pixels from 1 frame away in the average (3 frames total will be considered counting the current frame).
    At 7 it would be including pixels from up to 7 frames away (15 frames total will be considered).
    With the way it checks motion there isn't much danger in setting this high, it's basically a quality vs. speed option. Lower settings are faster while larger values tend to create a more stable image.
    Default: 3.

  • ythresh
    Luma threshold for differences of pixels between frames.
    TTempSmooth checks 2 frame distance as well as single frame, so these can usually be set slightly higher than with most other temporal smoothers and still avoid artifacts.
    Must be between 1 and 256.
    Also important is the fact that as long as mdiff is less than the threshold value then pixels with larger differences from the original will have less weight in the average. Thus, even with rather large thresholds pixels just under the threshold wont have much weight, helping to reduce artifacts.
    Default: 4.

  • uthresh, vthresh
    Same as ythresh but for the chroma planes (u, v).
    Must be between 1 and 256.
    Default: uthresh = 5; vthresh =5.

  • ymdiff (only for pmode=0)
    Any pixels with differences less than or equal to mdiff will be blurred at maximum.
    Usually, the larger the difference to the center pixel the smaller the weight in the average.
    mdiff makes TTempSmooth treat pixels that have a difference of less than or equal to mdiff as though they have a difference of 0. In other words, it shifts the zero difference point outwards.
    Set mdiff to a value equal to or greater than thresh-1 to completely disable inverse pixel difference weighting.
    Applied only to the luma plane.
    Must be between 0 and 255.
    Default: 2.

  • umdiff, vmdiff (only for pmode=0)
    Same as ymdiff but for the chroma planes (u, v).
    Must be between 0 and 255.
    Default: umdiff = 3; vmdiff = 3.

  • strength (only for pmode=0)
    TTempSmooth uses inverse distance weighting when deciding how much weight to give to each pixel value.
    The strength option lets you shift the drop off point away from the center to give a stronger smoothing effect and add weight to the outer pixels.
    It does for the spatial weights what mdiff does for the difference weights.

    • 1 = 0.13 0.14 0.16 0.20 0.25 0.33 0.50 1.00 0.50 0.33 0.25 0.20 0.16 0.14 0.13
    • 2 = 0.14 0.16 0.20 0.25 0.33 0.50 1.00 1.00 1.00 0.50 0.33 0.25 0.20 0.16 0.14
    • 3 = 0.16 0.20 0.25 0.33 0.50 1.00 1.00 1.00 1.00 1.00 0.50 0.33 0.25 0.20 0.16
    • 4 = 0.20 0.25 0.33 0.50 1.00 1.00 1.00 1.00 1.00 1.00 1.00 0.50 0.33 0.25 0.20
    • 5 = 0.25 0.33 0.50 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 0.50 0.33 0.25
    • 6 = 0.33 0.50 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 0.50 0.33
    • 7 = 0.50 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 0.50
    • 8 = 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00

    The values shown are for maxr=7, when using smaller radius values the weights outside of the range are simply dropped. Thus, setting strength to a value of maxr+1 or higher will give you equal spatial weighting of all pixels in the kernel.
    Must be between 1 and 8.
    Default: 2.

  • scthresh
    The standard scenechange threshold as a percentage of maximum possible change of the luma plane.
    A good range of values is between 8 and 15.
    Set to 0 to disable scenechange detection.
    Must be between 0.0 and 100.0.
    Default: 12.0.

  • fp (only for pmode=0)
    When true will add any weight not given to the outer pixels back onto the center pixel when computing the final value and it's much better for reducing artifacts in motion areas and usually produces overall better results. When false will just do a normal weighted average.
    Default: True.

  • y, u, v
    Planes to process.
    1: Return garbage.
    2: Copy plane.
    3: Process plane.
    Default: y = 3, u = 3, v = 3.

  • pfclip
    This allows you to specify a separate clip for TTempSmooth to use when calculating pixel differences.
    This applies to checking the motion thresholds, calculating inverse difference weights, and detecting scenechanges.
    Basically, the pfclip will be used to determine the weights in the average but the weights will be applied to the original input clip's pixel values.

  • opt
    Sets which cpu optimizations to use.
    -1: Auto-detect.
    0: Use C++ code.
    1: Use SSE2 code.
    2: Use AVX2 code.
    3: Use AVX-512 code.
    Default: -1.

  • pmode
    Which mode to use.
    0: Checking the motion thresholds and calculating inverse difference weights.
    1: It tries to select "the best" sample from the total pool of input samples in the temporal axis for current position in frame using some algorithm of selecting "most equal/probable" value. More info here.
    pmode=1 with thresholds = 255 is more or less equal to the MedianBlurTemporal AVS plugin. It is slower with maxr=1 and possibly with maxr=2, and significantly faster with maxr > 2 Also it provides additional functions of possible threshold to limit max difference with source sample and some optional temporal IIR filtering. More info here.
    Default: 0.

  • ythupd, uthupd, vthupd (only for pmode=1)
    IIR memory update threshold respectively for luma and chroma planes.
    If y/u/vthupd > 0 - enable IIR-type filtering. It remembers previous "best" sample output and compares current "best" with memory value. If the difference below thupd - the memory sample used for output. More info here.
    Must be greater than 0.
    Default: 0.

  • ypnew, upnew, vpnew (only for pmode=1)
    Penalty for the new 'best' sample update in the IIR-memory respectively for luma and chroma planes.
    It is added to the internal metric of the current "best" sample in the algorithm of replacing the current stored value with new "best".
    Allow to decrease the number of memory replacement operations and create more stable temporal output. Close to idea of penalty to update current best motion vector with new candidate with a bit better metric in MAnalyse from mvtools. Must be greater than 0.
    Default: 0.

  • threads (only for pmode=1)
    How many logical processors are used.
    0: Maximum logical processors are used.
    Must be between 0 and maximum logical processors.
    Default: 0.

Building:

  • Windows
    Use solution files.

  • Linux

    Requirements:
    - Git
    - C++17 compiler
    - CMake >= 3.16
    - OpenMP

    git clone https://github.com/Asd-g/AviSynth-vsTTempSmooth && \
cd AviSynth-vsTTempSmooth && \
mkdir build && \
cd build && \
cmake .. && \
make -j$(nproc) && \
sudo make install