Software » Useful Tools » AviSynth+ 3.7.6

Changes from 3.7.5 to 3.7.6
Additions, changes
Added new pixel format constants

YUV(A) 4:4:0(:) 8~32bits
YV411 aliased to more standard YUV411 naming
YUV 4:1:1 10~32bits
YUVA 4:1:1: 8~32bits
Y plus alpha 4:0:0: 8~32bits
YUV9 (added in 2.6 but not used except in the constants list) aliased to more standard YUV410 naming
YUV 4:1:0 10~32bits
YUVA 4:1:0: 8~32bits
VfW: YUV 4:1:0 is served as YVU9 (Y, V, U) and YUV 4:4:0 as I440 (Y, U, V); 4:1:1, 4:1:0 and 4:4:0 formats over 8 bits (with or without alpha) are served as their 8-bit YV411/YVU9/I440 counterparts.

VfW: 8-bit Y plus alpha (YA8) clips are now served with the Y2[0][8] FourCC (a packed Y,A interleaved layout; matches ffmpeg's AV_PIX_FMT_YA8).

Added ConvertToYA, besides ConvertToY, mirroring the existing ConvertToYUVxxx/ConvertToYUVAxxx pairs.

New array syntax:

dictionary-style ArraySet (replaces or appends).
dictionary-style ArrayDel (deletes by key, no-op if not found).
dictionary-style ArrayIndexOf (returns the actual index of the dictionary-style array search).
See :doc:`Arrays <./script_ref/script_ref_arrays>`.

Added SetFilterProp, GetFilterProps, and SetFilterPropPassthrough:

SetFilterProp — automatically inject a frame property on the output of a named filter every time it is instantiated, without modifying the script. Rules are typically placed in an auto-loaded .avsi file. The property value may be a static int/float/string, a function object evaluated per frame (e.g. to derive _Matrix from output resolution), or captured from the filter's own call argument (undefined() capture mode).
SetFilterProp conditional form — inject a property only when a specific named call argument equals a given match value or any element of an alias array. String comparison is case-insensitive. Both positional and named call sites are matched.
SetFilterPropPassthrough — compatibility shim for old filters that predate frame-property support (e.g. ColorMatrix). Copies all input frame properties to the filter's output; self-healing: becomes a no-op per frame once the plugin is updated to use NewVideoFrameP.
GetFilterProps — returns the registered rule table as a JSON string for diagnostics (valid JSON: escaped strings, NaN and infinity as null).
Addresses issue #393 and more. See :doc:`SetFilterProp <./corefilters/setfilterprop>`.

Bump year to 2026

Overlay: "add" and "subtract" direct RGB mode

Overlay: "add" and "subtract" supports 32-bit float.

Added utf8 parameter to AddAutoLoadDir

Added utf8 parameter to ListAutoLoadDirs

Added utf8 parameter to LoadPlugin

Added utf8 parameter to DumpFilterGraph

"Info": Optimize AVX512 features display, group features, make a bit more compact

"Info": add L2 cache size display

"SetMaxCPU": add "avx512base" and "avx512fast" options to enable/disable AVX512 grouped features. see :ref:`SetMaxCPU <setmaxcpu>` . Users of base-only AVX512 CPUs can enable Avisynth-optimizations with SetMaxCPU("avx512base+").

ARM64 (aarch64) area:

"Info": add ARMV8-A features display (NEON, DOTPROD, SVE2)
Add ArmV8-A cpu feature detection (NEON, DOTPROD, SVE2) on ARM64 Windows/Linux/macOS builds. On Windows, only up-to DOTPROD can be detected due to OS limitations.
New CPU flags in cpuid.h and avisynth_c.h: CPUF_ARM_NEON, CPUF_ARM_DOTPROD, CPUF_ARM_SVE2
"SetMaxCPU": add "neon", "dotprod", "sve2", "i8mm", "sve2.1" options to enable/disable ARM64 (aarch64) features.
"ConvertToPlanarRGB(A)": bits parameter: on-the-fly bit-depth conversions to YUV->RGB conversion. See :doc:`ConvertToPlanarRGB <./corefilters/convert>`.

ConvertToPlanarRGB(A): added quality parameter: forces 32-bit float internal processing instead of S18.13 fixed-point arithmetic when converting from YUV, regardless of source or target bit-depth. See :doc:`ConvertToPlanarRGB <./corefilters/convert>`.

"ConvertToYUV(A)xxx" and legacy 8 bit name versions: bits parameter: on-the-fly bit-depth conversions to RGB->YUV conversion. See :doc:`ConvertToYUV444 <./corefilters/convert>`.

ConvertToYUV(A)xxx and legacy 8 bit name versions: added quality parameter: forces 32-bit float internal processing instead of S18.13 fixed-point arithmetic when converting to YUV, regardless of source or target bit-depth. See :doc:`ConvertToYUV444 <./corefilters/convert>`.

"ResetMask": add parameter float "opacity"

"AddAlphaPlane": add parameter float "opacity"

"Layer": YUY2 is handled as YV16 (lessen source code bloat)

"Histogram" Color and Color2 mode additions and fixes:

Added matrix parameter (Color and Color2): specifies the YUV matrix for chroma interpretation (BT.601, BT.709, BT.2020, etc.), following the same convention as the ConvertToYUV family. If not set, the matrix is read from the clip's _Matrix and _ColorRange frame properties.
Added graticule string parameter (Color and Color2): controls the danger zone shading (Color) or valid chroma boundary square (Color2). "on" (default) always draws it, preserving pre-3.7.6 behavior; "off" never draws it; "auto" draws it only for limited-range clips and suppresses it for full-range clips.
Added circle parameter (Color and Color2, default false for Color, default true for Color2): draws the hue circle with 15° tick marks. In Color2 mode this was previously always drawn; it can now be disabled.
Added targets parameter (Color and Color2, default false): draws target boxes at the six 75%-amplitude ColorBars Cb/Cr positions (Yellow, Cyan, Green, Magenta, Red, Blue). Positions are computed from ground-truth linear RGB values through the active matrix, giving accurate coordinates at all bit depths.
Added targets100 parameter (Color and Color2, default false): draws target boxes at the same six ColorBars Cb/Cr positions as targets, but placed at 100% luma level instead of 75%.
Added axes parameter (Color and Color2, default false): draws horizontal and vertical crosshair lines through the vectorscope center.
Added iq parameter (Color and Color2, default false): draws target boxes for the NTSC I, +I and +Q chroma-phase references, using the luma-corrected broadcast convention (Y raised until the most negative RGB component reaches Code 0, avoiding illegal RGB values).
Added iq_lines parameter (Color and Color2, default false): draws radial lines at the fixed NTSC subcarrier phase angles of 33°, 123°, 213° and 303°, marking the I and Q chroma axes.
Fix: "Color" and "Color2" modes: copy alpha channel from source for alpha-carrying formats (YUVA, RGBPA, RGB32, RGB64); initialize alpha to zero in the histogram panel area.
Fix: accurate pixel positioning and scaling throughout the histogram panel, limited/full range aware. Marker positions use the same conversion path as the vectorscope signal dots, ensuring exact alignment at any bit depth and color range.
Vectorscope is now fully matrix-aware: all overlay marker positions are computed from ground-truth linear RGB values converted through the active YUV matrix, giving accurate Cb/Cr coordinates at all bit depths including 32-bit float.
ConvertToYUV4xx: ChromaInPlacement/ChromaOutPlacement are defined per axis (co-sited, centered or bottom, horizontally and vertically), and every placement name is accepted for every subsampled format; on a non-subsampled axis the positions coincide (e.g. 4:4:0 "left" equals "center", 4:2:2/4:1:1 "top" equals "center", "DV" equals "left" there). 4:2:2 and 4:1:1 no longer reject "top", "bottom" and "DV". See :doc:`Convert <./corefilters/convert>`.

Resizers placement: same per-axis chroma siting as ConvertToYUV4xx (shared code).

"CombinePlanes": YA support: planes="YA" from greyscale clips gives a YA clip, and the default source planes for a YA target are "YA". Target planes not listed in planes keep the first clip's plane at the same position (Y/G, U/B, V/R, A), or get a neutral value when it has none (U/V center, alpha opaque, Y/R/G/B black by _ColorRange); they were undefined before.

"UToY8", "VToY8", "UToY", "VToY", "ExtractU", "ExtractV", "PlaneToY", "YToUV": YA support: U/V extraction from a YA clip reports that it has no such plane; "YToUV" accepts YA clips (their Y plane is used, as for Y clips).

"ConvertToYUY2": rewritten to route all conversions through YV16 as intermediate format, using the full ConvertToPlanarGeneric infrastructure. ChromaOutPlacement parameter added (was missing, present in ConvertToYV16). All source formats (YV12, YUV420/422/444, planar/packed RGB, high bit depth) are now handled correctly. bits parameter accepted; must be 8 if specified. See :doc:`Convert <./corefilters/convert>`.

"ConvertBackToYUY2": kept for backward compatibility; now forwards to ConvertToYUY2. The pre-2.5 left-pixel-only chroma hack is no longer needed or applied; the YV16 lossless repack path avoids chroma resampling loss entirely for roundtrip workflows.

"ConvertToYV12" (legacy 8-bit name): the YUY2 fast-path shortcut is removed; now routes directly through ConvertToPlanarGeneric::CreateYUV420. bits= parameter semantics corrected: the legacy 8-bit-named functions (ConvertToYV12, ConvertToYV16, ConvertToYV24) now check the target bit depth rather than the source bit depth, allowing high-depth sources when bits=8 is explicitly specified.

8 bit packed RGB formats are converted to planar RGB before 444 conversion. Stop using direct rgb-yv24 conversions (16 bits were already converted for long time).

Add :doc:`ColorBarsUHD <./corefilters/colorbarsuhd>` ITU-R BT.2111-3 (05/2025) with three signal variants: HLG narrow range (mode=0), PQ narrow range (mode=1) and PQ full range (mode=2)

"Layer": add "mulovr" op — Overlay-style multiply for YUV(A) formats only (RGB raises an error). The overlay luma (Y) plane drives darkening of all base planes: dark overlay Y pulls base luma toward black and simultaneously desaturates base chroma toward neutral (128 for 8-bit integer, 0.0 for float). Bright overlay Y (max value) leaves the base unchanged. Matches ...

Changes compared to AviSynth

64-bit version available
Faster script startup and substantially faster core filters
Support for multiple ("shadow") plugin directories
Autoloading of C-plugins
Script language extensions, with support for multiline conditionals and loops.
Improved still image support and TimeStretch

Developer features

Easy and fast compilation, with current compilers
2x increased compilation speed
Improved debugging experience
Leaner and more logical project structure
Large parts of the code refactored