v12

• Add parameter "combed_only". If it's True, only frames where the property "_Combed" is 1 will be processed. The default value is False.

nnedi3_weights.bin

v11

• Allow the use of the int16 prescreeners with 9..16 bit input. This means the faster "new" prescreener (pscrn=2..4) can be used and so 9..16 bit input is processed faster with default parameters.
• Allow the use of the int16 predictor with 9..15 bit input. This means 9..15 bit input is processed faster (theoretically) with default parameters.
• Fix memory leak when nnedi3_weights.bin is missing or has the wrong size (bug introduced in v3).
• readme.rst finally explains what all the parameters are.

nnedi3_weights.bin

v10

• Fix prescreener with 16 bit or float input (bug introduced in v9).
• Add parameter "show_mask", for debugging purposes.

nnedi3_weights.bin

v9

• Float pixels are no longer clamped to the usual ranges of [0.0, 1.0]/[-0.5, 0.5].
• Fix a crash in OS X and possibly 64 bit Linux.
• Fix a bug in the NEON intrinsics.
• Always set the _FieldBased frame property to progressive.
• Ignore the _FieldBased frame property when "dh" is True and use the _Field property instead.
• Remove the deprecated parameters "Y", "U", and "V".
• Replace the bitfield parameter "fapprox" with the parameters "int16_prescreener", "int16_predictor", and "exp".

nnedi3_weights.bin

v8

• Fix crash with more than 8 but less than 16 bits per sample (bug introduced in v6).

nnedi3_weights.bin

v7

• Fix corrupt output with more than 8 bits per sample (bug introduced in v6).

nnedi3_weights.bin

v6

• Normalise the frame rate when doubling it.
• Add support for 32 bit floating point images.
• Deprecate the Y, U, and V parameters in favour of the standard planes parameter.
• Only allocate temporary buffers for the planes that are actually processed.
• Remove the nnedi3_rpow2 filter.

And a change that affects only (some) users of ARM CPUs:

• Add NEON intrinsics (translated from the x86 ASM). They make it go fast(er).

nnedi3_weights.bin

v5

• Adjust the frame durations when doubling the frame rate.
• Fix buffer overflow with images wider than 8192 pixels or so (inherited from the Avisynth plugin).
• Hopefully prevent crashes with images that require more than 2 GiB per plane.
• Refuse to create clips longer than INT_MAX (often 2**31-1).
• Use the _FieldBased frame property to determine each frame's field order, for sources where it changes.

nnedi3_weights.bin

v4

• Fix copy-paste error in the FMA functions
• Rename nnedi3 weights.bin to nnedi3_weights.bin

nnedi3_weights.bin

v3

• Use FMA (fused multiply-add) instructions in some functions. Speeds up the float paths a bit.
• Add another SIMD function for a little more speed with 16 bit input.
• Automatically select the best functions if opt=True, use only C functions if opt=False.
• Don't embed the weights into the DLLs. nnedi3 weights.bin needs to be in the same folder as the DLL.