entranskerneldeint

[based on Avisynth filter documentation]

This filter deinterlaces using a kernel approach. It gives greatly improved vertical resolution in deinterlaced areas compared to simple field discarding.

If you set the threshold property to 0, you can get totally artifact free results (due to lack of thresholding) but with much less loss of vertical resolution than simple field discarding. For optimal results, however, set a motion threshold that allows static areas of the picture to be passed through. In this mode, the kernel-based deinterlacing of the moving areas preserves their vertical resolution compared to simple interpolation.

This filter assumes that top field is first. If not, use a filter (e.g. entransfields) to swap them. Note that whether the top field is indeed first can be verified by separating fields (using e.g. fields) and verifying that motion is progressive (and not jumping back and forth).

The threshold parameter defines the "motion" thresold. Moving areas are kernel-deinterlaced while non-moving areas are passed through. Use the entranskerneldeint:map parameter to tweak the threshold parameter so that just the combed areas of the frame are deinterlaced.

sharp, when set to true, selects a kernel that provides better vertical resolution and performs some sharpening of the video. For less sharpening but also less vertical resolution, set this parameter to false.

two-way, when set to true, selects a kernel that includes both the previous and the following fields for deinterlacing. When set to false, the kernel includes only the previous field. The latter one-way kernel is faster, crisper, and gives less blending (this last advantage makes the filter perform better on anime).

map, when set to true, shows the areas that are "moving" as determined by the threshold parameter and which will be kernel-deinterlaced.

History

• Avisynth kerndeint filter [Donald A. Graft]
• Also available in mplayer (kerndeint filter)
• Also available in avidemux (kernel deinterlacer)