Video filter TDeint

This is a port of Tritical TDeint deinterlacing filter. The original file can be found at http://www.missouri.edu/~kes25c/. Discussion at http://forum.doom9.org/showthread.php?t=82264. Original filter documents at http://avisynth.org.ru/docs/english/externalfilters/tdeint.htm.

TDeint is a bi-directionally, motion adaptive, sharp deinterlacer. It can adaptively choose between using per-field and per-pixel motion adaptivity, and can use cubic interpolation, kernel interpolation (with temporal direction switching), or one of two forms of modified ELA interpolation which help to reduce “jaggy” edges in moving areas where interpolation must be used.

This port is somehow limited compared to its upstream, but has very interesting results.

• Field Order
  • This setting specifies the field order of the source video. It needs to be set according to your input video to make TDeint work correctly, so be careful here!

• Interpolate
  • This setting specifies which field is to be interpolated. In each frame TDeint keeps one field and (adaptively) interpolates the other one.

• Type
  • Specifies the interpolation mode. Cubic interpolation is the fastest mode, Modified ELA produces smooth pictures and Kernel interpolation produces sharp pictures. ELA interpolation performs best on Anime and Cartoon material, because it produces less “jaggy” edges. Kernel interpolation is recommended for “real life” footage and will cause significantly less flickering. Note: ELA-2 is faster than ELA.

• MntMode
  • This setting controls whether a 4 field motion check or a 5 field motion check is used. 5 fields will prevent more artifacts and can deal with duplicate interlaced frames. However, the 5 field motion check is quite a bit slower than the 4 field motion check. Default is 5 fields.

• Link
  • Controls how the three planes (YUV) are linked during comb map creation. Default is Y to UV.

• AP Type
  • Specifies the mode used for Artefact Protection. Modes 1 and 2 provide a way to catch more artifacts (low AP values) without completely sacrificing static areas. Default is 1.

• Use Chroma to evaluate
  • Includes chroma combing in the decision about whether a frame is combed. Some weird sources will require this setting. For normal sources, keep it disabled.

• Try Weave
  • This setting allows TDeint to try to use per-field motion adaptation instead of per-pixel motion adaptation where possible. If enabled, TDeint will try to weave the current field with the previous filed or the next field. If weaving is successful, then the progressive frame is returned and no further deinterlacing is applied. In case weaving fails (results in a combed frame), the usual per-pixel motion adaptative deinterlacing will be used. This method is useful for processing video with (inconsistent) field shifts. But it's useless for purely interlaced video! Disabled by default.

• Denoise
  • This setting will denoise the motion map in order to avoid inadvertently detecting noise as motion. This will not denoise the actual video, it will only improve the motion detection. Keep this enabled, unless you have a good reason to disable it! Enabled by default.

• Sharp
  • Controls whether the “sharp” or “normal” kernel is used. The sharp kernel includes more pixels and produces a sharper result, but is slightly slower. This has no effect for Cubic interpolation or ELA modes. Enabled by default.

• Evaluate all frames
  • If this option is enabled, then all frames are processed as usual. If not enabled, all frames are first checked to see if they are combed. If a frame isn't combed then it is returned as-is, otherwise it is processed as usual. This allows TDeint to act as a post-processor for IVTC (Inverse Telecine) without the need for hints. Disable this option only if your video contains a mixture of interlaced and progressive frames! Default is enabled.

• Motion Threshold
  • The motion thresholds for luma and chroma. If the difference between two pixels is less than this value they are declared static. Those “static” pixels will not be deinterlaced. Smaller values will reduce residual combing, larger values will decrease flicker and increase the accuracy of field construction in static areas. Default is 6. Setting both values to 0 or below will disable motion adaptation (thumb bobbing).

• Area Combing Threshold
  • Area combing threshold used for combed frame detection. This controls how “strong” or “visible” combing must be to be detected. Good values are from 6 to 12, if you know your source has a lot of combed frames set this towards the low end (6-7), if you know your source has very few combed frames set this higher (10-12). Going much lower than 5 to 6 or much higher than 12 is not recommended. Default is 6.

• Combed Threshold
  • While Area Combing Threshold controls how “visible” or “strong” the combing must be, this setting controls how much combing there must be in any localized area on the frame. Default is 64.

• Artefact Protection Threshold
  • Artefact Protection helps to prevent motion adaptive related artifacts. Before outputting a deinterlaced frame, TDeint will scan all weaved pixels to see if any create a value greater than AP threshold. Any pixels that exceed the threshold will be interpolated. A large AP value (25 or higher) is recommended. Set AP as high as possible without visible artifacts. Smaller values will destroy the benefits of motion adaptivity in static, detailed areas.

• Debug
  • If enabled, this option will cause TDeint to print additional debug information to the output video. This is especially useful to see whether the Try Weave and/or Evaluate all frames options have any effect on your individual source video (successful weave will be indicated, skipped frames have no text at all). Use this for testing only, but disable it for the final processing. Disabled by default.