akarinvs / vapoursynth-dpid

Rapid, Detail-Preserving Image Downscaler for VapourSynth

VapourSynth-dpid

Copyright© 2019 WolframRhodium

Rapid, Detail-Preserving Image Downscaler for VapourSynth

Description

dpid is an algorithm that preserves visually important details in downscaled images and is especially suited for large downscaling factors.

It acts like a convolutional filter where input pixels contribute more to the output image the more their color deviates from their local neighborhood, which preserves visually important details.

Supported Formats

sample type & bps: RGB/YUV/GRAY 8-16 bit integer or floating point.

Usage

dpid.Dpid(clip clip[, int width=0, int height=0, float lambda=1.0, float[] src_left=0, float[] src_top=0, bool read_chromaloc=True])

• clip:

  The input clip.

• width & height: (Default: clip.width / clip.height)

  The width and height of the output clip.

  One of which can be 0, and the downscaling will keep the aspect ratio.

• lambda: (Default: 1.0)

  The power factor of range kernel.

  It can be used to tune the amplification of the weights of pixels that represent detail—from a box filter over an emphasis of distinct pixels towards a selection of only the most distinct pixels.

  This parameter happens to be a python keyword, so you may need to refer to the doc.

• src_left, src_top: (Default: 0)

  Used to select the source region of the input to use.

  They can be arrays to specify different shifting for each plane. The last value is used for the unspecified planes.

  Only the first value in the array is passed to resize.Bilinear(), i.e.

  clip = core.dpid.Dpid(src, width, height, lambda_, src_lefts, src_tops)

  is equivalent to

  down = core.resize.Bilinear(
      src, width, height, 
      src_left=src_lefts[0], src_top=src_tops[0])
  
  clip = core.dpid.DpidRaw(src, down, lambda_, src_lefts, src_tops)

• read_chromaloc: (Default: True)

  Whether to read _ChromaLocation property.

---

dpid.DpidRaw(clip clip[, clip clip2, float lambda=1.0, float[] src_left=0, float[] src_top=0, bool read_chromaloc=True, int[] planes=[0, 1, 2]])

• clip: (Same as dpid.Dpid())

• clip2:

  User-defined downsampled clip.

  Should be of the same format and number of frames as clip.

• lambda: (Default: 1.0)

  (Same as dpid.Dpid())

• src_left, src_top: (Default: 0)

  Used to select the source region of the input to use.

  It can also be used to shift the image.

• read_chromaloc: (Default: True)

  (Same as dpid.Dpid())

• planes: (Default: [0, 1, 2])

  Sets which planes will be processed.

  Any unprocessed planes will be simply copied from clip2.