The BCC Median ﬁlter makes each pixel look like the majority of its neighboring pixels. It produces a smeary painterly look, but with sharp edges (at neighborhood boundaries). The Median ﬁlter also reduces noise by eliminating “spikes,” or pixels that are very different from their neighbors (also referred to as “salt and pepper noise”).
In addition to offering the ability to fine tune the look of the filter, BCC Median allows users to specify how similarity is determined between a pixel and neighboring pixels.
- When Pixel Selected by Channel is chosen, the median is calculated for the chosen channel, then the neighboring pixel(s) that have the median channel value are averaged and replace the current pixel. For example, if the Channel menu is set to Red, then the median Red value is calculated. All neighbor pixels with that same median Red channel value are averaged.
- When Selected Channel Only is chosen, each pixel is altered so that its chosen channel is the median value. The other aspects of the pixel remain unchanged. For example, if the Channel menu is set to Red, the Green and Blue channels are unaffected, but the Red channel is set to the median value.
- When RGBA Separately is chosen, each pixel is replaced by a new pixel. The four (or three, if Preserve Alpha is enabled) channels are the median values of those channels, calculated separately.
- When Most Common Pixel is chosen, each pixel is replaced by the neighboring pixel that is least different than the other neighbors.
The built in PixelChooser, with integrated mocha planar tracking allows for greater control over which areas of the image the ﬁlter should affect.
To learn more about this filter, and see a breakdown of the various controls, please click here.