| 10.9. Predator | ||
|---|---|---|
|
10. 艺术滤镜 |  |
| 10.9. Predator | ||
|---|---|---|
|
|
10. 艺术滤镜 | |
This filter adds a “Predator” effect to the image. The predator effect makes the image/selection look something like the view the predator has in movies (kind of a thermogram and that type of thing). This will reduce the image to edges in a few basic colors on a dark background.
If there is an active selection, the filter effect will be applied to the selected region, otherwise to the alpha channel (the filter will add an alpha channel, if necessary). The filter works best on colorful RGB images.
The “predator” filter will detect edges using the Sobel edge detector. The specified “Edge amount” will be passed to the Sobel filter. A high value will result in detecting more edges.
If checked, the filter will simplify the image into solid-colored squares using the Pixelise filter before the real predator effect will be applied. You can select the size of these squares with the option Pixel amount, which will heavily affect the result (see examples below).
“Pixel amount” is the size of the color blocks the image will be simplified to if Pixelize is checked. Actually you are decreasing the resolution with this option. In the examples below, you can see directly how increasing the pixel block size leads to something like “macro pixels”:
If checked, the filter will be applied to the active selection. Else, it will be applied to the active layer.
When this option is checked, a copy of the active layer will be created above the active layer and the filter will be applied to this copy, leaving the original layer untouched. If not checked, the filter will be applied to the active layer.
Since this filter delegates the essential parts to two or three other filters the algorithm is very simple:
The original image.
Optionally, the filter pixelizes the image: it renders the image by using color blocks instead of pixels, thus reducing the image resolution.
The colors will be reduced to pure red, green, blue (and possibly gray colors), using the minimal RGB channel for every pixel.
Applying the Sobel edge detecting filter, the image will be reduced further on to edges, usually on a black background, with very few colors.
