gd_filter.c

int gdImageScatter( gdImagePtr  im, int  sub, int  plus )

int gdImageScatterColor( gdImagePtr  im, int  sub, int  plus, int  colors[], unsigned  int  num_colors )

int gdImageScatterEx( gdImagePtr  im, gdScatterPtr  scatter )

int gdImagePixelate( gdImagePtr  im, int  block_size, const unsigned  int  mode )

int gdImageNegate( gdImagePtr  src )

Invert src image.

int gdImageGrayScale( gdImagePtr  src )

Convert the image src to a grayscale image.

int gdImageBrightness( gdImagePtr  src, int  brightness )

Set the brightness level <level> for the image src.

int gdImageContrast( gdImagePtr  src, double  contrast )

int gdImageColor( gdImagePtr  src, const  int  red, const  int  green, const  int  blue, const  int  alpha )

int gdImageConvolution( gdImagePtr  src, float  filter[3][3], float  filter_div, float  offset )

int gdImageSelectiveBlur( gdImagePtr  src )

int gdImageEdgeDetectQuick( gdImagePtr  src )

int gdImageGaussianBlur( gdImagePtr  im )

gdImageGaussianBlur performs a Gaussian blur of radius 1 on the image.  The image is modified in place.

NOTE: You will almost certain want to use gdImageCopyGaussianBlurred instead, as it allows you to change your kernel size and sigma value.  Future versions of this function may fall back to calling it instead of gdImageConvolution, causing subtle changes so be warned.

Parameters

Returns

GD_TRUE (1) on success, GD_FALSE (0) on failure.

int gdImageEmboss( gdImagePtr  im )

int gdImageMeanRemoval( gdImagePtr  im )

int gdImageSmooth( gdImagePtr  im, float  weight )

Return a copy of the source image src blurred according to the parameters using the Gaussian Blur algorithm.

radius is a radius, not a diameter so a radius of 2 (for example) will blur across a region 5 pixels across (2 to the center, 1 for the center itself and another 2 to the other edge).

sigma represents the “fatness” of the curve (lower == fatter).  If sigma is less than or equal to 0, gdImageCopyGaussianBlurred ignores it and instead computes an “optimal” value.  Be warned that future versions of this function may compute sigma differently.

The resulting image is always truecolor.

More Details

A Gaussian Blur is generated by replacing each pixel’s color values with the average of the surrounding pixels’ colors.  This region is a circle whose radius is given by argument radius.  Thus, a larger radius will yield a blurrier image.

This average is not a simple mean of the values.  Instead, values are weighted using the Gaussian function (roughly a bell curve centered around the destination pixel) giving it much more influence on the result than its neighbours.  Thus, a fatter curve will give the center pixel more weight and make the image less blurry; lower sigma values will yield flatter curves.

Currently, gdImageCopyGaussianBlurred computes the default sigma as

(2/3)*radius

Note, however that we reserve the right to change this if we find a better ratio.  If you absolutely need the current sigma value, you should set it yourself.

Parameters

Returns

The new image or NULL if an error occurred.  The result is always truecolor.

Example

FILE *in;
gdImagePtr result, src;

in = fopen("foo.png", "rb");
src = gdImageCreateFromPng(in);

result = gdImageCopyGaussianBlurred(im, src->sx / 10, -1.0);