imagecolortransparent
The comment below is an extremely good workaround
however, php did throw lots of warnings at me when i had error reporting set to E_ALL
this can be avoided with one line of code and no impact (as i can see) to the rest of the function
<?php
//include this file whenever you have to use imageconvolution...
//you can use in your project, but keep the comment below :)
//great for any image manipulation library
//Made by Chao Xu(Mgccl) 2/28/07
//www.webdevlogs.com
//V 1.0
if(!function_exists('imageconvolution')){
function imageconvolution($src, $filter, $filter_div, $offset){
if ($src==NULL) {
return 0;
}
$sx = imagesx($src);
$sy = imagesy($src);
$srcback = ImageCreateTrueColor ($sx, $sy);
ImageCopy($srcback, $src,0,0,0,0,$sx,$sy);
if($srcback==NULL){
return 0;
}
#FIX HERE
#$pxl array was the problem so simply set it with very low values
$pxl = array(1,1);
#this little fix worked for me as the undefined array threw out errors
for ($y=0; $y<$sy; ++$y){
for($x=0; $x<$sx; ++$x){
$new_r = $new_g = $new_b = 0;
$alpha = imagecolorat($srcback, $pxl[0], $pxl[1]);
$new_a = $alpha >> 24;
for ($j=0; $j<3; ++$j) {
$yv = min(max($y - 1 + $j, 0), $sy - 1);
for ($i=0; $i<3; ++$i) {
$pxl = array(min(max($x - 1 + $i, 0), $sx - 1), $yv);
$rgb = imagecolorat($srcback, $pxl[0], $pxl[1]);
$new_r += (($rgb >> 16) & 0xFF) * $filter[$j][$i];
$new_g += (($rgb >> 8) & 0xFF) * $filter[$j][$i];
$new_b += ($rgb & 0xFF) * $filter[$j][$i];
}
}
$new_r = ($new_r/$filter_div)+$offset;
$new_g = ($new_g/$filter_div)+$offset;
$new_b = ($new_b/$filter_div)+$offset;
$new_r = ($new_r > 255)? 255 : (($new_r < 0)? 0:$new_r);
$new_g = ($new_g > 255)? 255 : (($new_g < 0)? 0:$new_g);
$new_b = ($new_b > 255)? 255 : (($new_b < 0)? 0:$new_b);
$new_pxl = ImageColorAllocateAlpha($src, (int)$new_r, (int)$new_g, (int)$new_b, $new_a);
if ($new_pxl == -1) {
$new_pxl = ImageColorClosestAlpha($src, (int)$new_r, (int)$new_g, (int)$new_b, $new_a);
}
if (($y >= 0) && ($y < $sy)) {
imagesetpixel($src, $x, $y, $new_pxl);
}
}
}
imagedestroy($srcback);
return 1;
}
}
?>
imageconvolution
(PHP 5 >= 5.1.0)
imageconvolution — Apply a 3x3 convolution matrix, using coefficient and offset
Description
bool imageconvolution
( resource $image
, array $matrix
, float $div
, float $offset
)
Applies a convolution matrix on the image, using the given coefficient and offset.
Parameters
- image
-
An image resource, returned by one of the image creation functions, such as imagecreatetruecolor().
- matrix
-
A 3x3 matrix: an array of three arrays of three floats.
- div
-
The divisor of the result of the convolution, used for normalization.
- offset
-
Color offset.
Return Values
Returns TRUE on success or FALSE on failure.
Examples
Example #1 Embossing the PHP.net logo
<?php
$image = imagecreatefromgif('http://www.php.net/images/php.gif');
$emboss = array(array(2, 0, 0), array(0, -1, 0), array(0, 0, -1));
imageconvolution($image, $emboss, 1, 127);
header('Content-Type: image/png');
imagepng($image, null, 9);
?>
The above example will output:
Example #2 Gaussian blur
<?php
$image = imagecreatetruecolor(180,40);
// Writes the text and apply a gaussian blur on the image
imagestring($image, 5, 10, 8, 'Gaussian Blur Text', 0x00ff00);
$gaussian = array(array(1.0, 2.0, 1.0), array(2.0, 4.0, 2.0), array(1.0, 2.0, 1.0));
imageconvolution($image, $gaussian, 16, 0);
// Rewrites the text for comparison
imagestring($image, 5, 10, 18, 'Gaussian Blur Text', 0x00ff00);
header('Content-Type: image/png');
imagepng($image, null, 9);
?>
The above example will output:
Notes
Note: This function is only available if PHP is compiled with the bundled version of the GD library.
add a note
User Contributed Notesimageconvolution
Jase
14-Sep-2007 06:53
14-Sep-2007 06:53
mgcclx at gmail dot com
01-Mar-2007 11:55
01-Mar-2007 11:55
imageconvolution() does not appear in PHP with non-bundled GD libraries. It is a rare situation, but it still happens. That's why I wrote a replication of imageconvolution() in PHP. Compare to the post below, this one makes the use of offset and 30% faster.
Because it's written in PHP, it is 50 times slower than the bundled version.
Actually, this is a replication of gdimageconvolutaion() of GD library, it does not support data validating feature imageconvolution() have. But I guess people who uses this function knows their stuff.
THE SCRIPT:
<?php
//include this file whenever you have to use imageconvolution...
//you can use in your project, but keep the comment below :)
//great for any image manipulation library
//Made by Chao Xu(Mgccl) 2/28/07
//www.webdevlogs.com
//V 1.0
if(!function_exists('imageconvolution')){
function imageconvolution($src, $filter, $filter_div, $offset){
if ($src==NULL) {
return 0;
}
$sx = imagesx($src);
$sy = imagesy($src);
$srcback = ImageCreateTrueColor ($sx, $sy);
ImageCopy($srcback, $src,0,0,0,0,$sx,$sy);
if($srcback==NULL){
return 0;
}
for ($y=0; $y<$sy; ++$y){
for($x=0; $x<$sx; ++$x){
$new_r = $new_g = $new_b = 0;
$alpha = imagecolorat($srcback, $pxl[0], $pxl[1]);
$new_a = $alpha >> 24;
for ($j=0; $j<3; ++$j) {
$yv = min(max($y - 1 + $j, 0), $sy - 1);
for ($i=0; $i<3; ++$i) {
$pxl = array(min(max($x - 1 + $i, 0), $sx - 1), $yv);
$rgb = imagecolorat($srcback, $pxl[0], $pxl[1]);
$new_r += (($rgb >> 16) & 0xFF) * $filter[$j][$i];
$new_g += (($rgb >> 8) & 0xFF) * $filter[$j][$i];
$new_b += ($rgb & 0xFF) * $filter[$j][$i];
}
}
$new_r = ($new_r/$filter_div)+$offset;
$new_g = ($new_g/$filter_div)+$offset;
$new_b = ($new_b/$filter_div)+$offset;
$new_r = ($new_r > 255)? 255 : (($new_r < 0)? 0:$new_r);
$new_g = ($new_g > 255)? 255 : (($new_g < 0)? 0:$new_g);
$new_b = ($new_b > 255)? 255 : (($new_b < 0)? 0:$new_b);
$new_pxl = ImageColorAllocateAlpha($src, (int)$new_r, (int)$new_g, (int)$new_b, $new_a);
if ($new_pxl == -1) {
$new_pxl = ImageColorClosestAlpha($src, (int)$new_r, (int)$new_g, (int)$new_b, $new_a);
}
if (($y >= 0) && ($y < $sy)) {
imagesetpixel($src, $x, $y, $new_pxl);
}
}
}
imagedestroy($srcback);
return 1;
}
}
?>
interghost at crovortex dot com
01-Apr-2006 03:26
01-Apr-2006 03:26
an implementation of this function for PHP versions <5.1
<?php
if(!function_exists("imageconvolution"))
{
function imageconvolution(&$img,$mat,$div,$off)
{
if(!imageistruecolor($img) || !is_array($mat) || count($mat)!=3 || count($mat[0])!=3 || count($mat[1])!=3 || count($mat[2])!=3) return FALSE;
unset($bojainfo);
for($nx=0;$nx<imagesx($img)-1;$nx++)
{
for($ny=0;$ny<imagesy($img)-1;$ny++)
{
$rgb=imagecolorat($img,$nx,$ny);
$bojainfo[$nx][$ny][r]=($rgb>>16)&0xFF;
$bojainfo[$nx][$ny][g]=($rgb>>8)&0xFF;
$bojainfo[$nx][$ny][b]=$rgb&0xFF;
}
}
for($nx=1;$nx<imagesx($img)-1;$nx++)
{
for($ny=1;$ny<imagesy($img)-1;$ny++)
{
$nr=$mat[0][0]*$bojainfo[$nx-1][$ny-1][r] + $mat[0][1]*$bojainfo[$nx][$ny-1][r] + $mat[0][2]*$bojainfo[$nx+1][$ny-1][r] + $mat[1][0]*$bojainfo[$nx-1][$ny][r] + $mat[1][1]*$bojainfo[$nx][$ny][r] + $mat[1][2]*$bojainfo[$nx+1][$ny][r] + $mat[2][0]*$bojainfo[$nx-1][$ny+1][r] + $mat[2][1]*$bojainfo[$nx][$ny+1][r] + $mat[2][2]*$bojainfo[$nx+1][$ny+1][r];
$nr=intval(round($nr/$div));
if($nr<0) $nr=0;
elseif($nr>255) $nr=255;
$ng=$mat[0][0]*$bojainfo[$nx-1][$ny-1][g] + $mat[0][1]*$bojainfo[$nx][$ny-1][g] + $mat[0][2]*$bojainfo[$nx+1][$ny-1][g] + $mat[1][0]*$bojainfo[$nx-1][$ny][g] + $mat[1][1]*$bojainfo[$nx][$ny][g] + $mat[1][2]*$bojainfo[$nx+1][$ny][g] + $mat[2][0]*$bojainfo[$nx-1][$ny+1][g] + $mat[2][1]*$bojainfo[$nx][$ny+1][g] + $mat[2][2]*$bojainfo[$nx+1][$ny+1][g];
$ng=intval(round($ng/$div));
if($ng<0) $ng=0;
elseif($ng>255) $ng=255;
$nb=$mat[0][0]*$bojainfo[$nx-1][$ny-1][b] + $mat[0][1]*$bojainfo[$nx][$ny-1][b] + $mat[0][2]*$bojainfo[$nx+1][$ny-1][b] + $mat[1][0]*$bojainfo[$nx-1][$ny][b] + $mat[1][1]*$bojainfo[$nx][$ny][b] + $mat[1][2]*$bojainfo[$nx+1][$ny][b] + $mat[2][0]*$bojainfo[$nx-1][$ny+1][b] + $mat[2][1]*$bojainfo[$nx][$ny+1][b] + $mat[2][2]*$bojainfo[$nx+1][$ny+1][b];
$nb=intval(round($nb/$div));
if($nb<0) $nb=0;
elseif($nb>255) $nb=255;
$nrgb=($nr<<16)+($ng<<8)+$nb;
if(!imagesetpixel($img,$nx,$ny,$nrgb)) return FALSE;
}
}
return TRUE;
}
}
?>
it's a bit slowish so I wouldn't recommend big images, also offset is not implemented (don't know what it's suppose to do)
mlconnor at yahoo dot com
22-Mar-2006 01:00
22-Mar-2006 01:00
I've seen many people come up with ways to do a drop shadow behind a rectangle such as a picture. I haven't found one yet that was fast, PHP 4 complaint, and nice looking. Here is one I came up with last night. It takes an image, fills it with the background, and creates a blurred drop shadow at the specified coords using the colors and the distance offset specified. It looks great!!!
function blurRect(&$image, $distance, $rectX1, $rectY1, $rectX2, $rectY2, $shadowR, $shadowG, $shadowB, $backR, $backG, $backB) {
$potentialOverlap = ($distance * 2) * ($distance * 2);
$backgroundColor = imagecolorallocate($image, $backR, $backG, $backB);
$shadowColor = imagecolorallocate($image, $shadowR, $shadowG, $shadowB);
$imageWidth = imagesx($image);
$imageHeight = imagesy($image);
imageFilledRectangle($image, 0, 0, $imageWidth - 1, $imageHeight - 1, $backgroundColor);
imageFilledRectangle($image, $rectX1, $rectY1, $rectX2, $rectY2, $shadowColor);
for ( $pointX = $rectX1 - $distance; $pointX < $imageWidth; $pointX++ ) {
for ( $pointY = $rectY1 - $distance; $pointY < $imageHeight; $pointY++ ) {
if ( $pointX > $rectX1 + $distance &&
$pointX < $rectX2 - $distance &&
$pointY > $rectY1 + $distance &&
$pointY < $rectY2 - $distance ) {
$pointY = $rectY2 - $distance;
}
$boxX1 = $pointX - $distance;
$boxY1 = $pointY - $distance;
$boxX2 = $pointX + $distance;
$boxY2 = $pointY + $distance;
$xOverlap = max(0, min($boxX2, $rectX2) - max($boxX1, $rectX1));
$yOverlap = max(0, min($boxY2, $rectY2) - max($boxY1, $rectY1));
$totalOverlap = $xOverlap * $yOverlap;
$shadowPcnt = $totalOverlap / $potentialOverlap;
$backPcnt = 1.0 - $shadowPcnt;
$newR = $shadowR * $shadowPcnt + $backR * $backPcnt;
$newG = $shadowG * $shadowPcnt + $backG * $backPcnt;
$newB = $shadowB * $shadowPcnt + $backB * $backPcnt;
$newcol = imagecolorallocate($image, $newR, $newG, $newB);
imagesetpixel($image, $pointX, $pointY, $newcol);
}
}
}
jonathan dot aquino at gmail dot com
18-Jan-2006 01:30
18-Jan-2006 01:30
For those who don't have PHP 5.1 and thus don't have imageconvolution, I found some code to do blurring: http://www.hudzilla.org/php/11_2_25.php Works quite nicely.
dyer85 at gmail dot com
07-Jan-2006 06:24
07-Jan-2006 06:24
Took me a while, but thanks to a couple of the user notes on the array_values PHP documentation page, I was able to come up with a way to dynamically compute the divisor.
I'm using PHP 5.1.0b2 on Win32 with the bundled GD library. When I try and use the imageconvolution function, whether normally, or via the functions below, the resulting image (I've only tried JPEGs and GIFs), always comes out far too bright, even when the divisor makes matrix sum equal to 1. The only thing that would reduce the brightness was to make the offset argument ridiculously large. So, I'm not sure if this effects anyone else.
Here are the functions with an example:
<?php
$im = imagecreatefromjpeg('path/to/pic.jpg');
$matrix = array( array(5,5,5),
array(5,15,5),
array(5,5,5) );
makeFilter($im, $matrix);
header ( 'Content-Type: image/jpeg' );
imagejpeg($im);
imagedestroy($im);
/**
* functions
*/
// This flattens the 3X3 array matrix, so we can get the sum of all the values
function array_flatten($array) {
(array)$tempArray = array();
foreach ( $array as $value ) {
if ( is_array($value) ) {
$tempArray = array_merge($tempArray, array_flatten($value));
} else {
$tempArray[] = $value;
}
}
return $tempArray;
}
// Creates the divisor value dynamically, and passes offset
function makeFilter($resource, $matrix, $offset=1.0) {
global $$resource;
(float)$divisor = array_sum(array_flatten($matrix));
return imageconvolution($resource, $matrix, $divisor, $offset) ? true : false;
}
?>
27-Oct-2005 08:32
The example below didn't provide a 3x3 matrix. Correctly, it's a multidimensional array.
<?php
$matrix = array( array( -1, -1, -1 ),
array( -1, 16, -1 ),
array( -1, -1, -1 ) );
?>
timeshifting at gmail dot com
25-Aug-2005 06:40
25-Aug-2005 06:40
Matrices can be used for sharpening, blurring, edge detection, etc, ala Photoshop.
A sharpening example:
<?php
$sharpenMatrix = array(-1,-1,-1,-1,16,-1,-1,-1,-1);
$divisor = 8;
$offset = 0;
imageconvolution($myImage, $sharpenMatrix, $divisor, $offset);
?>
Below is some information on building different kinds of matrices. (If you have photoshop (or PSP, GIMP) you can test out your matrices before applying them in PHP)
http://loriweb.pair.com/8udf-basics.html (covers blurs)
http://loriweb.pair.com/8udf-sharpen.html
http://loriweb.pair.com/8udf-edges.html
http://loriweb.pair.com/8udf-emboss.html