Open bobosch opened 8 years ago
I've wrote a function to create the image thumbnails from a .jpg file
<?php
var_dump(create_face_thumbnails('data/media/family.jpg'));
function create_face_thumbnails($file) {
$xmp_data=get_xmp_tag_from_file($file);
if($xmp_data) {
$faces=get_face_data_from_xmp_tag($xmp_data);
if($faces) {
$image=imagecreatefromjpeg($file);
$info=array(
'x'=>imagesx($image),
'y'=>imagesy($image),
);
$path_parts=pathinfo($file);
$base=$path_parts['dirname'].'/'.$path_parts['filename'].' ';
foreach($faces as $name=>$face) {
$rect=get_face_coordinates($face,$info);
if($rect) {
$crop=imagecrop($image,$rect);
$exif=exif_read_data($file);
$crop=rotate_image_by_exif($crop,$exif);
$thumb=imagescale($crop,100);
imagejpeg($thumb,$base.$name.'.jpg');
$rects[$name]=$rect;
}
}
return $rects;
}
}
}
function get_xmp_tag_from_file($file) {
$content=file_get_contents($file);
$xmp_data_start=strpos($content, '<x:xmpmeta');
$xmp_data_end=strpos($content, '</x:xmpmeta>');
$xmp_length=$xmp_data_end - $xmp_data_start;
$xmp_data=substr($content, $xmp_data_start, $xmp_length + 12);
return $xmp_data ;
}
function get_face_data_from_xmp_tag($xmp_data) {
$faces=array();
$name='';
$xml_parser=xml_parser_create('UTF-8');
xml_parser_set_option($xml_parser,XML_OPTION_SKIP_WHITE,0);
xml_parser_set_option($xml_parser,XML_OPTION_CASE_FOLDING,0);
xml_parse_into_struct($xml_parser, $xmp_data, $vals, $index);
xml_parser_free($xml_parser);
foreach($vals as $val) {
switch($val['tag']){
case 'rdf:Description':
if($val['type']=='open' && !empty($val['attributes']['mwg-rs:Name']) && $val['attributes']['mwg-rs:Type']=='Face') {
$name=trim($val['attributes']['mwg-rs:Name']);
}
if($val['type']=='close') {
$name='';
}
break;
case 'mwg-rs:Area':
if($val['type']='complete' && $name) {
$faces[$name]=$val['attributes'];
}
break;
}
}
return $faces;
}
function get_face_coordinates($face,$info) {
if($face['stArea:unit']=='normalized') {
$rect=array(
'x'=>intval(round($info['x']*($face['stArea:x']-$face['stArea:w']/2))),
'y'=>intval(round($info['y']*($face['stArea:y']-$face['stArea:h']/2))),
'width'=>intval(round($info['x']*$face['stArea:w'])),
'height'=>intval(round($info['y']*$face['stArea:h'])),
);
return $rect;
}
}
function rotate_image_by_exif($image,$exif) {
if($exif['Orientation']) {
switch($exif['Orientation']) {
case 2:
imageflip($image,IMG_FLIP_VERTICAL);
break;
case 3:
$image=imagerotate($image,180,0);
break;
case 4:
imageflip($image,IMG_FLIP_HORIZONTAL);
break;
case 5:
$image=imagerotate($image,270,0);
imageflip($image,IMG_FLIP_VERTICAL);
break;
case 6:
$image=imagerotate($image,270,0);
break;
case 7:
$image=imagerotate($image,90,0);
imageflip($image,IMG_FLIP_VERTICAL);
break;
case 8:
$image=imagerotate($image,90,0);
break;
}
}
return $image;
}
?>
You can find an implementation on https://github.com/bobosch/webtrees
Since some years I sucessfully use this plugin: https://github.com/UksusoFF/webtrees-photo_note_with_image_map
In addition see this thread: https://www.webtrees.net/index.php/en/forum/2-open-discussion/30219-how-to-mark-individuals-on-group-photo
Hi @bobosch I recently added your code example to master branch in my extension (yep this is same is @marku5 link but renamed). If you still want see face tags in webtrees please try it :)
Face thumbnails improves the ancestors chart. But it's a exhausting work to create all these face thumbnails. Another difficulty is to use always the same aspect ratio to get a consistent look of the charts with images.
I'm using Google's Picasa to manage my pictures. Picasa can automatically mark faces on a photo. If you have enabled the "Store name tags in photo" option
Picasa stores this information in the file as XMP tag. The information also includes position, width and height of the face
By the way: The selections have always the same AR (and the nose in the middle ;-) )...
It would be a great help, if the media management would use this information to create additional thumbnails for each individual. If the media object is linked to a person, it should display the face thumbnail instead of the thumbnail of the full image.
If a photo has multiple face tags, a match algorithm is needed. Comparing the name should be enough, otherwise the additional information, which face tag belongs to which person has to be saved in the gedcom file. But I have no idea how. Perhaps as reference on the individual