newt : I really would like to have that, too !

There's a bit of a solution with Actionscript, I think. Yesterday, Adobe released the version 2 of Alchemy, their compiler allowing to recompile C++ as Flash code, allowing the use of direct memory access and special opcode acceleration. With that, you can gain a huge boost, especially when crunching numbers (like what's needed for face identification).

I used the first build of Alchemy inside the code used for my Mic plugin (to accelerate mp3 and ogg encoding). Without that, it would have been impossible to encode audio in a browser...

Edit : by the way, I'm asking myself the question if I should do a post each time I create a new plugin inside the "Creation" section of C2, because I think that there is a lot of user "afraid" to go inside the "develope section".

All : new build on first post. Now every function where it makes senses have the option to respect white pixels or not.

bjadams : I updated the first post with a new build. You were right, I only added the option to respect white pixels to the "Invert" effect. Now every effect where it makes sense offers that option.

I also added a new .capx for the demo, because on my machine, with Chrome, the behavior works with Families.

If it doesn't work with your white Sprite having a black border, can you post here the .capx (or give me a link via PM) to help me debug it ?

newt : <img src="smileys/smiley2.gif" border="0" align="middle" />.

Facial recognition isn't doable in Javascript, you need a native language to do that. It's based on various things, like Eigenfaces, and for that you need to capability to invert and resolve big matrices really fast (PCA, SVD and the like.).

Update v1.1 : now minification works properly !

The plugin : https://dl.dropbox.com/u/1412774/FaceDetectDemo2/pode_facedetect.1.1.zip

Here's a new behavior, allowing you to detect a face in the Sprite it's attached to. You can get the (X,Y) of the found face, its width and height, the confidence (only the first found face is reported).

<img src="https://dl.dropbox.com/u/1412774/FaceDetectDemo/demo1.png" border="0">

As you can see in the screenshot, there's also the possibility to retrieve a cropped version of the found face (you can give the margin you want to add around the original found face, since the bounding box can be a bit "tight"). The cropped image is given as a base64 string that you can load in a regular Sprite.

<img src="https://dl.dropbox.com/u/1412774/FaceDetectDemo/demo2.png" border="0">

The whole thing was made using the helpful jQuery FaceDetection plugin : http://facedetection.jaysalvat.com/

The demo : https://dl.dropbox.com/u/1412774/FaceDetectDemo/index.html

The plugin : https://dl.dropbox.com/u/1412774/FaceDetectDemo/pode_facedetect.1.0.zip

The .capx : https://dl.dropbox.com/u/1412774/FaceDetectDemo/FaceDetectDemo.capx

(The whole detection thing is using the Viola-Jones algorithm, implemented first in OpenCV, and reimplemented back in Javascript by liu liu in ccv.js : https://github.com/liuliu/ccv & http://liuliu.me/ccv/js/nss/. It's a fast algorithm - that's why it's viable in JS, relatively simple, but you can get some (few) false positives, and sometimes no detection, although you know there's a face in the image !)

I'm planning to add several face detection in the same image in a future release.

— : you provide a zip file to the launcher. Everytime you change something in the game, you need to recreate the zip file.

puoidirloatutti : You can use iframe if you need. Just set the url of the iframe as the url of your game.

bjadams : does it works that way when you say to respect white pixels, not to respect white pixels, or all the time ?

DeMayunn : I suggest to go the WebWorker way for your plugin, because WebCL isn't available in browser, apart from installing a dedicated plugin or using a special built (from Ericsson, or other vendors).

— : they decided that after you have started your application with the launcher, there is no way to go back to the launcher "homepage" without killing the app as a prior. Their argument : your final app will be seen as completely independent of the launcher by the final user, so, in the end, you won't have a "option" to "go back to homepage".

chrisbrobs : I made it to work with simple properties. Your gradient is a declared filter. I need to check how I implemented it <img src="smileys/smiley2.gif" border="0" align="middle" />.

bjadams : I'm going to check the families problem. Does it happens when you have several Sprites in a Family, and one of them has the behavior, or all of them ?

By the way, what kind of coloring book are you doing ? I don't know if you saw it, but I gave some code to R0j0Hound to integrate the "floodfill" ('paintbucket') in the Canvas plugin. With that, you can colorize black & white images with color where you put your finger.

(My question : your "anxious little kid", is it your inner child, or your real child ? <img src="smileys/smiley2.gif" border="0" align="middle" />)

bjadams : I tested it on Safari :

BUT there is a problem on Safari!

1. Press save

2. Press colorize Red. Works ok

3. Press restore. Works ok.

4. Press colorise Blue. IT DOES NOT WORK

and you are right, there's a problem. It seems that when you restore an image with it's initial value, you can't apply effects anymore on Safari. I need to dig deeper inside Safari inner working to understand how they handle security policies regarding base64 images.