mostly a codeBlog

I was lurking about over at Justin Windle’s blog Soulwire, wish my blog had an alter ego with a cool name..

He had been building some flow fields using Perlin noise, which my colleague think is Merlins evil brother for some reason.

I was writing a vector library for a particle engine Im going to publish here some day and thought that implementing theSoulwire flow field would be nice. Justin has been so kind as to put out the source, which is awesome  when your code is that good and visually stunning. As I played around with the code I got an idea, I think it is inspired by something I saw on the Processing site the idea was to use the input from my webcam as the noise map, instead of the MPerlin noise. I made a quick mockup, that actually looks quite promising. It needs some sort of effect, maybe an IR webcam, a 5×5 meter black wall to project it on, before it is ready, also the manipulation of the webcam image before it get’s read by the particles could give it a little extra. It is ticking along quite well on my Macbook Air, running the webcam and 5000 vectors.

For now I’ll just put the source code up here:

(It is build using Eclipse (Flex Builder Plugin), so it extends UIComponent, not something I would recommend, but if someone can tell me how to add Sprites to the stage of an AIR app, I will surley change it)

package
{
	import com.rickigregersen.flow_bump.Arrow;

	import flash.display.Bitmap;
	import flash.display.BitmapData;
	import flash.display.Sprite;
	import flash.events.Event;
	import flash.filters.BlurFilter;
	import flash.filters.ColorMatrixFilter;
	import flash.geom.ColorTransform;
	import flash.media.Camera;
	import flash.media.Video;

	import mx.core.UIComponent;

	public class Main extends UIComponent
	{
		//theese are the values I had fun playing with, but you might want something a bit more crazy:)

		private static const TRAIL : Number = 0.95;
		private static const BOOST : int = 110;
		private static const VECTORS : int = 1000;
		private static const BLUR : int = 4;
		private static const SCALE : Number = 6.0;
		private static const BACKGROUND : uint = 0x000000;
		;

		private static var fading : ColorTransform = new ColorTransform( 1, 1, 1, TRAIL );
		private static var blurFilter : BlurFilter = new BlurFilter( BLUR, BLUR, 3 )
		private static var grayMatrix : Array = [                       0.3, 0.59, 0.11, 0, 0, 0.3, 0.59, 0.11, 0, 0, 0.3, 0.59, 0.11, 0, 0, 0, 0, 0, 1, 0 ];
		private var grayFilter : ColorMatrixFilter = new ColorMatrixFilter( grayMatrix );
		private var brightMatrix : Array = [                      1, 0, 0, 0, BOOST, 0, 1, 0, 0, BOOST, 0, 0, 1, 0, BOOST, 0, 0, 0, 1, 0 ]
		private var brightFilter : ColorMatrixFilter = new ColorMatrixFilter( brightMatrix );

		private var camBMD : BitmapData;
		private var camBMP : Bitmap;
		private var renderBMD : BitmapData;
		private var renderBMP : Bitmap;
		private var vid : Video;
		private var cam : Camera;
		private var w : int, h : int;
		private var flowImage : Sprite;

		public function Main( _w : int, _h : int )
		{
			//width and height, also a sprite for holding the little particles;
			w = _w;
			h = _h;
			flowImage = new Sprite();
			//get the video;
			cam = Camera.getCamera();
			vid = new Video( w, h );
			vid.attachCamera( cam );

			//make a bitmap for drawing the video into
			camBMD = new BitmapData( w, h, false, BACKGROUND );
			camBMP = new Bitmap( camBMD );

			//a bitmap to stuff it all into and give some sort of effect
			renderBMD = new BitmapData( w, h, false, BACKGROUND );
			renderBMP = new Bitmap( renderBMD );

			//add the cam bmp and the rendered bmp
			addChild( camBMP );
			camBMP.y = h;

			addChild( renderBMP );
			//build the little vectors
			buildVectors();
		}

		private function buildVectors() : void
		{
			//place, rotate and scale them randomly
			var a : Arrow;

			for ( var i : int = 0; i < VECTORS; i++ )
			{
				a = new Arrow();
				a.x = Math.random() * w;
				a.y = Math.random() * h;
				a.rotation = Math.random() * 360;
				a.scaleX = a.scaleY = Math.random() + 0.1;
				flowImage.addChild( a );
			}
			addEventListener( Event.ENTER_FRAME, update, false, 0, true );
		}

		private function update( event : Event ) : void
		{
			//draw the cam to the stage, appy the grayscale Matrix, sometimes blur and brightfilters too
			camBMD.draw( vid );
			camBMP.filters = [                         grayFilter /*, blurFilter, brightFilter */  ];

			var a : Arrow;
			var brightness : Number;
			var radians : Number;
			var angle : Number;
			var speed : Number;
			var pixel : int;

			for ( var i : int = 0; i < VECTORS; i++ )
			{
				a = flowImage.getChildAt( i ) as Arrow;
				//get the brightness values of the pixel positioned directly under the vector
				pixel = camBMD.getPixel( a.x, a.y );
				brightness = pixel / 0xFFFFFF;
				//adjust spped and angle acording to the brightness
				speed = 0.1 + brightness * a.speed;
				angle = 360 * brightness * a.wander;
				radians = angle * Math.PI / 180;
				//make sure the vector is moving in the same direction as it's velocity
				a.x += Math.cos( radians ) * a.speed;
				a.y += Math.sin( radians ) * a.speed;
				//scale and rotate the vector according to brightness and
				a.rotation = angle;
				a.scaleX = a.scaleY = 0.1 + brightness * SCALE;
				//wrap around
				if ( a.x > w )
					a.x = 0;
				else if ( a.x < 0 )
					a.x = w;
				if ( a.y > h )
					a.y = 0;
				else if ( a.y < 0 )
					a.y = h;

			}

			//apply the colortransform, it determines the transparency of the bitmaps being stacked
			renderBMD.colorTransform( renderBMD.rect, fading );
			//draw the Sprite that hold all the vectors.
			renderBMD.draw( flowImage );

		}
	}
}

Here is the Arrow Class:

package com.rickigregersen.flow_bump
{
	import flash.display.Shape;

	public class Arrow extends Shape
	{
		public var speed : Number = Math.random() * 3;
		public var wander : Number = Math.random() * 0.5;

		public function Arrow()
		{
			draw();
		}

		private function draw() : void
		{
			/* 			graphics.clear();
			   graphics.lineStyle( 0, 0xFF0000, 0.4 );
			   graphics.moveTo( 0, 0 );
			   graphics.lineTo( 3, 0 );
			   graphics.moveTo( 2, -1 );
			   graphics.lineTo( 3, 0 );
			 graphics.lineTo( 2, 1 ); */

			/* 			graphics.beginFill( 0xFF0000, 0.9 );
			   graphics.drawEllipse( 0, 0, 6, 2 );
			 graphics.endFill(); */

			/*
			   graphics.beginFill( 0x00FF00, 0.9 );
			   graphics.drawCircle( 0, 0, 5 );
			   graphics.endFill();
			 */
			graphics.clear();
			graphics.beginFill( 0xff6600, 1.0 );
			graphics.moveTo( -1, -0.5 );
			graphics.lineTo( 1, 0 );
			graphics.lineTo( -1, 0.5 );
			graphics.lineTo( -1, -0.5 );
			graphics.endFill();
		}
	}
}

A few months back I finally got to clean up some of my classes in an old project. I realized that it was way overdue with a custom PictureLoader Class, a lot of the clutter that made my classes hard to read was try catch and Event handling. When loading a simple piece of graphics into an AIR app(Flash, Flex also of course) you will go through at least these steps:

1. declare loader, url and eventListener.
2. write listener method
3. start loading
4. catch complete event and add the loader to the displayList

Ok, not that much work and only a bit of clutter, however I am programming these AIR apps for a 52″ touch screen, meaining no mouse or keyboard, and any exception not caught will result in the UI being rendered useless and me having to VPN into the system and restart it. I really do not want to do this so here is the list that everyone should abide by when loading anything

1. declare loader, url and eventListener for ProgressEvent, IOErrorEvent and CompleteEvent
2. write listener methods for all of the above events
3. wrap the loader into a try/catch block to catch exceptions
4. catch complete or error event and remove all the EventListeners
5. cast the content of the loader into the appropriate type
6. copy this new content into a new object so that the loader can be null’ed and memory freed up

Ok, suddenly a lot of work and subsequently heaps of clutter. This is a result of the Event driven way of ActionScript 3.0 but a small price for the OOP architecture it now builds upon. I wished to encapsulate all the Event logic into the loader and have that deal with it, the Event code adds clutter. The solution uses the same approach as the Proxy Class so if You tried extending that this will make perfect sense. The main idea is to instantiate a PictureLoader object and pass it the methods that should be executed when events happen and let the PictureLoader handle all garbage collection and event clutter (Ok I’ll stop using the word clutter now, promise).

/**
* Author: Ricki Gregersen www.rickigregersen.com
* GPL: All code is snatch and grab
*/
package com.rickigregersen.loaders
{
	import flash.display.Bitmap;
	import flash.display.Loader;
	import flash.events.Event;
	import flash.events.IOErrorEvent;
	import flash.events.ProgressEvent;
	import flash.net.URLRequest;

	public class PictureLoader extends Loader
	{
		private var pic:Bitmap;
		private var loader:Loader
		private var percent:int;
		private var failFunction:Function;
		private var progressFunction:Function;
		private var successFunction:Function;

		public function PictureLoader()
		{
			//the worlds blankiest blank...
		}
		public function loadPicture(url:String, success:Function = null,  progress:Function = null, fail:Function = null):void
		{
			//set the local functions to your global functions
			failFunction = fail;
			progressFunction = progress;
			successFunction	= success;
			//make the needed objects ready
			pic = new Bitmap();
			loader = new Loader();
			//Add a listener for all the different events the could be fired
			loader.contentLoaderInfo.addEventListener(ProgressEvent.PROGRESS, pictureProgress, false, 0, true);
			loader.contentLoaderInfo.addEventListener(IOErrorEvent.IO_ERROR, pictureFailed, false, 0, true);
			loader.contentLoaderInfo.addEventListener(Event.COMPLETE, pictureLoaded, false, 0, true);
			//Wrap the load method in a try-catch to catch any wildcard errors that could break the VM
			try{
				loader.load(new URLRequest(url));
			}catch(error:Error){
				pictureFailed();
			}
		}
		//getters for the percent loaded and the picture
		public function get progress():int
		{
			return percent;
		}
		public function get picture():Bitmap
		{
			return pic;
		}
		//Functions to catch the events
		private function pictureProgress(event:Event):void
		{
			percent = (loader.contentLoaderInfo.bytesLoaded / loader.contentLoaderInfo.bytesTotal) * 100;
			//call the parents progressFunction to alert it that a loadProgress has happend
			progressFunction();
		}

		private function pictureLoaded(event:Event):void
		{
			//cast the picture into a bitmap, before this the picture is of type "loader.content"
			pic = (Bitmap)(event.target.content);
			//remove the eventlistener
			loader.contentLoaderInfo.removeEventListener(Event.COMPLETE, pictureLoaded);
			loader.contentLoaderInfo.removeEventListener(IOErrorEvent.IO_ERROR, pictureFailed);
			loader.contentLoaderInfo.removeEventListener(ProgressEvent.PROGRESS, pictureProgress);
			//make sure to free the memory and null the loader
			loader.unloadAndStop(true);
			loader = null;
			//call the parents pictureLoaded function to alert it the picture was loaded and ready to be picked up
			successFunction();
			//successFunction(pic); could have been used, just send the parent function a reference to the picture.
			//But I like getter/setters. As long as the method in the parent accepts a bitmap as input you don't have to worry about it
			//in here.
		}
		private function pictureFailed(event:Event = null):void
		{
			//Loading failed, so clean the loader and alert the parent
			loader.contentLoaderInfo.removeEventListener(Event.COMPLETE, pictureLoaded);
			loader.contentLoaderInfo.removeEventListener(IOErrorEvent.IO_ERROR, pictureFailed);
			loader.contentLoaderInfo.removeEventListener(ProgressEvent.PROGRESS, pictureProgress);
			loader.unloadAndStop(true);
			loader = null;
			failFunction();
		}
	}
}

This is how it is used, notice that this is complete with all the logic for listening for the complete, progress and error events.

			import com.rickigregersen.loaders.PictureLoader;

			private var picLoader:PictureLoader;
			private var pic:Bitmap;
			private var container:Sprite;

			private function init():void
			{
				picLoader = new PictureLoader();
				picLoader.loadPicture("assets/picture.png", picLoaded, picProgress, picFailed);
			}

			private function picProgress():void
			{
				trace(picLoader.progress);
			}
			private function picLoaded():void
			{
				pic = picLoader.picture;
				rawChildren.addChild(pic);
			}
			private function picFailed():void
			{
				trace("error loading picture");
			}

The functions passed to the PictureLoader are not mandatory, they are all set to null as default.

This means that the really short version of using this Class is like this:

			import com.rickigregersen.loaders.PictureLoader;

			private var picLoader:PictureLoader;
			private var pic:Bitmap;
			private var container:Sprite;

			private function init():void
			{
				picLoader = new PictureLoader();
				picLoader.loadPicture("assets/picture.png", picLoaded);
			}

			private function picLoaded():void
			{
				rawChildren.addChild(picLoader.picture);
			}

The use of rawChildren.addChild(picLoader.picture) is because this originates from a UIComponent in Flex so all other places you would use addChild.

So when building Components or just Classes that handle lots of Events and loading of external assets, this approach is something to consider.