hmm... i just did a quick test, just with one image that get's reloaded though but this one is 5000x5000px.

it does drop dramitcly the frameRate but is running now constantly with 1fps for around 15min

did you try to increase your memory already via the processing settings?

maybe also try this simple sketch to narrow down the problem.

1.  
2. PImage a;
   
3. void setup() {
4.   size(200, 200);
5. }
6. void draw() {
7.  
8.  println(frameRate +" | "+millis());
9.  
10.   a = loadImage("Blob.png");  // Load the image into the program 
11.   image(a, 0, 0);
12. }

Hi,

I increased the memory size to 2gigs and no luck :(

I' don't hava a camera to run the code. But i thought: is it needed to save the frames? Maybe just make an array of images and add to the end of it removing the first image. Make the array the size that fits into memory. Like in pseudo

Thanks for the feedback v.k.

How would I go about removing the the first image?

You can just shift the content of the array.
Or you can use a specialized data structure, like ArrayDeque.
A simple example: http://bazaar.launchpad.net/~philho/+junk/Processing/view/head:/_QuickExperiments/_Interactive/LineQueue/LineQueue.pde

Okay, I tried it creating an array of 50 images and copying in the captured Image. This seems to be stable memory wise.

Strangely though:

images[i] = myCapture;

Doesn't seem to result in images[i] having a copy of the particular frame captured at that time, it's content seems to continually update like a connection is made between the image and the capture instead of just copying the data.

Any ideas how that can be resolved? I'm a bit of a noob and have trouble coming up with anything useful while googling.

It is a common error, your = just copy the reference of the myCapture object. You might want to do a real copy, eg. with get().

My bad. I lead to the mistake. There is this cool example  from processing.org about storing inputs, its very simple and thus even better. I tweaked it to display images form camera with a delay. Just a quick try. I'm not drawing all stored frames as the example does, so i might made some wrong logic with frameCount, but looks good :). When starting it will display a faded image until buffer is filled. 

I've now got a fixed array of 25 320x240,RGB images, and only update those from the canvas with get();

Memory still grows and grows until the sketch crashes. That must be a bug.

Well the code above is not crashing and memory is behaving fine, what's your code now?

The sketch posted didn't work for me, trouble with my webcam I suppose. So I tried to implement as much of it as I could:
Some of the 'get()' stuff I didn't manage to get to work, I just ended up with a gray screen. I would also like to copy the webcam data 'off screen'. No idea how to do that so now I draw the webcam on the screen and then get() it to the array:

 
import processing.video.*;

String[] cameras = Capture.list();
Capture myCapture;

int captureFrame = 0;
int playbackFrame = 0;

int captureDelay = 0;
int stopMotionFrame = 0;

int totalFrames = 25;
PImage[] stopMotion = new PImage[totalFrames];
PImage emptyImage = createImage(320, 240, RGB);

//boolean sketchFullScreen() {
//  return true;
//}

void setup()
{
  frameRate(30);
 
  size(320, 240);

  // The name of the capture device is dependent on
  // the cameras that are currently attached to your
  // computer. To get a list of the
  // choices, uncomment the following line
  //   println(Capture.list());
 
  // To select the camera, replace "Camera Name"
  // in the next line with one from Capture.list()
  // myCapture = new Capture(this, width, height, "Camera Name", 30);

  // here is somebody else with problems of getting a camera to work
  //
  // http://processing.org/discourse/beta/num_1234875866.html
 
  // This code will try to use the camera last used
  myCapture = new Capture(this, 320, 240, 30);
  myCapture.start();
 
  for (int i = 0; i < totalFrames; i++){
          stopMotion[i] = emptyImage;
        }

  //myCapture.stop();
}

void captureEvent(Capture myCapture) {
  myCapture.read();
}

void draw() {
  background(0,0,0);

  image(myCapture, 0, 0, 320, 240);

  // Captures a frame at a time interval.
  captureDelay++;
  if (captureDelay>15){
    captureDelay=0;
    captureFrame++;
      if (captureFrame>=totalFrames){
        captureFrame=0;
        }    
  stopMotion[captureFrame] = get(0,0,320,240);
  }

  stopMotionFrame++;
  if (stopMotionFrame>=totalFrames){
    stopMotionFrame=0;
  }

  image(stopMotion[stopMotionFrame], 0, 0, 320, 240);
 
}

Hi, you are using 2.0, right? I still using 1.5. I had to change the "myCamera.start()" to "myCamera.read()" to run your code. It works like a charm in 1.5, very cool indeed. Memory used is steady around 170mb and virtual memory steady around 440 mb. I run it for some time and no problem. So it might be a bug as you said. You may try it under 1.5 to see.