Question

I have been working on a Java 2D game for a little while. It is a raster graphics system with an array of pixels (integers).

private BufferedImage image = new BufferedImage(WIDTH, HEIGHT, BufferedImage.TYPE_INT_RGB);
private int[] pixels = ((DataBufferInt) image.getRaster().getDataBuffer()).getData();

I then create an array of integers separately, manipulate it with the screen objects (rendered from external image files, such as .png), then copy that array into my main one, which is projected on the screen. I found no performance difference (and really didn’t expect to) in using array copying methods over iteration. Regardless, the graphics render well and the game is coming along swimmingly.

However, I have found this to be extremely CPU intensive. My activity monitor says that the application is using more than 100% of my CPU. This is obviously because I am iterating through a pixel array (76k integers) each update (60 times per second).

I chose this technique for educational purposes. This is a personal project and I simply wanted to get insight into Java graphics. I am, by no means, married to this rendering technique.

My question comes in three related parts…

1. Obviously there is a better way to do this. What libraries/frameworks would do it better?
2. Will those libraries essentially do the same thing (loop through the pixels), just in a more efficient way?
3. Is there a way I can optimize this technique without using external tools, or is it just not worth it?

Answer 1

OpenGL is most often associated with 3D graphics applications, but it can also be used for 2D applications. Using an orthographic projection and textured quads to take the place of sprites and background images, you can construct the graphics and interface for a game with absolutely no 3D elements at all, despite the fact that the drawing is being done, in a sense, in 3D.

I have no experience using OpenGL in Java. However, I know that it is possible using the LWJGL library and possibly others. You should check it out if you want to drastically improve your graphics performance.

To answer your second question, OpenGL in this application actually works very differently from your approach, using the same techniques it would use to draw textured polygons in 3D rather than simple block image transfers into a framebuffer, though pixel data can be accessed and manipulated.

Here’s an example of 2D graphics using OpenGL with C, from a game project I never finished but nevertheless may serve as a good visual example of what I’m talking about. In this case, I did not use an orthographic projection, but rather a perspective matrix to get parallax effects by drawing the various layers at different depths in 3D space.