Computer Animation


Animation is one of the hardest kinds of programming to get right, but it's so obvious when it's not. Other kinds of programs might have errors that nobody knows about, but animation is in your face.

The human eye is designed by God to integrate successive images differing only slightly into the perception of a single scene in motion. The eye will perceive flicker at less than 20 frames per second (fps), but most sense of motion persists down to closer to 1 fps. Commercial movies are shot at 24 fps, American TV at 30, and some of the high-end HDTV monitors refresh up to 72* fps. For our little experiments we will be satisfied with something under 10 fps, because JavaScript is interpreted and probably not capable of doing better. There is a further problem that JavaScript has no direct graphical output, so all our images must be constructed from character graphics (using text to simulate images) and/or pictures prepared offline in some kind of image editing program. GIMP, as its name suggests, is somewhat lame (like all so-called "open source" software, it takes a lot of effort to make it to do anything useful), but it's a free download (try here or here or Google "GIMP download"). You will want to create GIF pictures, although PNG also works in most browsers.

For full animation control you need a separate image for each frame. GIF also lets you do animation within a single file, but that's a different game. For your program to control it, you need separate images. I made eight beachball images, each slightly different from the next (but all identical in size), which you can see here:

For your animation to be credible, you need to observe how real balls (or whatever) move, so you can reproduce the same flow of motion. You also want the differences between the images to be a lot smaller, but this will demonstrate the techniques. Here I tried to get a little precession (wobble) in the ball rotation, but have it come back to the starting point so I can re-use the same eight images as the ball rolls across the screen.

In HTML we display an image thus:

<img SRC="Ballz/Ball0.gif" BORDER=0 height=29 width=29>
where the height and width are the actual dimensions of your image. When you output this text using document.write() inside your script, it shows the picture at that place in the window. If you give some other sizes, the image will be stretched or shrunk to fit, but that will take extra time that you might prefer to use in a higher frame rate. Images in HTML are placed next in the normal text flow on a standard web page. We deal with that in this program by filling the lines above and the space to the left with something else, in my case another image that is plain white and stretched to the size I want. As before, I will pass the image state in the hash code attached to the URL.

For the animation to flow smoothly, you need precise timing between the frames. JavaScript has a couple timer commands (technically methods of the Window class), of which we will use one. Similar to our Tic-Tac-Toe programs, as soon as we start to redraw the screen, all our script goes away, so we will again alternate between two similar URLs with identical code.

As mentioned earlier, JavaScript is pretty weak on input and output, and browser differences make that worse. This is a fact of life in computer programming. Fortunately, with widely deployed programming languages such as C++, Java, and JavaScript, you almost never are the first to encounter your particular problems. Go on the internet and search for whatever problem you are having. Chances are, somebody already had the exact same problem (or at least something similar), and somebody else already answered it. I did that to find out how to detect when the player clicked on the ball, and then to know the exact click coordinates. After it worked on my computer but failed in Windows Explorer, I went back on to find out why. You can look at the code in KickBall.html (see also KickBell.html) for details. The event capture syntax is in my interpreter, but it doesn't do anything. The X and Y coordinate extraction properties of events are not supported in my interpreter, and I didn't try to make the StopIt() function work in Windows, because I only used it for early testing.

This should get you started. Have fun!

Tom Pittman
rev. 2010 December 23

* Note on Frame rate. I think the perception of flicker depends mostly on how much time the screen is showing something other than what the viewer is supposed to see. Movie projectors spend about 2/3 of their time showing the film, and 1/3 of the time (about 100th of a second) in the dark while moving the film to the next frame. Old-style TV tubes had a little tiny spot of electron brightness flying all over the screen, so 99% of the time any one part of the screen was dark. They fixed that by alternating odd and even lines (called "interlace") so the dark time was 1/60th of a second, then they used high-persistence phosphors in the tube which did not go out so quickly after the electron beam moved on. As a result, the image was not dark any longer than in the movies. HDTV projectors are "progressive scan" (not interlace), so the dark time is essentially doubled, requiring a much higher frame rate to overcome the flicker. Modern liquid-crystal (LCD) displays (including most so-called LED displays) depend on the sluggish quality of the organic LCD goo to keep the image there long after the electronics has gone on to refresh the other pixels -- in fact they work hard to speed that up, because otherwise rapidly moving images tend to blur; you will see more flicker from the flourescent backlight than from the LCD itself.

Out slow animations spend a lot of time white, while the JavaScript and image rasterizer are working to replace it with new image. Clever programmers often "double-buffer" the image -- that is they leave the old image up on the display while preparing the new display invisibly in memory; then when it's ready, the blit (a fast transfer of pixels) from the memory image to the screen. This way the use sees no dark (or white, as the case may be) screen at all, only the actual pixels. If they are not too different, the eye integrates them very well. For now we depend on the browser to do that for us (or not).