I am a bit late writing this this week. I normally post my site and this blog on thursday evening as that way I can keep my friday's free. This week I ended up going out on thursday. Last week on Blazing Games, I posted the open source version of my Easter Egg Hunt game where you navigate a 3D maze in order to find a dozen easter eggs. The eggs are randomly generated. The 3D maze was generated using a very old technique known as the painter's algorithm. I was thinking that none of the techniques used in the game are overly noteworthy then I realized that a lot of the techniques in the game, while not useful when you have OpenGL (including ES and WebGL) to work with, may still be relevant if you were limited to a 2D drawing system like the HTML 5 Canvas. HTML 5 developers, at least the ones that are targeting Internet Explorer, may need to fall back on these old techniques. This is due to MS deciding not to support WebGL.
The egg is simply a palette swapping varient. There are 12 patterns for the eggs, but the two colours that are used to draw the pattern are randomly selected. This is a really old technique that you will see in older video games where the same monster sprite is used repeatedly but given different colours so the same artwork can be used for a large number of monsters. This would be a little tricky to do with the HTML 5 canvas as I don't recall a way of changing an image's palette. I would have to look into this, but if you can't change the palette, you can use the image data to manipulate individual pixels.
The 3D maze uses a technique known as the painters algorithm. Essentially you treat the 3D maze as a series of blocks. You draw the furthest away blocks first, then the nearer blocks, and so on until you are painting the close blocks. The downside to this is you are potentially overdrawing many times. The nice thing about this technique is it is very easy to implement and the results can look very good. Of course, you aren't going to get 360 degrees of smooth scrolling movement, but it is something that can be done on any browser that supports the canvas.
It would be nice if there was a hardware accelerated 3D standard that would work on all browsers, but even if that does happen, it won't be for years. Perhaps for next Easter, I will port the game to HTML 5, though I am hoping I come up with a far better idea for an Easter game before then.