Ryan Witmer

I'm wrapping up my new game, Magical Fairy Force, which supports the mighty 5200 trak-ball controller. I had a bit of an adventure getting this thing to work, so I'd like to post everything I figured out about this controller in order help anyone else that might be considering a project that supports it. I will disclaim that my methods may not be the best, but they seem to be working for me and are giving consistent results for the two trak-balls that I own. First, I'd like to highlight this thread: The information in this thread pointed me in the right direction, but some of the details were not quite right. Building on what I found here, I came up with the following procedure: First, it must be stated that in general you cannot uniformly handle the joystick and the trak-ball. You first need to figure out which controller you're looking at so you know how to interpret the POT values later on. Here is the procedure I used for this: Disable pot scanning by writing 0 to CONSOL. Wait for the POT shadows to update. You might also be able to manually trigger a pot scan and read the values straight from POKEY, but I opted to use the shadows. Read one of the POT values and compare it to $e4. When scanning is disabled, the joysticks report $e4 when read. I believe this same value is also produced by an empty controller jack and I'm not aware of any way to discern between a joystick and no controller at all. If the value is not $e4, this is a trak-ball. The value that was read is the rest value for that axis on the trak-ball. Save this value somewhere, you'll need it. Read the POT for the other axis. This is the rest value for that axis, the two axes can have different rest values. Again, save this value. Repeat for each controller that you're interested in. Re-enable POT scanning by writing 4 to CONSOL. When it comes time to handle input, you need to check each POT on the ball and compare the value to the rest value for that axis. The difference from the rest value indicates the speed and direction of the ball's spin on that axis. Values lower than the rest value are left/up, greater are right/down. The values that you get tend to be in a very small range. My game handles values +/-10 from the rest values. Once my game is finalized, which should be within a couple of weeks, I'll release the complete source so any interested people can examine how I did this in detail.

The single matches are human vs. human only, and there's no record tracking in the vs. mode. I had thought of both of these things, but by the time I got everything else done I was pretty much out of cartridge space. Perhaps when massive bank-switched carts become the norm on the 5200 I can do a new version with features and characters back-ported from the PC version that I've started working on.

RealSports Curling doesn't support the trak-ball. I remember considering the possibility when I was making it, and although it might have been really great for aiming it would have been terrible for sweeping. I'm pretty sure I was even testing it for a while with a trak-ball hooked up to it just for fun. It almost kinda worked for aiming even without dedicated code to handle the input, but not well enough for actual use.

Here's a quick video of me playing the basically complete single-player mode: This is a cool video because you get to see several of the special attacks in action. No narration from me this time, just game audio. I've made tons of progress, and the game is very close to being done. I have two show-stopping bugs that I need to address. One of them I have a lead on, the other only seems to happen on real hardware so it's much more difficult to debug, but I have a few ideas. The game content is pretty much complete. I still need to write the endings (just a couple lines of text each, no room for anything else), and make a few tweaks and adjustments here and there. I want to try to fit some simple music over the character select screen, but I'm not sure I'll have enough space on the cartridge. With any luck, I'll have a test build that I can post later in the week for people to fool around with.

I took some time off to actually play some 5200 games last night, and this thread inspired me to put some time into Space Dungeon. I've definitely lost my touch over the years, but it was still really fun. I had forgotten how much I hate those rooms with the corner lasers. Then I got distracted by Centipede when my brain went into its usual "how was this done?" mode. I'm happy to say that I think I figured it out, and I have a whole bunch of new ideas for graphics techniques that I want to apply to future homebrew projects. Probably play some more Space Dungeon tonight, and Blaster, which is just amazing.

As you can see, the 5200 is my jam. On the horizon, I'm excitedly waiting for 2600 Zoo Keeper, 5200 Magical Fairy Force, 5200 Mr. Do!, and 7800 Arkanoid.

I've mentioned this here and there, but I'm also planning a greatly enhanced and expanded PC version of this game. I've been doing some early work on it whenever my brain burns out on 6502 assembly, and I got this early version of the character select screen made up using the existing artwork. The working title is Magical Fairy Force: Champion Edition and I definitely plan on adding more characters in this version, possibly as many as 8, doubling the cast. I'm also planning an nice gallery mode with lots of screenshots and info on the Atari version. I'm planning all new art for this version too, should be pretty slick.

Other than Xari Arena, is there another 5200 game that supports multiple trak-balls? I'm not aware of any.

The hype is real, now!

I passed up on a loose K-Razy Shootout at some convention one time, and I'm still mad at myself for it.

I have to LMS all over the place because I'm using vertical scrolling to move the falling objects down the playfield. That boundary thing really caught me when I was doing Ratcatcher and I was just super confused for a while until I stumbled across the reason in some documentation somewhere. I'm really looking forward to trying out some tileset stuff at some point. I'm planning to use it for my next project, it will fit it really well.

No, pretty much everything is ANTIC mode E. I made the images by down-scaling the original high-resolution artwork, and then running the images through a small program I wrote that converted them to mode E compatible bytes and a color table that I could run through using an hblank interrupt. I also ended up compressing the image data in the ROM, which saved a lot of space but unfortunately slowed down the draw rate, which you can see a bit in the video. Getting the portraits to flip for the second player side was an adventure. I originally reversed all the bytes on the fly during the image draw, which was super slow. Later I ended up creating a "flip table" in an unused page of RAM that contained pre-reversed versions of all the bytes, and each byte of data could be used as an index into this table to get the reversed data very quickly. That sped things up a lot and made the draw time a lot more tolerable.

As it approaches completion, I wanted to get a nice little demo video of Magical Fairy Force made to explain the game play and features, So here's such a video, narrated by me and my cheap static-inducing microphone.

Here are the ROMs for my two games. ratcatcher.bin rs_curling.bin

Thank you so much for posting that thread. I've been trying to figure out for years why those particular games don't work on the 2-port BIOS so I could make sure to not make whatever mistake they did in my own game projects. Now I know, and it's something that I never would have done anyway, so it's all cool.

Also wanted to mention that I finally managed to get a second trak-ball. Last year's PRGE was the first time the game had two trak-balls hooked up to it, and there were some issues that I hadn't seen before. Now that I'm able to do better testing I've worked out all the trak-ball kinks, as far as I can tell. At some point I'm going to do a nice post in the programming section with all my findings and lessons on 5200 trak-ball support to help anyone else that might want to do a project for these awesome controllers.

I've scoured the documentation looking for something like this, but haven't managed to find anything. I'm using 32KB output format (-f3) and I would really like to know how much of that ROM space I'm actually using, and how much is just the fill byte. Presently I'm forced to hexdump the binary and just estimate, it seems like there must be a better way to get this information. Is there some kind of switch for this that I just haven't tracked down yet?

I noticed this same behavior with Starmaster on my 7800, also by Activision. I haven't decided if it's the game that does this, or the 7800 is doing some kind of trick to simulate the color/BW switch through the difficulty switches. It would be interesting to find out the what's really going on here. Playing around with a non-Actiivision game that needs that switch on the 7800 would be enlightening, but I don't have one to test.