INTERVIEW
Joe Zbiciak

Joe Zbiciak is a Texas Instruments programmer which has created new
games for the Intellivision. Joe also coded the jzIntv Intellivision emulator,
and is considering emulating the rarely seen Intellivision Keyboard Componant

MT> Why don't we begin by having you share a brief synopsis of your background with programming and entertainment software.
JZ> Basically, I figure I've been an electrical-engineer-in-training most of my life. (Well, at least, starting with the Radio Shack "25-in-1 Lab Kit" that I got for my 5th birthday.)

Some of the earliest videogames I played were the classic Pong (two potentiometers on a plastic black cabinet with a built-in speaker), the "6-in-1-Pong-on-Steriods" type games, and of course the occasional Atari 2600 game when visiting relatives.

My dad got an Intellivision around 1981, as I recall. It was a gigantic leap forward compared to the other video games I had played (basically "Pong with a slight twist" games). I had a blast playing the Inty games, and in the back of my mind, I always wondered how they worked.

I first started programming when I was about 6, hammering out CBM BASIC programs on the babysitter's husband's VIC-20, and later on the Commodore PETs at school. I wasn't a real accomplished programmer, but I was definitely a rather avid spelunker into the wonders of "making it do things."

Fast-forward a couple years, to the great crash of 1983. The depressed computer prices finally allowed me to get my own computer, a TI-99/4A. There, my love of programming and my love of video games were cemented together in perpetuity. The TI sports the wonderful TMS9918A VDP, and with TI Extended BASIC plugged in, it actually makes a reasonable system for writing simple games. And believe me, I did (and they all sucked, but who cares?). (Also, it got me my current job. The person who hired me was on the TMS9918A design team, and he was impressed with my knowledge of the part... ;-)

From that point on, video gaming for me has been a combination of enjoying the game play and wondering how it was done. Throw a healthy hardware interest in on the side, and you have me in a nutshell.

MT> Why your love for the Intellivision, and which Intellivision games are your personal favorites?
JZ> The Intellivision had a consistent style all it's own, and a certain charm that other machines lacked. The times I played the Atari 2600, for instance, just never seemed as satisfying. There just seemed to be something missing -- a lack of real depth. (The recent homebrew carts often seem to center around fixing these issues -- witness the rewrites of PacMan, for instance.)

My dad's Intellivision itself still sees somewhat regular use -- it's played at least once a year. It's never been decommissioned. He keeps it up at his cabin, and we'll play games on it if the weather's crummy or if the mosquitos are just too obnoxious. (After all, the Mosquito is the Michigan State Bird.)

My personal favorite Inty games from "back then" included Astrosmash (top score ~830,000 on 'hard', at one sitting), Bomb Squad, Snafu (dig that music!) and Utopia. I was also fairly good at Worm Whomper.

Utopia was especially cool, because Alex (my brother) and I would sit for hours filling up the screen with fishing boats, PT boats and everything we could buy. It was the ultimate "keep ahead of the Joneses" type of game in that respect. Also, it had some neat bugs where you could push ships off the lower right corner and "sail on the islands" after coming back onscreen. A world of wonder crammed in 4K.

In recent years, Space Spartans has become one of my favorites. I was too young to appreciate the relatively rich gameplay it provided back when I was younger, but now I really enjoy the strategy elements.

MT> You created the Intellivision emulator, jzIntv. What technical problems and/or difficulties did you encounter while programming jzIntv?
JZ> The biggest problem I ran into was in finding reliable documentation. When you combine that with the fact that I didn't have an actual Intellivision handy (I was in Texas, my dad's Intellivision gear is in northern Michigan) kept my aspirations of writing an Inty emulator on hold for a couple years. I basically gave up on the idea before even starting.

Then, I stumbled onto the Blue Sky Ranger's website, and that really got the nostalgia stirred. After some poking around, I then stumbled onto Frank Palazzolo's 2-Bits site and found interesting tid-bits such as "De Re Intellivision" as well as scans from an old Osborne book. That really got me stoked. After a couple of false starts, I finally got rolling on jzIntv.

My emulator really owes alot to the fine work that Carl Mueller, Jr. did in developing INTVEM (later INTVPC) and documenting what he found along the way via William Moeller and De Re Intellivision. Truly, without that source, I would be very far behind where I am today.

Presently, the main difficulty I run into now is lack of time. :-) That's partly because I want to do a thorough job, and partly because I try to do a thorough job at my actual, paying day job. :-)

Carl's emulator broke new ground, providing the world with its first Intellivision Emulator. He did a fantastic job, but his emulator isn't perfect. It's very, very good, but not perfect. Currently, I'm investigating some of the darker corners of the Inty and am discovering rather interesting details, such as the fact that the STIC actually steals cycles from the CPU, and so on. I'm very interested in accurate emulation, rather than simply adequate emulation.

An emulator is adequate if it plays all (or most of) the existing games. However, if you want to make new games, such an emulator might let you get away with things that the real hardware would not allow. Carl's emulator falls into the "adequate" category, because it allows certain things (such as accessing the STIC registers and GRAM during active display) that the real hardware doesn't let you do. That doesn't matter if you're just playing already written games. However, I'd like to write a few new ones, so I need a bit more accuracy. :-)

MT> What differences are there between your Intellivision emulator, and other Intellivision emulators that have been released.
JZ> The main difference, I think, is in the generality of the implementation. INTVPC is written as a DOS program, complete with hand-coded assembler for some pieces, and direct accesses to the hardware. It's not very portable, but it runs great even on minimal hardware.

The Mac INTV emulator that comes on the Intellivision Lives! CD leaves me wanting more. It has the nice GUI wrapped around it, but overall it's nowhere near the level of quality of Carl's emulator. I'd say more, but I'm following the maxim "If you can't say something nice...."

Then there's jzIntv. From the beginning, I wanted jzIntv to be a fairly portable, clean, well documented Intellivision emulator. I wanted this partly so that others could learn about the Inty from it and write their own emulators. I also wanted a free emulator that anyone could use, so that nobody would be bound to the Intellivision Lives! CD-ROM. And most importantly, I wanted something that'd run under Linux. :-)

Currently, I feel jzIntv's strong points are its portability and its debugging functionality. (Also, I'm very proud of its sound emulation. Carl Mueller, Jr is borrowing some ideas from it for a future INTVPC revision, as well.) I'd like jzIntv to be the "programmer's emulator", one that's suitable for developing new games on in a timely fashion.

Already it has one feature in it that I only dream of seeing in gdb. The current development version has a "crash trace", which dumps out the last several thousand instructions (complete with register history!) leading up to the crash. Very useful.

MT> You've included a Linux kernel driver within jzIntv to use Intellivision controllers as regular joysticks via the Hand Controller Interface. Now's your chance to boast, anything to contribute on the subject?
JZ> Actually, I can only boast on behalf of Linux in general. What made this small feat possible was the fact that I can get the other joystick drivers for Linux, make a few modifications to one of the source files, and voila! I have a complete, robust, stable joystick driver of my own.

Try that with Windows sometime.

Vojtech Pavlik did a very good job of structuring the Joystick driver and providing great examples to work from.

MT> Tim Lindner has been porting your jzIntv code to the Macintosh. Any other contributors to the project?
JZ> Currently, nobody other than Tim and myself are contributing directly to the code-base. Frank Palazzolo donated his disassembler to the project early on, but that's about it for direct code contributions.

I'm very thankful that Tim has pitched in on this project. I think it's a great example of the Good Things that can happen when you throw something like this out to the world to see. At the time he ported jzIntv to the Macintosh, I didn't even have access to a Mac, let alone one with a compiler. (Now, my fiancee has a iBook, but we still have no development tools.)

Tim sends me an email one day saying, basically "Hey, I got your emulator to run on my Mac." I about fell out of my chair with surprise. It was awesome. I've never met Tim in person, but I hope to someday. He seems like a great guy, and I'm glad he's helping out.

On the indirect side, the list is alot larger. Carl Mueller, Jr., William Moeller, and Scott Nudds all provided the base from which I worked. Doug Parsons has provided valuable feedback by running various tests for me from his Foomboard. And of course, there's always the interesting tidbit here or there from the BSR's site.

Chad Schell, over the past few days, has been running a number of tests for me on a development board of his own. (Some of these actually provide some of the timing and other information I was planning to measure myself, that I mentioned in the interview.) He's also given me some additional insight into the weird GI bus protocol. (Basically, there were some bits that were all but undocumented, and we worked them out together, although I'd say he did most of the real grunt work.)

I'm sure I'm forgetting alot of people in this list, but these are the main ones.

MT> What motivated you to start creating new titles for the Intellivision?
JZ> One piece of motivation was Mike Hayes' Programming Contest. (I felt kinda bad after taking 2nd place in his last contest with "SIMON SEZ".)

Another was the fact that while I was growing up, I could never work out how to write a Tetris clone of my own. The main problem was that I was thinking too linearly, when really Tetris has some decoupled elements. Also I was thinking too hard about the problem... there's no real need to get clever with Tetris.

Looking back, I was a bit silly programming-wise back then, but that's what happens when an 7-yr-old teaches himself programming. ;-)

MT> How did you approach creating a new Intellivision title?
JZ> Well, the first step was to get my feet wet writing some routines and short demos. I took a vacation at my dad's cabin last summer, which was when I got the idea to do a Tetris clone. So, I sketched out some rough ideas, and wrote out some real basic routines for getting and putting pixels. Over the next week, I fumbled around with Carl's assembler and manage to put together my first couple demos -- Maze Demo and Bouncing Pixels. The main reason I was able to do that then was that I had with me a laptop computer that I had borrowed, so I had access to DOS so I could use Carl's tools. (At the time, I only had Linux boxen at home.)

Once I got home, though, I didn't do much programming on anything Inty for awhile -- it sat on the back burner. I did do development on jzIntv, but no Inty-native code.

I finally got frustrated by the lack of tools, and so I went and found a portable cross-assembler, the Frankenstein assembler, and hacked a new target into it, the CP-1600. (The CP-1600 is the Inty's CPU.) With this new assembler available, I was able to resume programming on Tetris and related efforts.

Most of the Tetris development occured while I was on vacation over Thanksgiving and Christmas. The first step was to instantiate a small framework -- get the cartridge header in place, get a couple small init-routines in place, and then think about how to structure the rest.

The first piece I built was a mini Real-Time Operating System (RTOS). It was pretty simple and could only barely be called an OS (it was really just a task scheduler), but it was fairly fundamental to the overall structure. The next bit of coding then revolved around writing the support routines (random number generation, print-string, print-number) so that I could at least get a title on the screen. :-) And then came sound.

The sound format was devised in a bowling alley while my fiancee's family was bowling. I had a sheet of graph paper, a pencil, and a few crazy ideas. For one, I wanted the music to be continuous -- if a sound effect played, it should avoid killing the music, or at the very least, the music should keep time. Second, I wanted music and sfx to share the same engine. The result of my brain-storming was an early version of my music format. I then coded it up (on paper!) and tried it out the next day, and to my surprise, it worked! Once sound was there, development plateaued for a bit while I played with the sounds. ;-)

After I got over the fanciness of the sound, I focused on getting the actual Tetris game logic infrastructure written. Before, I was writing general support routines -- getpixel, putpixel, play a sound, print a string, random number, etc. Now I started writing more complex routines such as "detect full lines", "clear lines", "calculate score", and so on. I built up little test drivers which would try to exercise these routines somewhat, and benchmark them to make sure I wasn't overrunning the machine.

Eventually, I got to the point where I could manually place Tetris pieces around in the well, and scoring worked, and music worked, and so on. But the game wasn't complete. I hit a slump. I could see the end from there, but I just didn't feel like taking the last few steps. I spent inordinate amounts of time tweaking and cleaning up bits that didn't need it all that much, but I didn't feel like taking the final step.

Then, I "recovered". I got that final gust of wind which pushed me over the finish line, and Voila! Tetris was a complete game. I then spent around a week or two polishing up the game, tweaking the music files and music format, and generally making it all it could be -- in 8K. I also ended up writing a Pong clone which I hid inside Tetris. (Later, I broke it out separately, added a chincy AI, and submitted it as a separate game).

Probably the biggest technology feat in Tetris was the sound engine. I ended up writing a "music compiler" which converts (in a round-about fashion) a MIDI file into a music sequence for the Intellivision. The music format itself has rather fancy forms of "sliding window" and "delta update" compressions built in for compact coding, and the compiler itself makes full use of these features. I'm rather proud of it, considering that it was brainstormed in a bowling alley. ;-)

If I had it to do over again, though, I'd encode it differently, perhaps with different SFX and music engines that at least cooperate to share the PSG. There are some definite limitations to my approach, but it works rather well in Tetris, so who am I to complain?.

MT> Which of your primary Intellivision creations are you most proud of: Tetris, Pong, or your Song Player?
JZ> I'd have to say Tetris, because it contains both Pong and Song Player in it. :-)

The Song Player is basically my sound engine broken out as a separate program. I did this so I could play around more easily with the sound format, and also so I could explore 6-voice sound some as well. (The Song Player just runs the same sound engine twice, once on each PSG, with one playing the "left" stream and the other playing the "right" stream.)

Pong is actually embarassing, coding-wise. I coded it totally as an afternoon hack, and it's got some pretty grotty code in it. :-)

MT> You recently won first place in Mike Hayes' Intellivision Library 1999 Intellivision Summer Development Competition with your Tetris entry. Is your mother proud?
JZ> ...and father and brother and sister and fiancee and... Yes. The family is proud of me. Right now, I'm looking at ways to make it playable on the _real_ Intellivision, but don't hold me to anything just yet. :-)

MT> Were there any monumental revelations you discovered while programming your demos such as maze, groovy, bncpix, or others? Something distinctive that made you think, "Gee, I didn't know the Intellivision hardware could do this?"
JZ> The main revelation-inducing program was "Groovy". The fact that you could play 4-bit audio via the PSG was hinted to in the PSG emulation in MAME -- it was never discussed or mentioned in any of the GI spec sheets that I saw.

So, I decided to try it out. It worked in jzIntv, but that didn't mean anything in the "Real World." So, Doug Parsons tested Groovy on his Foomboard, and to our pleasant surprise, Ash lives on in the real Inty! Groovy!

MT> What is the typical time cycle for development on your projects?
JZ> Anywhere from 1 hour to 1 decade. :-)

For instance, on Sunday, I spent about 3 hours writing a Navy DIF -> email form letter conversion program for my fiancee (in C, from scratch). She TAs a geology lab, and it was rather tedious to send out grades manually. So, now she exports her spreadsheet as a DIF file, and my program does the rest.

Tetris was a two-three month project, for the bulk of the work. (I don't really count the baby-steps I took in June. Rather I think of it more as about Nov 20th to Jan 20th.) I'm sure I'll tweak it some more (already I plan to try to squeeze a third song into the same 8K!), but it's pretty much finished.

jzIntv is my longest running publicly visable project. I started it with a few false starts back in 1997, with it really taking root in late 1998/1999.

At work, I've been coding on, extending, and maintaining an internal tool that we use there for the past three years. It's basically been a pet project of mine that whole time, and TI absolutely loves me for it.

And then there's those projects that never truly die, such as some of my graphics hacks that I occasionally revisit when I feel nostalgic. I actually whipped out my Apple ][e and did some assembly coding on it a few months back, for instance.

MT> Any plans or future projects concerning the Intellivision?
JZ> Currently, I'm in the process of designing a "SoftCart" for the Intellivision that would allow the user to download games and play them on the Inty. I'm still torn between a number of cost-vs-flexibility tradeoffs, although I think I'm coming to a reasonable solution.

I'm also working on figuring out more of the Inty's timing parameters at a hardware level, since I don't think those have really been explored all that greatly. I have an oscilliscope, and already I've discovered some interesting things.

These timing parameters and other details will probably be important for my ultimate project, which will be to write a Keyboard Component emulator. :-) I don't know if I'll ever get to it, but I'm certainly going to try.

Watch an animation of Joe's Maze Demo

Good Deal Games and the Intellivision community would like to thank Joe
for his part in helping enhance our Intellivision experience.

Visit Joe Zbiciak's website
E-Mail Joe Zbiciak

 

 

Are You Involved with Classic Games?!?
Let us know, and we'll interview you!



E-Mail: GOOD DEAL GAMES
GOOD DEAL GAMES HOMEPAGE

Copyright © 2003, GOOD DEAL GAMES