Tower Defense: Design discussion

[Adam Schmelzle]

Hello all,

I thought I'd open up discussion on my latest game, since I'm not settled on what I want to do. I know how I want the interface/controls to work, and I have the creep pathfinding hammered out already. I can alter the level in real-time and have the baddies act accordingly.

What I am trying to do now is decide on the towers. I could go the classic route, and have the usual mix: standard(air+ground), air only, ground+splash, freeze(ground+air) . This is my fallback.

I'm also not 100% decided if I want to have flying creeps at all. Gameplay wise the flyers bypass your towers completely, and all they do is force the player to dedicate some valuable land to anti-air towers. I don't know if I ever liked that, but is that extra little requirement necessary to keep the player on edge? It may be needed.

Every time I start to think of new and cool things to do, I find myself making the system complicated. I need to keep reminding myself that the average person playing this game will not be an avid Warcraft3 junkie. I probably can't get away with having an intricate upgrade system, without scaring people away.

Any thoughts?

[Adam Schmelzle]

Well, I see that we've generated a lively discussion on this topic!

I'm leaning towards having 3 basic tower types: rock, ice and lightning.

Rock is 1 target, fast shooting direct damage
Ice slows a single enemy down
lightning will target multiple enemies at once, but have a long recharge time. Like a splash damage tower.

My goal here is to keep things ultra-simple for a tower defense game. I don't want a complicated interface, and I don't want to have a lengthy tutorial, or any tutorial. In the mobile space, the simplest games perform the best. That being said, it'll still be a tower defense game, so there's always strategy in how you place your towers.

[vlad]

Hey Adam

Your design problem is not your ideas, is the medium. If you don't want tutorials, you are assuming that the player already knows the game. The first time I played a tower defense game, it had no tutorial and I had a problem until I understood the purpose of it. But I'm not the ordinary player and we do not represent our audience.

So I wonder what is the threshold of the common mobile player, is he willing to try to understand or will he quit the game ASAP? If he doesn't quit the game and he is willing to try it, why not raising the complexity even if just slightly?

[Adam Schmelzle]

I won't really have a tutorial, but I will have a bit of text before each level, and work up the difficulty. For instance, it almost works like a tutorial, but not interactively

"Place towers where they can cover the path" The first level will be very small, low resources(enough for a basic tower), and with a fixed path that only allows the player to build in like one of 3 spots. After a level like that, it should be clear what a basic tower can do, and that placement is important. If they place in one of the 2 bad spots, they WILL lose.

Second level: "Try an ice tower to slow 'em down!' And it'll be a straight path, so and 'ice' tower or two will be required to slow the baddies down enough to survive.

Third level: "Lightning hits large groups at once!" Level has a large number of baddies in a tight group, and the only way to kill em is with a lightening tower.

Fourth level: "If the path is not striped, you can divert the creeps with a maze of towers!" Small level with a straight path, but player is now allowed to place towers over the path. Player will win if they divert the path around the tower.

Fifth level+: Increasing difficulty.

There will be some first-time help text at the start of the game in the same way I have help in my other games. Also, I guess I didn't mention it, but the first 3 levels will have the path as a no-build zone. In my game the path of the baddies is actually shown visually on-screen unlike most(all?) other tower-defense style games. This helps the user determine the effects of placing towers, since they will be able to SEE the path of the baddies change in real-time.

I am not 100% decided if I will have strict build/creep phases, where you are unable to build while the current baddie wave is running through. I am leaning toward distinct built-time/creep-time as it's less stressful, and makes saving the game easier.

The 3 basic towers will be upgradeable as well, but I think it'll just be simple upgrades, like ice gets faster, rock gets more powerful, and lightening will get a larger radius. Multiple levels of upgrades of course.

[Adam Schmelzle]

If I do it this way, I guess there's no reason NOT to have some more advanced 'hybrid' towers that are only available after upgrading the basics to a target level. If the 'hybrid' towers are not even available until the upgrades are made, the first-timer's won't be confused by the strange tower that cannot be built.

So Maybe I'll put in 'hybrid' towers. These towers would have characteristics of 2 of the basic types of tower, and their footprint will be 50% bigger x and y (3x3, instead of 2x2). If you have upgraded a tower to level 5, and you have another level 5 tower of a different type anywhere on the board, you will have access to 'hybrid' towers (only of the combined type) for placement. You won't be able to upgrade old towers into 'hybrids' because of the change in footprint, but I'll make it obvious that a new tower is available to be built by using a pop up.

The player won't have the resources for the upgrades required to make hybrids until later on in the game, so it should confuse any of the new players.

[vlad]

Apr 21, 2008 8:46:54 GMT -5 Adam Schmelzle said:

If I do it this way, I guess there's no reason NOT to have some more advanced 'hybrid' towers that are only available after upgrading the basics to a target level. If the 'hybrid' towers are not even available until the upgrades are made, the first-timer's won't be confused by the strange tower that cannot be built.

Indeed. To increase the game's longevity a bit you can put levels in the middle at increasing numbers. The player will have time to grok the concept, have fun and you a more enjoyable game with a higher value. Something like this:

tutorial + level 1
level 2
tutorial + level 3
level 4
level 5
tutorial + level 6
level 7
level 8
level 9
tutorial + level 10

And so on. I don't know if you read "The Theory of Fun", but the basic principle is this. Player's fun comes from learning and automating a mechanic ("groking" in the book) and using it to conquer a certain objective at increasing rates of difficulty and wider time intervals. What I've been designing lately applies it using this principle: The higher the difficulty, the bigger the interval.

Hope this helps, although I think you are already right

[Adam Schmelzle]

Yeah, that's a good idea. I should ease the player into a bit more. It's been a while since I made a game that requires a bit of thought/setup from the user.

[Adam Schmelzle]

I've been slacking on this for far too long, and now I'm getting back into it. What I'm trying to hammer out now, is the tower building interface.

Of course, I want this to work well on screens with limited size, so as always it's a challenge to come up with the perfect interface. Unlike any of my other games though, tower defense is normally a mouse driver affair, with significant screen space. This is what I've been pondering for the last day or so. I think I may end up with a very boring, yet functional interface for choosing the tower to build, since anything cool&fancy takes up too much screen space, or isn't descriptive enough

Anyhoo, I'm mostly just rambling about what I'm doing to get myself motivated enough to actually get something coded. However any thoughts/suggestions are welcome.

[Adam Schmelzle]

Hello all!

I now have the basic game engine up and running, with creeps, and towers that shoot. The best part is that the game performance is independent of the number of towers. This was necessary, since I want to have levels with layouts that are much larger than the average mobile screen. So far I've run it on a few of my Midp2.0 devices, and I'm completely happy with performance. There still a bunch of low-level optimizing to be done, so performance will only get better.

I'm not completely finished implementing the splash and chain-damage. It's coded, but untested. It should be entirely possible to have levels with hundreds of towers. Performance will depend more on the number of creeps running around. I haven't played any mobile TD games with real-time path finding, only the fixed-path type. Maybe it'll be a first.

[Adam Schmelzle]

Chain lightening now works. I'm a bit undecided as to exactly how I want the chain hits to behave. I've got 2 choices, and both have merits..

1) Chain hits jump from one enemy to the next, not spreading out
2) Chain lightening spreads to nearby enemies on each jump, resulting in (potentially) an exponential increase in chain hits. This looks much cooler in game than standard jumping.

It's trivial to switch between the two(one line variable change). With the spread jumping, it behaves like the chain lightening in Diablo. Of course I'd have to reduce the damage overall becaue of the significant number of hits per tower shot. I could have the damage reduce on each jump as well, so that the later hits don't do as much.

When I'm 'done', it's going to take a good deal of time to balance everything out.

[Adam Schmelzle]

Working on the level editor today. I'm 90% done, since I'm just modifying my 'Attack Breaker Pro' editor. I'm going to make as many levels as will fit into 50KB(runtime mem). It'll probably only take up 5KB in the .jar file, but I don't want to eat any more than 50KB in game. If I want more levels I'll just start splitting the levels into 'packs' so that I don't need so much heap space.

I get 2 tiles per byte, plus up to 4 baddie paths. Each baddie path is comprised of up to 8 waypoints. Along with the level data, will be 'waves' of baddies. I'll probably have up to 50 waves per level, where each wave has spawn info for 1->4 different paths.

I could save a pile of work if I use the same set of waves in each level since the level layouts and starting cash will be different anyway. I'll only be able to decide after actually playing it though.

I think I'm going to have the player's 'gold' carry forward to the next level. So maybe you'll start level 1-1 with $50.00, and when you survive all 3 or 4 waves(whatever it happens to be), you get a reward of $100, and carry forward the remaining gold.

This way it could give people reason to go back to previous levels, and replay them to achieve a higher carry forward. I will of course let people go back and replay any level they want. So if I am starting level 35 with a mere $300, and want more, I can go back to level 2, and complete it as cheaply as possible, maybe freeing up an extra $10.00. That extra $10.00 would be added to the starting funds of the next level or possibly every subsequent level(including lvl 25).

The last level will have endless waves so you can use up all the money you have saved from all the previous levels. The online highscore will be the number of waves you can survive in the final level. Man, I want to play this now! That sounds awesome!

[Adam Schmelzle]

Level editor works, and I can load em up in game now as well.

[chrisl]

Hi Adam,
Out of curiosity, are you also coding your editor (and other tools) in Java?

[Adam Schmelzle]

Yuuuup, Java end to end. It's my language of choice, and I do absolutely everything with Java. I'm a Java nut.

It's actually been years since I did anything with C/C++/Lisp/VB/whatever. My last 'job' was as a phone support guy for HP and Compaq computers *shudder*

Back in my school days I was a Teaching Assistant for the first year Java course at the University of Western Ontario. I only did that one year, but enjoyed it greatly. Now I try to help people mostly on the sun CLDC/MIDP message board...

forum.java.sun.com/forum.jspa?forumID=76&start=0

I'm known as 'Hooble' over there. I'm always trying to draft people from that board over to here.... end of rant.

[Adam Schmelzle]

I hate this part of game development. I'm pretty much done the game mechanics, and level editor. Now I actually have to turn it into something presentable. I can't stand making menu screens and info windows, and all that graphical fiddyness. To me this part of game development is like writing a book report. It's times like this that I wish I had a team of minions that would do this stuff for me while I frantically wave my hands and point my fingers at them.

...which explains why I just released a new flower game. I'm easily sidetracked at the moment. I've also signed up for 'Dream Build Play', the M$ XNA game dev contest. Stupid thing is that I can't see other peoples game projects unless I pay M$ $60.00 for a 'premium' XNA account. What a load a bull.

[chrisl]

Same problem here. It's easy to get motivated and feverishly code the core game engine in a few days, but now I'm getting bogged down in the overall framework stuff, menus, auto-scaling graphics, saving, and so on.
And then there will be the daunting task of compatibility check/debugging/optimization...
I'm not there yet... So I try a keep a reserve of small interesting gameplay features that I can implement between more mundane tasks.

Also, in your case, I suspect a new flower game may have more revenue potential for far less development effort than a brand new game with unproven gameplay?

[Adam Schmelzle]

Well, a new flower game may just spread the flower players around a bit more. It's hard to say if people play more than one of them, or only decide to download if they like the particular flower in the game. It is relatively little effort, so worth a shot.

Hmm... I could cut out a ton of menu's and stuff if I make this a one-level survive as many waves as you can type game, with a high-score on the kill count. Then make a bigger and better version if it does well. I did that with 'Attack Cannon Challenge', where I did 'Water Stealers' based on the same code, but with many more features.

... that sounds really tempting right now. I want to spend some time with the XNA stuff, and could get to that a lot quicker if I shorten up the first iteration of this game. With 1 level, I don't even need a menu screen. It wouldn't even be bad, since it would be a lot like playing the standard level in 'Desktop Tower Defence' at that point.

[Adam Schmelzle]

Funny, how typing that out just made me motivated again. Now I see myself at the point where I'm 90% finished, instead of 50%, and have the 'finish it!' drive. 5 minutes ago I had the 'just get it over with' drive, which doesn't help much. I could be done this game next week.

Should be a good market test for the gameplay. oooooo I could do 2 different games, one with a large fixed path, and the other with the totally open board where you block the path with the towers. The exact same code(just 2 different levels), and it'll let me know which gametype is more popular. Sweet, I'm gonna get 2 games out of this in a week(or so).

[vlad]

Now that's motivation! Keep it up and let me know when the tower defense is done... I'm drooling in anticipation.

[Adam Schmelzle]

Hmm, If I rename it to : 'Strategic Defence Plan" or something like that, I can keep the 'programmer art', and say it's the interface for the general's computer...

[chrisl]

That leads to another question: is GreyStripe involved at all in the number and rythm of release of your games, updates or variants? Do they have any politics in place regarding their "virtual shelf space"? I understand you may be in a special position regarding this, due to the high success of your games, but still, I would be interested in knowing the kind of influence they have or not on your release strategy (from your last posts, it seems you do as you want, which is cool!)

[Adam Schmelzle]

I do as I please, they are my slaves.

Seriously though, they don't impose their own quality filter on the stuff publishers submit and release. When I submit a game, and click 'ready for release', it goes into their queue for release. At that point, they do some minor testing, and then release it. No crap filters at all.

I'm not sure what type of testing they do, but they certainly don't block crappy games. I think they're just looking for violations of terms of service like sex, child abuse, etc. They likely do a little bit of on device testing just to make sure it runs on some of the phones the publisher indicates are compatible as well.

They do releases on Tuesdays usually, and their official word is 2x a month, but it can happen on 'any given Tuesday' as far as I can tell. That's when their website refreshes the ranks for the games as well as the partner catalog.

They're pretty quick at releasing now. I haven't had to wait more than 1 week for the last few I put into the system. I remember a time last year when they were starting to get a lot more games for release than they were used to, and they decided to only release games once a month. That was painful, cause it could take 2 months before th game showed up if you didn't submit 2 weeks before that month's release day.

I think they started pruning games out the catalog that perform terribly based on the revenue generation. That only happens when a game has been out for quite some time, and is still a stinker. It'd have to be really terrible to be de-listed, probably crashing and unresponsive. I've got a couple horrid performers in there 'Attack Brain' and 'Mobile Rosary' that are still listed, and rosary has made an average of $0.17/day. Thats BAD.

[chrisl]

OK, so the key here is to manage to get an account from GreyStripe, and then you're free to adopt any strategy you like. That sounds good.

[Adam Schmelzle]

Yuppers.

[Adam Schmelzle]

I just got Greystripe's latest 'Consumer Insight' report for the period of Feb 15th to May 15, and of the top 20 revenue generating games in their system, I had 5. Interesting thing is #1 was 'Funky Monkey in Funky Monkey Land', which is a platformer I think. Gotta do one for sure now...

[chrisl]

Yes I saw the report also: impressive performance, all the more as an indie!
Platformers are nice but (usually) require a lot of graphics and animations, which takes time and/or money. It also must be much more difficult to keep a reasonable jar size, but maybe that attracts another audience in search of flashy graphics for their higher-end handset.
So many parameters...

[Adam Schmelzle]

I had another look, and it turns out that the games listed aren't actually ranked based on revenue, they simply list the top 20 revenue generating games. The list of the top 20 is ordered alphabetically by category name, so mine might actually be the top still.

I want to make a platformer after my TD style game. I plan on doing it in a simple manor though, with more emphasis on running/jumping like in Super Mario Bros 1. May not even have baddies, but I haven't gotten into any planing yet, since I'm busy with TD.

A link to that report is totally going into my resume, if I ever have to get a 'real' job again. The way things are now though I'm not exactly keeping it up to date After my utter failure the last time I had to 'get a job', it sure will be nice to not look like an average punk that just graduated.

Working for yourself has it's own kind of stress, even when things are going well. My wife and I are planning baby #2, and that's freakin me out a little, which would explain my random off topic ramblings...

[chrisl]

I totally see what you mean when you talk about the stress of working for yourself, especially when your family counts on you. That's the exact reason why I resist to quit my day job and start developing Java games for a living...
That said, if you never worked in a typical company environment, it's probably difficult for you to fully appreciate the kind of liberty you're currently enjoying: my wife being also her own boss, she's often mystified by the kind of constraints you're confronted to when working as an employee. You know, just simple things such as having someone telling you what you have to do tomorrow, or having to ask for permission for taking a day off. This is nothing special when you've always been an employee, but once you're working for youself, I think it's probably difficult to adapt (or readapt) to such a system.
Anyway, congratulations on your plan to expand the family, enjoy your way of life and keep up the good work!

[Adam Schmelzle]

I worked at Honeywell Hi-Spec for 18 months back between 3rd and 4th year of university. I was a VB/C++/SQL guy, sitting in a cube 9-5. I hated it. Stupid things always happened, like outsourcing a component, and then having to re-write the entire thing in-house when it was done. Waiting an entire month for the architects in another city/country to get some design specs finished so we could officially start working on something. Meaningless meetings. Unskilled co-workers that cared more about the spelling in your comments than the actual code, etc...

Never going to do that again. Ever. Rather die. Not that I don't like doing software, but I'd only be comfortable in a small team, that didn't have to wait for approval from 'higher up' for absolutely everything. 5-foot high cube walls, seem more like a prison cell.

[chrisl]

OK, so you see what I meant. At least this sad experience gives you a very high motivation to keep doing it the indie way!