Saturday, November 11, 2017

In Retrospect: Voodoo Cheval (Part 2)

The previous post left off with the re-introduction of snakes in Wave 6 of level one, so we'll pick up from there.

Wave 6

The snakes are significantly faster than zombies, and paired with their capability to split in two should the player make a wrong move, they prove to be a significant challenge when tossed into the spawn wave mix.  

Up until this point, the player has been able to get by with just drawing circles of various sizes on the screen, fighting off larger and larger hordes of zombies.  With the addition of snakes, I wanted to keep a similar pace to what has been developed so far without having to place a large emphasis on a new enemy type, hence why two snakes were placed in this wave, along with 4 zombies.  

As snakes appear first on the screen, it gives the player a moment or two to process what they are seeing and try to recall where they had seen it before.  This was, in hindsight, the wrong way to go about it.  Snakes should have been a part of the mix from the beginning, considering we introduce them so early on in the experience.  This could have even further prevented the use of the exploit (discussed in the last post) from the beginning, keeping it from even being called an "exploit" in the first place.

Thankfully, if they fail to react and kill the snakes in time, we have placed a check point here to avoid the frustration of starting over at Wave 1.

Wave 7

A bit of a throwback to the lesson I was aiming to teach in Wave 2, with the single, large grouping of zombies.  This wave, an arrangement of 8 snakes that were split into 2 straight lines, were meant to suggest visually that the player could chain together multiple snake kills with a single drawn line.  Two lines of snakes were used instead of one to, again, avoid slowing down the pace too much when introducing a (possibly) new idea.

Waves 8-10

More practice in iteration, applying concepts I attempted to develop since Wave 1.  Overall, I'm pretty happy with the way these placements turned out, they strike an odd balance between symmetrical and random.  Enemies are, for the most part, consistently pouring in from every direction into the screen, without favoring a particular side.  This keeps the player actively analyzing the entire scene, making quick decisions as to which enemy to take out first.

Wave 11

The last enemy that gets introduced in Level 1, the "Demon".  The demon comprised of 3 possesed skulls, all chained together by magic.  The idea was that to defeat the skulls, you would have to trace the chain that connected them.  Sometimes, this suggestion would be a little more vague than others (see above screenshot). 

This was a bit of an odd one to work with, as his prefab asset was designed to randomly place him on the map, with the skulls placed in a somewhat random arrangement up to a certain distance from one another.  Originally, we had aimed to have 10 waves in total for Level 1, but somewhere during development of the other levels, it was decided that the demon should be introduced earlier in the game.  I don't recall if we even had discussions of offering the demon a tutorial like we did the zombie and the demon...  But I liked the idea of giving him his own wave, and letting the player figure out how to defeat him.  The chains, being connected in straight lines, should have (hopefully) suggested what action to take.

Waves 12-15

The last set of iterative levels.  Now that I have played through the whole game a few times, these last few felt anti-climactic... I'm unsure as to why this ended up being the case, but the only thing that I can recall was worrying about the integration of the random demon spawn with the rest of the enemy placements.  Placing that asset required thought and timing in relation to when the other enemies would appear on screen; I didn't want to run into the situation where the player was unfairly overwhelmed.

I am pleased to see that, for the most part, I remained consistent in applying the concepts from earlier parts of the level (varying sizes of zombie hordes, grouping of snake-lines), as it certainly added variety to my input actions.  The game never quite felt stagnant, the only change between waves that I could feel upon re-playing the game was a rise and fall in flow .


"Voodoo Cheval" has it's flaws, but as "my first shipped title", and my time working with a large team on one project, I enjoyed the experience and am rather proud of the final result.  Joining VGDA made me a few friends, but I also got the opportunity to learn a lot of development practices and concepts first hand.  These lessons may seem obvious to you as a seasoned developer; they're obvious to me now even with only a year or two of experience under my belt.  But these lessons were important nonetheless, and sometimes the only way you will learn them is the hard way:
  • Play-test early, play-test often, play-test with everyone.
    • It's obviously important to play test your game frequently and thoroughly, but it's also important to try to play with a mindset of someone who has never seen your game before.  This is the job of the average QA tester, someone who will dive into a game and try to break all of the rules; accounting for every possible thing that could go wrong or could be done differently than what the designer anticipates (see 'The Exploit' in part 1).  This could yield better results in the long run when you take your game to external play-testing.
    • Play-testing the whole game (or level, in my case) sounds exhausting, and I'm pretty certain I had my aversions to the idea during development (even though my part of the game lasted only 5 or 7 minutes...).  But it's truly the only way to really gauge the difficulty curve of the experience as a whole.
  • Look outward for inspiration.
    • Something I realized looking back at this development process was that I never ventured outside of Voodoo's bubble to gather inspiration on how to approach the spawn-wave design.  I would often try to sketch waves on paper and try to determine the best way to use the pieces at hand, or simply drop assets into Unity and adjust their locations until I had a wave that simply "felt good".  Now that I have a few other projects under-way, I am constantly doing research and looking for inspiration to better the game-play experience overall.  I can't help but wonder how different this level could have been if I had put that into practice earlier.
  • Open and constant communication.
    • This wasn't a lesson so much learned as much as emphasized.  As a college student, working on a game part-time with a club, your priorities shift often.  Class work builds up, exams get brutal, and tasks that you either aren't payed for or graded on will get pushed aside.  Communication in this kind of situation is huge if a project is going to see completion by a deadline, as assigned tasks that you might not be able to complete can be handed off to someone else.  As VGDA did it's best to mimic an Agile environment, there were several techniques and tools that I had to become familiar with, like Trello or weekly Scrums, that assisted in keeping communication open and updates constant.
Overall, as small of a game as Voodoo Cheval is, my time developing it was worthwhile.  I've since continued working with VGDA on our next title, and I can't wait to see what challenges we overcome together.

Friday, November 10, 2017

In Retrospect: Voodoo Cheval (Part 1)

During a QA session with some of the recruiters at Blizzard, it was brought to my attention the importance of maintaining documentation for your design and development process.  While I reviewed the documents I had made for previous projects, I had the idea of making a collection of blog posts that review particular challenges or struggles that I encountered and what my solutions to them would be.  Looking back at my experience in this way, I hope, will help to shed some light on things that may have gone right, as well as things that I could have done differently.  Consider these posts to be bite-size post-mortems.

Voodoo Cheval is the first (and at the time of writing, the only) game that I have worked on that has been published onto a distribution platform.  If you are unfamiliar with the title, it is an action game intended for mobile devices, where the player is tasked with defending their beloved horse from an onslaught of zombies, snakes, and other monsters by drawing particular shapes on the screen to destroy them. 

I was brought onto the project rather early in development, signing up for a role that sat between design and programming, but leaning more towards the design end of things.  The team, like myself, were all full-time and part-time students of California State University, Long Beach, who had banded together to form the Video Game Development Association (or VGDA, as I will refer to it from now on).  What I found so brilliant about this opportunity was that very fact; we were all students.  We all had coursework to worry about, jobs to help pay for rent and tuition, very little free time, and a huge passion for video games.

The team was fairly large in scale, and with each of us only being able to contribute so much time each week to development, tasks were divided into smaller, more manageable pieces.  While I had the opportunity to work with other students designing systems, namely enemy behavior and player controls, the bulk of my work was in designing Level 1; an open setting where 3 types of enemies would appear on the screen in waves.  By the end of development, there were 15 waves in total, each with it's own purposeful design with the aim losing the player within "the magic circle".


Upon starting the game from the main menu, the player is met with a very brief and simplistic "scripted tutorial" scene, where the player is met 2 types of enemies that they will see throughout the remainder of the game.

Although this tutorial was not of my own design (will give credit when I find out who exactly built it), it had a large influence over how I built the rest of the introductory stages.  The goal was to make the tutorial completely interactive, without hindering the start of a play session too much.  As the enemies approach the horse on screen, the player will have the opportunity to defeat these enemies in the usual fashion, but if they do not know what to do, the snake and the zombie will pause, staying frozen in place before they can cause any harm.  When they freeze, the spell effect (usually visible on the screen when touch input is detected) is animated to suggest what action to take, almost in a "monkey-see monkey-do" kind of way.

No pop-up windows, no invasive companions to say "Hey! Check this out!".  Just a simple suggestive animation paired with a brief safety buffer before the game really begins.  For a new player, this brief encounter establishes the foothold that the following enemy spawn waves will build upon.

Wave 1

Looking back on it, I'm pretty sure I over-thought some of these enemy placements...  It was clarified before I started that enemy spawn would not be random, but placed.  With replay-ability being a concern, I felt it best to keep these placements a little on the unpredictable side.  Hence, this asymmetrical placement of the first two zombies.  An easy win, but it also should have served as an indicator that, from this point forward, there will never be just one of these guys on the screen at a time.

Wave 2

The idea behind this wave was to re-enforce an idea that might not have been obvious to every player: the fact that you can kill more than one enemy with the same attack.  Even if you attempted to draw a circle around a single zombie in this grouping, you are bound to rope in another one with how close together they are.  This was also meant to establish the idea that the size of the shapes you will draw can vary, which we will find out later can either help or harm you immensely.

Waves 3, 4, and 5

The third wave was intended to push the player to use the different sizes of circles that were demonstrated between waves 1 and 2.  The zombies on the left and right are the first to appear, and they are spaced relatively far apart.  It would be fairly easy to circle each one individually or circle them both together.  The grouping at the top of the screen comes shortly after, a little more tightly knit together, and finally there's the big grouping at the bottom, appearing slightly behind the grouping at the top.  

At this point, the player may start to feel a little frantic, drawing circles around zombies the second they pop onto the screen.  But if they are patient, they may be able to see that by allowing the zombies to draw in closer, they can draw one big circle to take them all out at once.

Unfortunately, we never implemented any sort of reward system for combo kills.  Something that would have made this design process a little more meaningful.

Wave 4, shown below, tries to make that idea of the risk and reward of patience a little more obvious, having the zombies enter the screen in the form of an actual circle.

In hindsight, it might have made more sense to swap Waves 3 and 4 to make this point, but I think it also works in regards to replay-ability.  Now that you are fully aware of how this kind of strategy could pay off, you might want to play through it again and see how many zombies you could kill with a single drawn circle.  An unexpected issue that this design may have contributed to, however, is the discovery of an exploit that many players would try to abuse,which I will discuss shortly.

Wave 5 was simply a zombie rush, with assets placed in various positions in an attempt to establish rhythm, but while avoiding predictability.  Designing waves like this one (there are many others in Level 1) was simply an iterative process, where I would place the zombies in various positions and re-position them until the wave felt like it could hold a "flow"; not too challenging, but not to boring either.

The Exploit

Something that become apparent to us rather quickly during play testing was the fact that the player could, in fact, draw circles big enough to wrap around the whole screen, if they were quick enough to make the  beginning and ending points of the line connect.  I don't quite recall at what point or which wave that play-testers would come to this realization, but I'm pretty sure it was around waves 3 or 4.

When a discovery like this one is made this early on, it makes the "strategic placement" of enemies essentially moot.  Even more unfortunate, the team never got around to properly addressing and fixing the issue.
A (rather poor) demonstration of the exploit being used in Wave 3.

Looking back, there are a few reasons I can think of as to why this wasn't fixed.  One is the deadline; as students, the goal was to make one small game over the course of a semester, all the way up to the week of final exams.  While we did do a fair amount of internal play-testing, this issue wasn't really brought to our attention until we had students from other colleges play the game at the Student Game Developers Alliance Summit in May, not long before our deadline.  At this point in time, we were no doubt scrambling to get bugs fixed and the final art assets in place, which ultimately still resulted in a cohesive experience we were all proud of by the end of development.

Another reason that comes to mind is how ineffective this strategy becomes after Wave 5, as snakes are re-introduced for the first time since the tutorial.

Something that might not have been made clear in the tutorial is this particular trait about snakes:  if you accidentally draw a circle around them, they "split", and form two smaller snakes that the player now has to deal with.  If the player were to simply sit there and spam large circles on the screen, without paying attention, they would be overrun within moments.

I recall at the time feeling that this enemy design would counteract the exploit so easily that it would be abandoned after Wave 6, which could also be how the rest of the team felt when the exploit was discovered.  In either case, it taught me to try to play-test my own work from a perspective of someone who doesn't know what to expect.  This seems so obvious to me now, after reading books and articles on bug testing or human centered design, but as this was one of my earliest projects, I'm willing to forgive myself and move on.