Touck KO: Tech Review

Environments were given a cap of roughly 3000 vertices and in conjunction with the main characters models utilizing roughly 4000 vertices, each frame consisted on average 7,000 vertices at any given time. Speaking with the brothers, one thing that really piqued my curiosity was the term “Static Batch processing”. As Matt explains it;

“We are benefiting from Unity’s static batching process. The arena level, for instance, is about 5400 vertices total (ca. 9000 triangles), but because we have cut it up to take advantage of static batching, almost half of those vertices are not really computed at any given time. Basically, because of how Unity’s batching process works, we can have a single screen worth of environment that packs as much detail as an entire environment would under other circumstances. In essence, thanks to Unity, the GPU limitations don’t really have too much of an effect on us. Things like script performance optimizations and memory from audio make a bigger difference at this point.”

Texture wise the boxers used up to 7 textures across 4 materials, and somewhat surprising to me was the use of two separate 1024×1024 texture sheets for the skin and tattoos, this would explain the great fidelity in graphics especially in the boxers faces. One 256×256 texture were used for the gloves, a 128×128/256×256 for the trunks and one 64×64/256×256 for the shoes. Yes, if you’re thinking it’s a lot you have to take into account customization was done with texture swapping. It’s important to note that reducing draw calls whenever possible is solid practice, but when unable too, further optimization is always key;

“While stating all of this, I am realizing that we can probably further optimize these now with little effort since I believe we can again do PVRTC compression on non-square textures.”

{click image for larger view}

Most environments consisted of 14 to 16 texture sheets across 7 to 8 materials. Each texture included accompanying light-maps ranging from 32×32 to 1024×1024 where needed to add depth to each scene.

Although animation sets are limited, there is some authenticity to the ones available, this is due to the fact that key-framed and mo-capped animations (to a greater extent) were used in conjunction with one another. Although mo-cap animations add “realistic” behaviour to your animations, they do carry a ugly caveat called mo-cap clean up. This is excluding the fact that if you don’t have direct access to a mo-cap studio, sifting through available libraries for that right animation tends to be a time consuming process, not to mention converting those libraries to a usable format can all be diplomatically stated as “less exciting”.

“Carnegie Mellon University’s Media Lab has a pretty large library of free motion capture data available online, including a variety of boxing moves. Someone handily went to the trouble of converting it all into BVH format, which we can use in Maya. The biggest hurdle was really browsing it effectively to find stuff we could use and then of course cleaning it up. Adam’s technical artist background came in handy here, as he made a bunch of custom tools in Maya that helped speed a lot of this up.”

Technicalities

Utilizing standard industry software, the brothers worked with Maya 2009, Photoshop CS4 with it’s newly added 3D painting system and their treasured Unity iPhone 1.1 license.

“We should also mention that we use UnityDevelop for coding, which is what we use at Flashbang. Matthew Wegner made some adjustments to the open-source FlashDevelop environment and created some handy tools to let us easily jump into Windows through Parallels from within the Unity Editor. It basically gives us things like code completion that helps speed up work.”

Hardware wise, the brothers use in their words, rather outdated Macbooks one of which was a Pro, and a couple of 3G iPhones for testing. So with them both having 3G iPhones for their testing bed you maybe wondering how they could account for 1G models of the past;

“We both have 3G iPhones, and our friend RC Torres helped out by providing his ancient relic — the iPhone 1G — when we needed it for testing.”

If you’ve read up to this point, it should be of no surprise that Unity iPhone 1.1. was used for development, and those of you who frequent TCG would also note that Unity is an engine highly touted by myself and I recommend it’s use whenever asked. When asked what they thought of Unity’s capabilities.

“We used Unity iPhone 1.1. We tell people if you are wanting to make 3D games for the iPhone and you are not using Unity, there is a good chance you are doing it wrong.
Seriously though, for a couple thousand bucks (or a couple hundred if you don’t need pro features) you can get an engine that easily outperforms anything you could make with a couple thousand bucks worth of your time — unless you are Tommy Refenes maybe.

Unity has some absolutely brilliant engineers working on their iPhone team, so they have been able to write a fair bit of more complex operations in assembly for the iPhone’s vector coprocessor. That is how we can have two high-detail characters with lighting in detailed 3D environments. The cost of doing animation and skinning — really most transformations – is incredibly minimal as a result.”

Technical hurdles are forever present in any project you undertake, the key to reducing this eventual obstacle is from experience and the right tools. For the brothers, using Unity in their pipeline greatly reduced most technicalities that became apparent, this was namely due to Unity handling many of the more complex issues for them. In addition Adam’s technical artist background and Matt’s programming and mathematics background were essentially all they needed to steer Touch KO in the right direction while keeping it on course. But of course when it came to things out of their hands namely the reoccurring culprit – memory, problem solving and clever workarounds kept everything in perspective;

“In terms of programming, our biggest enemy was memory. Not the amount of memory per se, as textures, sound (especially sound), and the like take up much more, but simply reducing the rate that we allocated it. The trick with Unity, since it is using Mono, is to try to keep down the number of allocations for objects on the heap. Otherwise Mono wants to invoke garbage collection quite a bit during game play, which can slow things down.”

Finally on the audio end of things, TKO uses 1 licensed track with an option to create your own set list from music stored on your iPod. When asked about the difficulties in acquired licensed tracks for TKO, they brothers had this to say;

“That all depends on what the music is you are trying to license. If we had wanted to license something like Bad Religion or NWA, we would have no idea how to go about it!
We used a professional music licensing service, Jamendo Pro, which lets users license music from their library of artists pretty painlessly — they basically help put people like us in touch with musicians. The only real difficulties are finding something you like that fits, and then being able to afford it. We had originally wanted two or three other tracks by the same artist, but we simply could not afford it up front since our budget was pretty much out of pocket.”

Coming away with a victory

After the successful release of a title, the learning opportunities surface in an abundance, you quickly become aware of tricks to reduce time, pitfalls you should avoid the next go around and so fourth. For many, myself included, it’s precisely the reason I migrated to the game industry, the fact that challenges are always readily waiting to be triumphed over did it for me.

“We have certainly gotten better at optimizing Unity content for the iPhone. If we were to create another game of similar complexity, it would probably take about 75% of the time this one took.

We have also learned a fair bit from our partnership with Chillingo. Because they have such a large portfolio of titles with lots of very successful products, they have had plenty of great suggestions not only about development but also in handling communications with the outside world and responding to players’ needs and wants.

Finally, it can be challenging trying to balance our inclinations as gamers with the realities of the iPhone market place. The vast majority of users are simply not hard-core gamers willing to sit down for hours and devote themselves to the mastery of a really deep system — they want something that they can start, play, and quit in a couple of minutes, since they are gaming in different contexts from a PSP, DS, or console gamer.”

Up for Round 2?

In addition, the brothers are still hard at work with their thinking caps well worn and their ears permanently affixed to their user base regarding additional features and fixes. The real work really does begin after release.

“Adam has already started working on stuff for body shots as well, so that should be available as a second update very soon after the first. In general, we are trying to get lots of feedback from players and implement things that
they want as we are able. Ultimately, this all depends on how well the game does and what we can realistically accomplish.

Some stuff on our radar is:

1. Tweaking physics properties to improve the appearance of rag-doll falls
2. Making slow-mo closeup animations more interesting
3. Adding licensed boxers (Chillingo has already done some of the foot work necessary
for this)
4. Adding female boxers (Adam actually really wanted to do this quite some time ago,
so it will almost certainly happen at some point — Million Dollar Baby was a huge
influence for him)
5. Adding more environments
6. Adding training mini games
7. Improving movement around the ring
8. Adding some kind of multiplayer (Matt is really interested in this, but it is also an enormously complex problem for a team of our size, so don’t expect this right away)”

In no small way I would like to give a big thanks to Adam, Matt and Chillingo for participating in this truly in-depth and very informative tech review. Definitely one for me to remember.

Touch KO Game Summary
Developed by: Adam & Matt Mechtley
Published by: Chillingo
Platforms: iPhone and iPod touch
Development time: 6 months
Number of Developers: 2 in-house
Interesting unknown fact: Adam – “One of the characters, Jersey Boy, is actually a friend of mine doing his PhD in political science at Princeton University! He told me he had always wanted to be in a videogame, so I hooked him up.”
Purchase price: $2.99
Purchase link: Touch KO – itunes
Official website: Touch KO – Chillingo

Pages:1 2

Pages: 1 2