Postmortem: Twisted Pixel's Splosion Man
December 2, 2009 Page 3 of 4
What Went Wrong
We knew from the very beginning of development that we would have a co-op multiplayer mode in Splosion Man and that we wanted people to be able to play together online. We didn't have the manpower to add networking support to our engine internally so we decided to contract someone to help us out.
Unfortunately, it took a little longer than we had hoped to find someone to fill this role which left only a short four months to implement our network layer and get the whole game synched up over the network.
Because of the short development cycle, many of the game's features weren't implemented until quite late causing network synching of these features to be pushed back even later.
Late in development, we started realizing that we were using way too much bandwidth and that the game played poorly when packets are lost or when lag is significant. We improved this as much as we could without risking the game's stability in the short amount of time we had left.
Luckily, our contractor was quite skilled and did a fantastic job given the time that he had. Unfortunately, after Splosion Man's release, we received a lot of complaints from players having problems playing over the network.
So, we took it to heart and have spent some time working on a Title Update that immensely improves network performance. The good thing that comes out of all of this is that we learned a lot and have managed to improve our network code for use on future games.
2. An Over-Worked Chainsaw
At Twisted Pixel we work with a very talented audio studio called Gl33k, who has worked on all of our titles. When bringing them on for Splosion Man, we were unsure of how many sound effects the game would need, and how long they would have to create the audio assets for the game.
We originally based our sound estimates on The Maw, but that proved to be a very wrong assumption, as Splosion Man ended up having way more sound effects, including special musical numbers and other prizes than The Maw.
Our "go to" guy at Gl33k was Matt "Chainsaw" Chaney, and he worked extremely hard to make sure all the new things that went into the game everyday received the sound attention that they deserved. Making things even harder on him was the fact that most of the final art and animation assets weren't going into the game until towards the very end, which made it difficult for him to figure out what we wanted and where it should go.
While this is a negative point in our postmortem (bad scheduling on Twisted Pixel's part, no clear path for Gl33k when sound should go in), it luckily became a positive as well. Chainsaw loved the game and loved what he was doing for it, so he put in an insane amount of effort to make sure that the sound design was of a higher caliber than what most people would expect from a $10 downloadable title.
Along with Matt Piersall (President of Gl33k), they did an incredible job of making the game memorable, especially with Chainsaw's own "Donuts Go Nuts" song, which some players may associate with Splosion Man more than the actual gameplay or characters.
With future titles, it will be important for us to not underestimate the amount of planning it takes for an awesome audio implementation. Luckily for us, our friends at Gl33k worked with us to make it possible to ship Splosion Man on time, which I'm not sure anyone else could have pulled off except for them.
3. Too Many Concurrent Projects
Although Twisted Pixel's primary focus is developing original games using our own IP, we also do contract work for other games. During Splosion Man's development, we had more concurrent projects going on than we have had in the past.
It was typical for us to have a small team of programmers working on outside contract work while our main team is working on our own game. However, during Splosion Man we were also finishing up The Maw downloadable content, had another small team working on a port of another game, had a couple of people working on some contract work, and were working with another developer who was porting The Maw for release on Windows.
This stretched those of us who were involved with managing these projects pretty thin, especially given that we were also active key members of the Splosion Man team. There were just too many things to monitor and do with too few people to make sure that it was all running smoothly.
Fortunately, time was able to ease this problem as projects were completed. The Maw DLC was completed fairly early on in development, the port that the small team was working on got cancelled for external reasons, and The Maw port to Windows was completed a few months into the development of Splosion Man.
Thanks to the progressive completion of these projects, we were able to avoid any development delays because of limited manpower. In the future, we need to be more cognizant of the amount of time it will take to manage multiple teams of people on different projects no matter how small the teams are.
Page 3 of 4