What Went Wrong
1: AGS
Since 2009, AGS has been my engine of choice. Even for non-adventure games, I've enjoyed stretching its capabilities to produce turn-based strategy and arcade games. While it's even better for making adventures, the challenges that it created while making
Technobabylon have started to outweigh the benefits of its dedicated adventure-creation for me. If you're looking to start somewhere, I'd still wholeheartedly recommend AGS, but personally I'm going to be making the move to Unity or Unreal.
At first, using an adventure engine to make an adventure game seems perfectly logical. But problems began to manifest as the development process went on. It isn't a system that likes to share – there isn't a way to split development of different assets and modules across a team, beyond a separation into “art”, “audio” and “everything else”. Saved games aren’t backwards compatible, which makes patches nearly impossible to implement. While some may see this as a boon (after all, how would
Assassin's Creed Unity have been if they'd had to get everything correct out of the door?), it means that, with the exception of major game-breaking issues, we are unable to update the game following its release. This also prevents us from releasing extra language updates in the future.
The third major issue with AGS, and the one that is chiefly driving me to change, is compatibility. The games will run in Windows, and by default, that's just about it. Some tweaking can make them Mac and Linux compatible, but this takes specialized engineering by Wadjet Eye's staff. Even when the games work in Windows, there are often compatibility issues with graphics. AGS has been a useful place to start, but the number of difficulties it's added to this production have encouraged me to move on.
2: Stop-Start-Stop-Start
Technobabylon began its development in 2010, had three episodes, and then kind of stalled for about four years. Several attempts to get Part 4 off the ground were not satisfactory, partly because of the game's original intention. Starting as a practice project, I'd get half-way through Part 4, then find that my own art style had improved (or simply changed), meaning that characters and scenery no longer resembled their counterparts from the beginning of the episode's production. Deciding to apply this newer style across the board, I set out to remake the earlier sections of the game, but each time had the same result. Because of this, there are roughly five different versions of the first section of
Technobabylon, almost word-for-word the same, but differing in presentation.
Unsatisfied with each effort, I would keep clearing out and beginning again, a cycle which continued until 2014, when I finally decided to leave the art in the hands of someone more competent than myself. Unfortunately, my previous turnover of a new section every few months had been completely scuppered by these delays. For me, being able to complete a project in sections is a tremendous aid in getting through something as large as this. By reaching these tangible goals, it feels less like a marathon, and more like a series of shorter, more manageable races.
Wholly rebuilding the first sections not only delayed getting the game itself finished, but created a feeling of going around in circles, creating lingering doubt as to whether I'd ever be able to complete it. Doubts would always pop up, and it was very difficult to stay on track while I was handling the art assets as well. In the same sense that different people have different preferred styles of learning, I think that there's a lot to be said for finding a work approach that is effective for individuals – I need to identify a way to stay on-track for any future projects, rather than simply relying on the skill of others to fill the gaps that are causing me to fluster.
3: Bugs in the review build
As mentioned earlier, games built in AGS are not easy to fix once released, as saved games are rendered incompatible. Towards the end of development, as with other games, we had a few months of “crunch time”, where we'd work through the game with testers to root out and destroy any bugs in the game. The danger with this is that you keep on finding things to fix, just a little tweak here and there, and it ends up becoming a cycle of tiny changes (some that cause new problems themselves) that never ends.
Eventually of course, we'd have to release something to show to the press for reviews before the game itself came out, and we did. We'd made sure that the game was playable from start to finish, and sent it out to the reviewers – this of course was when we started getting reports of a couple more game-breaking bugs that the testers had found since the review build went out. A couple of them were fairly egregious, the kind of thing I feel like I should have spotted months earlier in the development process.
My first hope was that reviewers may simply miss them – if they were following the walkthrough provided, they might never find the errors. Obviously I'd underestimated them, as Wadjet Eye's public relations soon started to receive e-mails reporting these problems.
Dave assured me that reviewers would understand that this was an unfinished version, and that bugs were to be expected (as long as we'd squashed them by the final release). Based on the
reviews that the game eventually received, this seems to have been the case. Part of me feels that we could have made a better impression with a couple more weeks of testing, and I would kick myself when one of those reports came in. These things have to be released eventually though, and at least I can say we got through it in the end. There are still a few issues here and there, but nothing serious.
4: The Ending
As I planned
Technobabylon, I wanted to give it a suitably grand ending, with the concept of the human mind fusing with an all-knowing artificial intelligence. Then, I remembered that this was in fact the ending to
Deus Ex, and that I'd have to come up with something else.
I'd spent several years planning what was going into the game, but realised that I'd put remarkably little thought into the ending. The downside with adventure games is that, unlike many other genres, one can't finish with a boss-battle. The changes I made to it also prevented me from taking the player's actions throughout the game into account to anywhere near the extent I would have liked. I wanted multiple endings and I detest the
Mass Effect 3 style of “Press button A for red explosion, button B for blue explosion” way to end a game, but unfortunately a large part of
Technobabylon's ending ended up being dominated by a binary choice towards the end of the game. I aimed to ameliorate this by giving more detail and difference to the endings than the colour scheme, but I knew that it wasn't going to please everyone, and I'm sure I could have done it much better.
The other issue we had was that the endings (and the part leading up to them) didn't get nearly as much testing as they should have. The continued problem of saves breaking meant that, with each update to the game's beta (weekly), the testers would have to restart the game from the beginning. As a result, we got a lot of feedback about the first half of the game, but not a great deal for the latter. A mad rush of testing towards the end of production led to fixes for the most serious issues again, but I still feel that I ought to have given the last section more work.
5: The Style
This is a controversial one – I like the style, and our artists (Ben Chandler and Ivan Ulyanov, who dealt with the character portraits) have done tremendous work. '90s-style pixel art exactly captured the feel of the game I was looking for, evoking the memory of games like
Beneath a Steel Sky. It's straightforward to make, not resource-intensive, and a signature style of Wadjet Eye games today. However, if I could send e-mails backwards in time, I would advise myself to consider other options – maybe something higher-resolution, or potentially 3D.
While I and other players of indie games like
Technobabylon have a fondness for the style, one of the most frequent comments I see from casual observers on the internet is “ugh, those graphics!” Many players today are unable to parse the blocky style, and perceive the game as somehow cheaper or inferior based upon its style. While I personally disagree with them, I can't help but wonder if sticking to the style has potentially cost us sales, and am endeavoring to explore other options for future games.
Conclusion
Making
Technobabylon has been a fantastic experience. It started as a learning experience, and in many ways has continued to be so as I've gotten to grips with the realities of making a full-length game. There are certainly a few things which, in hindsight, could have been done differently, but these now become sources of experience from which to make the next game even better.
As to where we go next, there are all kinds of options. Maybe another adventure, maybe something different, perhaps even a
Technobabylon sequel since we've had people asking about it, and others suggesting I get into VR with Oculus Rift. I’ve got to start experimenting with Unity though, so the chances are that the next game you see from me will be another practice project – but as we've seen, who knows where that could go?