4. Dropping the ball at the last moment
When your game is divided into separate levels, in what order should they be created? We didn't struggle with this question -- the tutorial was made half-way through the project so that it would reflect the game reasonably well, and the last level and the end boss were left until the very end of the project. This was improvement over the Shadowgrounds games, where the development followed the games' level progression.
Naturally, when the time came to create the last level for Trine, we only had a couple of weeks until the game was supposed to be finished. Therefore all of our earlier thoughts on having a huge boss enemy were thrown out the window -- bosses take considerable time and we didn't have much.
An alternative plan B was schemed on the spot -- what if the player travelled upwards, avoiding rising lava? Several elements were added, such as an evil wizard (or the "old King" for those who tried to follow the story) who conjured objects such as boxes, planks, and spiked balls to hinder the player's progress.
The last level was to be the ultimate test of the players' abilities and how well they had understood the cooperation between the characters.
Everyone in the team had fun with the level. It was a considerable challenge and took many tries to get to the top. At the top a small regular fight awaited, and the game would end when the player reached the topmost platform. The feeling of accomplishment when the level was finally beaten was great, given the schedule constraints, and we were happy with the work-in-progress final cutscene -- the only one that uses the in-game engine and characters.
As time was running out, the level did not get nearly as much playtesting as e.g. the tutorial had got. We had made several changes to the tutorial -- for example, a few weeks before gold master we changed the Wizard's English tutorial hint message from "Draw a box" into "Draw a square" as we found out that some players started looking for a box to drag around instead of using the cursor to conjure a box out of thin air.
We knew the last level was a departure from the rest of the game, but we felt it was a nice change of pace. We had one outsider, an ex-employee, who tested the level before gold master. He did not enjoy it at all, but we ignored his feedback. He had only played a few levels of the game before so his opinion clearly wasn't valid.
We've never got so much negative feedback on a single level before. The majority of gamers did not enjoy the change -- the somewhat slow-paced gameplay had been transformed into a quick test of skill. There were some who enjoyed the level, but they were a minority. It could be speculated that Trine had reached out to a much larger target audience than we had perhaps envisioned, and that resulted in the clash. Many gamers also lamented the fact that there was no real boss fight.
In a way most of the decisions we made with the last level were perfectly sound -- they just belonged to a different game, such as the one Trine was originally supposed to be. Some parts of the level were (and still are) impossible to get past on the first try, as some of the triggers are simply unfair and do not give enough time for the player to react. This kind of memorization gameplay, in addition to hectic speed of the level, did not fit with Trine.
We made the speed of the rising lava slower with patches and removed it from the easiest difficulty. However, this did not help as much as we had thought -- while the Game Over screen featured a difficulty change option, many players did not take advantage of it and instead banged their heads against the spiked balls, and became frustrated.
As a result we slowed down the speed of the lava even more with further patches. While we are not believers in automatic difficulty adjustments, we will most likely consider some alternative approaches to difficulty levels in the future. Even something as simple as detecting player deaths and then changing the difficulty accordingly would have been much better and guaranteed a better experience for many gamers.
It's also worth noting that to this day, many gamers still think the last level of the game is too hard, because their own experience does not reflect the game in the post-patches stage. Some in fact think we made it harder, because we added a few skeletons into the mix to spice it up, and this is easier to notice than the adjusted speed of the lava.
This kind of inaccurate word of mouth does not change easily, and further emphasizes how important it is to get things right before launch. (And of course, making matters worse as of this writing is that the North American PSN version still features the original, tough-as-nails last level, although there are finally plans to have the patch released.)
In the future we'll definitely want to have the last levels made a bit earlier in the production and modify the difficulty afterwards, instead of pushing it to the end of the project. We will also focus more on bosses and such memorable events -- gamers and reviewers seem to have an innate need for "bosses", at least in certain game genres, and it would help to fulfill those expectations.
5. Leaving online multiplayer out of the game
Trine, like our earlier games, features an offline co-op mode for up to three players and it is considered one of the best aspects of the game. It changes the gameplay drastically, especially in three player co-op, where players truly have to work together -- and often a certain amount of hilarious mishaps are bound to happen.
So why do we get hundreds of negative emails and forum posts about co-op? Because it's not online. Gamers these days expect that co-op means "online" by default, and are sorely disappointed when they find out that Trine does not have any online multiplayer features.
Not having online multiplayer has been the biggest criticism against our games since day one, and it all goes back to 2002 when we were working on a realtime strategy game concept and made a decision not to pursue online/LAN features in favor of faster development times.
The same code base has been used in all of our games since, including Trine. As layers and layers of new code have been added to the engine, the cost estimate of adding online multiplayer has always increased with each project, and it became very hard to justify the additional cost of rewriting the game engine -- or rather, getting enough money to complete both the rewrite and whatever game we had in production.
While this decision has been a key factor in our ability to ship any games at all, it is hard not to dream of a time machine. A little bit more forward-planning in 2002 would have gone a long way. It's easy to fall into the "oh it's just a prototype, it doesn't have to be good" mentality, but if things don't go to plan and the project doesn't bring in the money everybody is expecting, the prototype will end up being used in more projects because it's cheaper and faster.
Trine probably could have doubled its sales and success had online co-op been included.
Trine as a game turned out great. Trine as a project was a big mess -- a series of unfortunate events that ended up just shy of miserable failure. But we got the game done and at the end of the day it's the game that counts. Many of the business issues and other hardships we went through during the project did not affect the final game at all. We were -- and still are -- very proud of Trine and what we accomplished.
And it's not just us. Trine got good review scores (Metacritic 80 and 83) and great feedback from gamers. Many have even said said that Trine is the best game they have played in a long time and it refueled their passion for games. It doesn't get much better than that.
Bittersweet as it is to think about all the lost potential, there is no denying that Trine is the project that has catapulted Frozenbyte into a new stratosphere, and the great player feedback and reasonably good sales remind us of this. With more experience under our belt and a full code rewrite in progress, the next project will go without a hitch and right all wrongs. Right?
Publishers: Frozenbyte (PC online) / Nobilis Games (PSN EU & US, PC EU) / SouthPeak Games (PC US) / Square Enix Downloadable Games (PSN JPN)
Release Dates: July 2, 2009 (PC online), July-August, 2009 (PC retail EU), September 10, 2009 (PC retail US), September 17, 2009 (PSN EU), October 22, 2009 (PSN US), March 31, 2010 (PSN JPN)
Platforms: PC online, PC retail, PlayStation Network
Number of Developers: 3 at start, 16 at peak/end
Length of Development: 18 months
Budget: 850,000 euros
Lines of Code (Game): 298,000
Lines of Code (Engine and earlier games): 426,000
Main Development Tools: MS Visual Studio 2005, PhotoShop CS2, LightWave 9, Hansoft
Lines of code in the shipped game that include comment lines with "hack" or "todo": 1438
Outsiders who tested the last level before release: 1