No Level Editor
For all of the big commercial games I worked on, including Railroad Tycoon 2 and 3 and Tropico, we created elaborate level editors. So good and polished, in fact, that we shipped them with the game.
It was, for the most part, a waste of time. It took a lot of effort to polish those editors to shippable quality. Very few users spent much time creating maps – perhaps 0.1% of the people who bought the game ever created a semi-polished map. A slightly larger percentage, perhaps 5%, ever downloaded user-made maps. But the programming effort spent on the level editor could have gone to the game itself.
As I was thinking out level designs today, I debated making a simple level editor for my casual game. But I know if I started with a simple editor, I’d add bells and whistles, and try to get it to be good enough for end users, leading to a whole new set of support issues.
In the end, I decided to just edit a big header file called LevelDesigns.h. Which, incedentally, is exactly what I did for Bonnie’s Bookstore. It took an hour to get code support for the features I wanted, versus days if I’d developed a bona-fide editor.
I know an editor is necessary for some types of games – certainly for more advanced commercial games. But I think in general, programmers should resist the urge to create uber-editors, and focus on their game instead.
March 21st, 2006 at 9:32 pm
A level-editor is more of a “hobbyist” thing to implement. It is targetting niche and creating a game with a cetain degree of greater depth that doesn’t have much breadth.
Generally level-generators have a lot of utility as development tools whether you release them or not but that utility depends a lot on the game in question. One argument for the level editor is that freedom and ease of developing levels may allow you to do a much better job of the actual development of levels. If you are just tweaking your level data in arrays you may not “see” some interesting designs. That value of course increases as the game/level complexity increases. For a lot of casual games it’s probably overkill. For a 3-d shooter like Half Life of course these level-editors are an absolute must whether they get released to the end user or not.
Also, I think that you can get some mileage out of in-between solutions. One solution, for example, is to store their levels as .XML files and include a short document with the game that details how one might change that .XML file to create their own level sets. Using XML (or some other standardized) format isn’t that bad of an idea to begin with from the development side of things. Sure, this makes your level creation tool even less accessible. So maybe .05% instead of .1% use it. A large portion of the benefit is just in advertising the thing. Users who never use the feature may none-the-less be more likely to by the game thinking that they might.
March 22nd, 2006 at 3:26 am
I saw an interesting approach used in Sprout, a 48-hour game coding contest entry. The author uses .BMPs as level files. The colour of each pixel denotes the kind of wall or item that the cell contains.
Naturally, he didn’t bother writing a BMP loading routine, he just asked SDL to dump the surface data. Pretty easy, and if you’re doing 2D tiled graphics then I bet even MS Paint has a better flood-fill function than your text editor!
March 22nd, 2006 at 11:57 am
Do you know what impact, if any, the level editor had on sales? I have always assumed that putting “Level editor included” on the side of the box increased sales, even though very few people would actually use it. But even if true, I’ve wondered if the impact was worth the cost.
March 22nd, 2006 at 12:25 pm
In general, you can never really evaluate the isolated impact of a single game feature on overall sales – it’s just guesswork.
Railroad Tycoon 2 got the most benefit from it’s editor – it had the largest editing community, and we even used about 40 user maps in the Railroad Tycoon 2 Platinum compilation (we paid the map authors a small amount).
If I had to make a totally wild guess, I’d say that RT2’s map editor added 3-8% in sales, and for Tropico and RT3 it was less than 2%.
March 22nd, 2006 at 4:08 pm
When I figured out the specs of different files in RT3, I asked Phil if it was ok to create a “building editor” myself. Phil said I could go ahead, but didn’t believe I’d ever finish it.
He was right 🙂
Even if 20 people on some RT3 forums would ask me to hex-edit some buildings, that’s still a lot less time invested than finishing an editor. The modding-community for RT3 is only now starting to take off with people adding custom (hex-edited) locomotives and buildings for their scenario’s. Maybe someone will write a hex-to-xml and xml-to-hex program so people who don’t like hexeditors can edit hapily too. No need for a “real” editor here.
March 24th, 2006 at 2:05 am
Our players *wanted* the level editor (it was for GEOM). The editor was very simple & easy to use (‘casual level editor’ 😉 – that was well taken.
“But I think in general, programmers should resist the urge to create uber-editors, and focus on their game instead.”
Well said.