Saturday, May 26, 2012

The making of a Portal 2 test chamber

Those with their finger on the pulse of contemporary gaming will be aware that Valve recently released the new user-friendly level editor for Portal 2. Consequently, there are now over 100,000 user-made levels available, all of them uniformly terrible. (DISCLAIMER: exceptions may exist; I have only played a few.) There's a reason for this, of course. It's not that easy to design a good map. There are only about six different things you can include in a Portal 2 puzzle. I haven't done the mathematics on this, but I'm pretty sure there are fewer than one hundred thousand ways to combine these things.

I'm new to mapping. I did once boot up the Hammer map editor, but it's tricky to get to grips with. Certainly there's no point even starting with it unless you know what you are intending to create. And that was just the problem; I didn't have an idea. There's no point haphazardly arranging walls and floors at random in the hope that a level will emerge. The advantage to the new editor is that you can do this - I wouldn't necessarily say you can start with no idea at all, but you can start with a sketchy, half-thought-out idea and refine it as you go along.

To my mind, there are three kinds of levels you can make in Portal:
  • Good levels, where you have to solve them by doing something unexpected.
  • Gimmick levels where it's clear what you have to do, and the designer probably thought it was cool, but it takes at least a dozen tries to actually pull it off.
  • And huge levels where the only challenge is in figuring out what each of the buttons do, and in which order the designer wanted you to press them.
Before continuing I should point out that I'm not casting stones here; I have made levels in all of these categories. (Maybe not the first one.)

Before the editor came out, I was convinced I had only one idea for a level. Since then I have confounded my own expectations by releasing upwards of three levels. Doing this has made me realise that there are two stages in designing a level, and consequently two challenges. The first challenge is having an idea for a level. This was the one I never imagined I would overcome, and I have to give credit to the flexibility of the map editor for putting me in a position where I have been able to come up with almost five level ideas, some of them bordering on reasonable.

The second step/challenge to designing a level, however, is one which I had never anticipated. It turns out that having had the idea, you subsequently have to design the level. This was entirely new territory.

I thought that it might be interesting to detail the procedure by which one of my levels was produced (note that I said might.)

Dual Bridge Deactivation: the making of
(NB: if you have not played this level yet, this post will give away the solution. So maybe you should play it first. Also, most of what's written below is going to seem utterly nonsensical otherwise, so maybe you should definitely do that.)

Wednesday 9th May
On my way to work, I wonder whether I can think of any ways to combine puzzle elements that did not already receive ample coverage in the main single-player game. (My first map was based on a combination of light bridges and repulsion gel.) At random, I select lasers and excursion funnels. How can they be combined? The obvious concept seems to be that of a laser that has to be diverted by moving a reflective cube through the excursion funnel. So I decide to go with this.

Why should the player have to do this instead of diverting the laser through traditional means? After some further thought, I decide: maybe they have to activate two laser receptors in succession, and something stops them from activating the second one once they are part-way through the solution.

By the time I reach the office, I conclude that the player should need to deactivate two light bridges in succession (extra doors being currently unavailable as a test element) so as to drop down and reach the exit at the bottom. Deactivating only one bridge will lead to a middle area which contains only an excursion funnel to get back to the top: thus hiding the solution in plain sight. It seems to me that in order to create an effective puzzle, it should seem like the crucial elements are there for a different reason than they really are.

The laser receptors will be high up in the main chamber. Opposite the first one will be a portalable surface, making it easy to deactivate. Opposite the other will be an alcove with a portalable surface facing the wrong direction, making the player think this is where they need to use the cube.

In a moment of down-time at the office, I sketch out this blueprint on a blank self-adhesive label which I have printed by mistake:
As you can see, the design is basically complete.

Thursday 10th May
The only problem with coming up with this idea on my way to work was that I would subsequently be on the road for the rest of the week, and thus unable to actually make the level. Needless to say this is a little frustrating. I while away some of the time by wondering what problems may emerge in the design. I am unsure of the best way to approach the alcove area opposite the second laser receptor. There is nothing wrong with a little deliberate misdirection, but I tend to think it's not quite right to have an area which exists solely to wrong-foot the player. Perhaps I will put a button up there which the player pushes to release the cube? This would provide a justification for its existence and my conscience would be clear.

I'm also not sure what to put in the area below the first light bridge. If I put a portalable surface there, will that enable players to bypass the puzzle? I can't be bothered to work it out. That is what play-testing is for.

Friday 11th May
I do not get home until the evening, at which point I am required to attend a friend's birthday celebration. This wastes several hours. I am unable to start building the level until late evening.

Upon starting to create the level, I become aware of some unanticipated problems.
  • For the cube to activate the laser receptors while in the funnel, the receptors must be placed in the centre of the tile. However, for the player to activate them by placing the cube on the floor, the receptors must be placed at the bottom edge of the tile. This, I feel, tends to signpost the solution somewhat.
  • When the laser is redirected into the receptor by the cube floating past, it does not deactivate the bridge long enough for the player to fall through it.
I give up on the level as unworkable.

Saturday 12th May
While lying in bed it occurred to me to wonder; what if instead of the player having to fall through the light bridges, it was a cube? Maybe there would be enough time for a smaller object to drop down. I put this idea to the test and it seems to work. However, the level will clearly need to be revamped.

Reluctantly, I am forced to delete the area below the main room which once housed the exit. I try to think of a way that the cube could fall through a separate set of bridges, and in the process somehow cause the original bridges to deactivate for longer, but quickly decide this would be contrived and unworkable. I leave the tiles grey where the hole used to be, in its memory.

The problem now is that the excursion funnel has no obvious reason for existing other than to solve the puzzle - in other words, it is obvious how to solve the puzzle. (Or so I feel.) Initially I put the final button used to open the exit on an inaccessible high-up platform which must be accessed via the funnel, but this feels inelegant. Eventually I hit upon the idea of the funnel forcing the player away from the exit and needing to be switched off. This seems gimmicky enough that players might just take it at face value.

Originally the player would have been unable to do anything after manually deactivating the first bridge, as they would have fallen halfway down. So, I introduce a fizzler to destroy the sphere if certain conditions are not met (oh yeah, it's a sphere now. Otherwise you could just deactivate the funnel using the original cube.)

I still haven't solved the other problem from the original design: the laser obviously cannot be redirected by placing the cube on the ground. To get around this, I come up with what seems to me the perfect finishing touch. I add a second cube, and a button it must be placed on to solve the level. Armed with two cubes, players are free to assume the level might be solved using cube-stacking - the hallmark of hack level designers the world over.

I am pleased with the level of variety introduced by this utterly redundant cube and button. Players are now faced with a plethora of options in terms of placing cubes on buttons, standing on the buttons themselves, placing cubes on cubes... It's not possible in every design, but I think the more non-workable options you give players, the better the challenge.

Sunday 13th May
After some play-testing I publish the map. It's after midnight, so, technically Sunday. An acquaintance has suggested I turn to popular internet web site Reddit for feedback on my maps, so I post this map among others for the perusal of a kind soul offering free play-testing.

Upon going to bed, I suddenly realise that due to the placement of the second fizzler it's now possible to destroy the two cubes. I get back up, add cube dispensers, and republish.* It's important.
Unfortunately this is the only existing screenshot of the original version of the level. It is a bit small.

The following morning I watch the video of the play-test. Aside from the minor niggle that the puzzle was not solved in the intended way, it is a resounding success.

(*I am certain that this happened, but it is not reflected in the level version history displayed in the Steam Workshop. The dispensers are present in the tiny screenshot above, but not in the video. Maybe the version number is not incremented for a very early edit, and it wasn't rolled out until after the video was recorded? Or is my mind going? We will never know.)

Monday 14th May
Nothing happens.

Tuesday 15th May
Away on business again, I decide to solicit more feedback on the map. I am concerned that I may have inadvertently left in more unintended solutions, so I post my link with the comment that there is "only one solution that I am aware of". This is intended as a hook to entice people to play the level, and in no way constitutes arrogance on my part.

Wednesday 16th May
The "Redditors" have done themselves credit by finding several different solutions:
  • Holding the cube up very high to redirect the laser while standing on the button
  • Using a portal near the button to reflect the laser at both receptors
  • Balancing the cube on the edge of the button to reflect the laser upwards
  • Turning the excursion funnel back on itself to reach the exit without switching it off
Suddenly I understand the true value of crowdsourcing. It would have taken me forever to find all these solutions. I vow to reward their ingenuity by changing the level so none of these exploits can ever be used again.

First, I make the room one unit taller and move up the laser stuff so it can't be reached from the ground. This removes the first solution on the list and hopefully the third. I move the second button away from the wall, and make the wall adjacent to the first button unportalable, in the expectation that this will preclude using a portal to redirect the laser from ground level. (These are some really incoherent explanations, aren't they? It's lucky those people took screenshots. It was this one, remember?)

After some further experimentation I find that the exploit is still possible if the wall is within two tiles of the button. So, I add another angled surface on the one remaining wall. This has the bonus consequence of removing the slightly-wrong solution found by the first play-tester. (I would have been fine with leaving it in, but for some reason the bridge doesn't always deactivate for long enough when you're moving the cube vertically. For this reason, it has to go - it doesn't feel as intentional as the correct solution. Because it isn't.)

I add a third angled surface by the ceiling for no reason, other than to make it look like I wasn't just putting them in to stop people from solving the level. And for aesthetic purposes.

It looks a lot nicer now! See how the grey tiles are still there where the hole used to be?

It's still possible to get to the exit using the exploit with the excursion funnel, but I don't mind leaving that in - it's so obviously not the intended solution that it can remain as a trick for speed-runners. Also, I can't think of a workaround that isn't going to really signpost that part of the ceiling as the required place to put the portal.

The level is complete, or at least for the time being. It's impossible to be sure whether someone else might point out something terribly wrong with it further down the line. For now, I am happy with it, and the rave reviews it has subsequently received (average rating: three stars out of five. I think it's weighted so that only maps with lots of reviewers can get a higher score. Yes, it must be that.)



This has been a lot longer than I anticipated, and slightly less interesting. It also suffers from something of a dearth of screenshots, but I did actually anticipate that. I hope it's potentially of some interest to someone somewhere, anyway.

There are exceptions, of course, but I think generally the best levels have a lot of portalable surfaces. I had to block some of the surfaces because they would have made the level easier, but I could have made it very easy by blocking everything except the two portal placements required to solve the puzzle. All solutions are obvious in retrospect; the trick is in trying to make them less obvious the first time round. It's hard to tell whether I have succeeded in this, as I have next to no experience and my maps have only been played by a handful of people.

Oh yeah, that's the other thing. I should point out that you shouldn't listen to my opinions on level design, because I don't know what I am talking about.

0 Comments:

Post a Comment

<< Home