How do you build your levels?

So I'm getting pretty close to finishing the 'engine' for my game. All the systems are 99% done, ignoring any actual art or sound effects...

So I've started building out some levels and damn, it's hard. Turns out level design is an entirely separate skill-set that is really hard to fake. So I'm very curious how others here build out there levels.

Do you scribble out a level on pen and paper first? If so, does that not make it hard play-testing what the player might do?

Do you go into the layout editor and throw a bunch of things randomly onto the map, then try and build a level around it?

Do you think of certain challenges you want the player to have, and then carefully plan a level around that?

What about building out the first few levels that slowly introduce the character to the game? Do you leave these for last or do you start on these first?

I'd love to hear some of the processes you all go through.

I always draft out ideas on pen and paper, I let my imagination go wild, and then I circle the good ones.

For puzzle games, I'll create a "difficult area/thing" and then base my level around that sort of difficult thing. Something like a trap in my head.

You have it pretty spot on.

I usually draw everything first.After that i just create a rough stage or level with multi coloured tiles red for dangerous areas and blue for safe areas etc. I only use the assets i made for the game after i tested the rough level for bugs and problems.

I build as I go. Full on. Generally I start with a thematic idea in my head and have some idea what 'shape' I want a section of game to be. A sine wave of rooms going both up and down? A 'stepped' progression? Just going straight up? I try and vary these shapes because that maximizes the effects of spatial memory in the player. Flow, pacing and aesthetics are also almost 'fractal'. So while the whole stage might have a shape, each seciton has a shape and each interaction has it's own pacing curve and blahblahblahblah, you don't actually explicitly DESIGN all this, but you FEEL it to make sure the level feels good and is varied. Since I tend to do rather long 'levels', I find it best to not plan too far ahead. I need to have a good feel for how one section of a level feels to make the next. Things rarely come out how I plan anyways.

Priorities for each game vary but you want to have uniqueness of movement and uniqueness of aesthetic. Aesthetics doesn't mean you have to have new assets in each section, but it means you should try and use the assets you have differently. I find that what I'm ultimately doing a lot is adjusting ratios. If I have a foresty level with some ruins/bricks, some part of the level will be all forest, some of it some brick and one more ruins heavy. Mix that up with things like the direction the player is traveling and a few set peieces, and you can have very memorable stages.

If you're doing longer stages, go look at Castlevania 1 and Castlevania 3 on the NES and look at how much they do with so little. You can head over to vgmaps.com to look at some of the techniques old level designers use.

If you're working smaller, go look at Super Meat Boy, which is surprisingly varied for many of the reasons I describe, despite rarely having unique per-stage assets. If you wanna go per screen, look at my game (I Wanna Be the Guy) or VVVVVVV.

Gameplay is obviously a top concern, but there is no good standard advice I can think of. Conversely, most amatuer games I play suffer from horridly bland level design, even on a movement and visual level. Everything is just samey and bland. Don't be one of those guys.

I draw out all gameplay assets and keep them in view at all times.

Then I sketch out a number of different ideas. Don't be afraid to branch out and experiment when actually building, the paper sketches should just be treated as a starting point.

Only just getting around to responding now, busy evenings.

Rory: I really like the idea of having the "difficult thing" and then planning around that. I think part of my problem is getting overwhelmed of where to start. Is the exit to close? To far away? Is the level to small? Too large? Are there too many traps? And so on and so on. I like the idea of saying "Ok this level will have lasers that will fire in a certain pattern and you need to jump through them" and then build out from there. Get the core aspect of it working and then add to it. I think that will help a lot.

DravenX: Seems quite a few of you all start with pen and paper first. I have a pile of printer paper in front of me right now so I'll start sketching and see where it goes!

kayin: I really like the "try and vary these shapes because that maximizes the effects of spatial memory in the player" note. I think I've found myself constantly battling with this as I tend to pick out boxy repetitive patterns. I'll check out VVVVVV and...wait...you did "I Wanna Be the Guy"?! Nice! Ha, I was already using your game as reference <img src="smileys/smiley4.gif" border="0" align="middle" />

Awesome feedback guys, I really appreciate it.

Boxy can be fine. You can do a lot with a box. U shape and C and C shaped levels, Spirals, circles, Vs and ^s, Each screen of IWBTG and VVVVVVV are boxy. It's what you do with the box that is the question! And yeah IWBTG can be fun to learn some of this too because I think a lot of the graphical stuff is hilariously weak (I was just sorta finding my feet with that sort of thing) and the tiles I made my self are kinda awful, but the aesthetic SHAPE of the screens and world are still pretty good. The only thing it has going for it is screen composition, since the elements within those compositions look like they were made in MS Paint. If you're careful about where you put things, even if you're doing it in a boxy manner, you can get surprisingly good results!

Same here. I just throw boxes around in Construct2 when I have an idea. As an artist, I can't be bothered to draw stuff before I prototype it as I can get into the graphics too much, therefore taking up a lot of my time, before I've even tested if the idea is fun or not!

If it's fun with crude shapes, then it'll be fun with decent graphics.

Workflow is definitely important. If you're basically drawing your levels and filling them with unique assets, you obviously HAVE To prototype. If you're working with tiles? Of course the disadvantage there is an area someone prototypes might not work as well as expected when 'arted' over. If your workflow involves tiles, you can make good looking places and make sure the composition works out while testing other things. Ultimately everyone needs to find their own workflow that works for their project, but I'd say as a rule, try and prototype in a way that both assures you a level will be cool both gameplay wise and visually. Some people can predict what all the collision rectangles will look like when painted over, some can't. Do what makes the most sense for you.

Same here. I just throw boxes around in Construct2 when I have an idea. As an artist, I can't be bothered to draw stuff before I prototype it as I can get into the graphics too much, therefore taking up a lot of my time, before I've even tested if the idea is fun or not!

If it's fun with crude shapes, then it'll be fun with decent graphics.

I pretty much do this. if it's a fun game with solid color blocks, it has a lot more potential to be a great game with better graphics tacked on.

You do have to have a strong minds eye for this though.

My minds eye isn't all that great and never really thought about the connection tbh. Could explain why I do find it hard to resist the temptation of opening up Photoshop. Sometimes I do give in way too early in the prototype though.

Glad you brought that up. Now that I'm aware of it, maybe I can control my urges..

Planning, then testing.

Followed by more planning, and retesting.

I've played enough indie games by developers who decided to just sort of throw things together, and it's not fun in the least. Bloated levels, levels with nonsensical progression, levels that are obviously debug test-room levels, etc. etc. They're all sh*t.

Start making the best, hardest stage you can think of, and start building levels around that.

Level design without the "design" is just pointless.

I do a rough sketch on paper (just boxes and lines) to plan out the progression, obstacles, flow, and gameplay theme. Then I often do a rough mockup in-game (boxes or basic wall elements) and then I build from the ground up. If its a platformer, that means all of the ground before decorations--and I spend a good deal of effort making sure parallax layers really add depth while maintaining gameplay. For top-down games, it means all of the ground (and return later for more detail), then walls, then decorations. I've seen plenty of people get so tied up in adding details that they forget to make the level fun/useful. Graphical flare is extremely important to me, but only for selling the gameplay or game concept idea to the player.

Depending on what the levels assets are (I usually do this with tiles) I'll draw a bunch of tiles and cut them out and then place them around pieces of paper.