Keep in mind that I am not as experienced (by a long shot) as some of the other people in this thread, but this kind of brings a different perspective on the question as well, which might be valuable to some.
First off, despite using ZC on and off since 2007, I never did real, substantive dungeon design until the end of last year extending into this year. I'm very comfortable with OW screens, with their organic shapes and kind of loose functioning - and I think a lot of people have the same issue I have, which is turning that energy into dungeons. But I kind of learned to really enjoy dungeon designing with Crucible Crest and [Redacted], so I wanted to share some things not strictly about how to design dungeons (there are loads of good resources and thoughts about this already) but about how to approach designing dungeons.
Overplan - this is the single hardest thing to transition from if you're stuck in OW-brain. When I do an OW screen, I usually leave the screen connections and details of the screen to however I feel they should go in the moment as long as it matches up to screens I've already designed. Dungeon screens can't be approached like this - you really need to have a concept for the room, what it does, and how it fits with the rest of the dungeon before you start. I did a lot of this page in my notebook before I even started Level 1 in Crest:
You can see that I've identified where most key dungeon items are, where miniboss/boss are, and where locked doors are. I've also sketched out general outlines for each room and made notes about their function and puzzle components. It's not noted here, but I also knew in my mind what the general enemy palette would be for the dungeon.
It doesn't have to be 100%, and some of these screens did change between plan and final - but this really helped firm up the dungeon flow. Some people don't like to plan because it "interrupts the design flow" but actually, having the plan before you start helps the design flow since you don't get "stuck" deciding what a room should do. I found that I always had a starting point for each screen and even if I deviated in little ways, it was way easier to push through and complete the design - whereas in the past when I tried dungeons, I end up getting blocked because I have no idea what to do next or worse find out I need to change screens I've already completed since I designed myself into a corner.
Playtest Early and Often - This is something I picked up from doing some Doom levels over last summer. I playtest almost every single room in a dungeon individually as I build them. I also tend not to move the player spawn point too much if I can help it, so I end up doing repeat testing of a lot of rooms as a block. When John Romero playtested his own Doom levels, he always tested from the start of his level - even if the thing he changed or added was way later in the level. I find this a bit extreme for my own workflow (Romero worked really damn fast) but I try to do a "lite" version of it. I find that this helps in a couple of ways:
- You internalize the puzzle/combat/layout elements in a much deeper way when you do this, and it helps you make decisions on what works and what doesn't, where your pressure points are in terms of flow and pacing, and can help you eliminate redundant elements.
- You experience the range of RNG that your combat has. This can help you refine enemy spawn locations and patterns.
- You figure out the design quirks of things as they appear and can exploit this in future rooms. For example, by refining the initial sleep moth encounter in Crest and playtesting this dozens of times as I tweaked the script, I learned certain things about their movement that allowed me to make better layouts with them in future screens. I couldn't have done that if I had built all of the screens up first.
- Sometimes it feels like setting up undercombos, secrets, enemies, puzzles, etc. is a bit like "eating your vegetables". You know you have to do it, but you'd rather just stick to the steak. If you don't playtest as you go, you can end up with a pound of green beans to eat at the end - and this can really kill momentum on a project. Playtesting helps the digestion and when you get to the last screens, you really are almost done, which definitely helps mentally.
Pots for instance affect pacing. They give the player an opportunity to refill/restock items (or grind if they're desperate) but also the opportunity to full heal off them cycling the screen means the player will move past those groupings of screens slower, especially if it's deeper into the dungeon where health is (relatively) more valuable.than earlier on.
I also want to add that this is something that is very easy to tweak in 2.55 with modifiable item drop sets. For instance, the bushes before the mini-boss and boss in Crucible Crest level 1 are guaranteed to drop arrow ammunition and hearts, which
- constrains the difficulty experience a bit since it fits the design goals of level 1
- also means that players aren't forced to "grind" out screens to prepare for these fights, which helps the pacing
I have seen a few pre-2.55 quests try to fix this by just putting a ton of pots/bushes in a room, but this has a pacing impact and constrains the room design of these spaces.