Advertisement

Game levels: top-down verses bottom-up design strategy

Started by October 13, 2024 04:31 PM
3 comments, last by JoeJ 2 months, 1 week ago

We are developing a first-person shooter example for our engine and are having a lot of trouble with level design. I read a lot of level design guides and articles, and they all say to start with a bubble or other diagram, block out the scene, play test it, and then in a magical final step use unspecified techniques to make it look good. With this approach we end up with large empty rooms, blank walls, and no idea how to populate them, or add additional geometry to make it not look like a block-out. Playtesting this stuff feels like trying to code a game using programmer art.

Now what?

The only approach I have ever had any success with is the complete opposite. Start with the small details you want, find interesting ways to combine them, build one room, then build another, and another, until you have enough rooms, without paying too much attention to the overall layout. Of course this will only work in single-player underground bunker-like levels where you don't have to resolve indoor/outdoor spaces, and you can expand the level in any direction you want, but it produces results. With this approach we can produce visually impressive results with relatively small effort.

What's going on here? I can't be the only person who is experiencing this. Please tell me your thoughts.

10x Faster Performance for VR: www.ultraengine.com

I agree with the graybox approach. Fit all the important bits, make the gameplay fun, make sure strategic areas like cover areas and overlooks and whatever else you need works well. It is much easier to iterate with gray boxes.

Absolutely start dropping in the assets as they are developed, but be mindful that changes now cost a bit more than they did with basic boxes. The more elements are in the world, the more that fundamental changes force re-working and therefore losing old work, doing it all a second time when it could have been done once.

Measure twice, cut once. It applies here, too.

Advertisement

@frob What I am specifically saying is that approach tends to be a road to nowhere for us. The blockout-to-good-looking-scene step never happens, and we just stare at blank abstract geometry with no idea how to fill it in.

10x Faster Performance for VR: www.ultraengine.com

Josh Klint said:
The blockout-to-good-looking-scene step never happens, and we just stare at blank abstract geometry with no idea how to fill it in.

Maybe it does not happen becasue you do not have detailed set decoration assets for the blocked outdoor level?

But you have them for the rotten basement scene. So you could try to block out such indoor scene, meant to represent rotten basements.
And after that you could add the details using those assets.

If the workflow works you could attempt the same for the outdoor level, beginning with deciding which sorts of decoration models would fit the setting. Street lamps, hydrants, curb stones, advertisement signs, posters of propaganda, etc.

Once you have a big library of such assets it should not be so hard to add the missing details. But you need the library.

(Talking about axperience from former artwork. The size of the library was more important than the talent of the artists, actually.
I never did this for a game, but guess it's the same here. Looking up YT videos of people turning grey box levels into final results confirms it.)


This topic is closed to new replies.

Advertisement