I’m looking for thoughts regarding building kit tile sizes for a largely city-based scenario.The scenario I’m considering is either isometric or 3rd person view without rapid platforming movement, present day. (well, zomboid would be a good example), where assets are likely placed on a grid.
Basically, it goes like this…
As far as I can tell… A single humanoid character takes 1x1 meter square to stand on. The distance between floors is normally 3 meters, and from floor to ceiling 2.5 meters. (0.5 meter floor thickness). That gives vertical height in threes, which is quite odd.
So, from here we could use a 1x1x3 tiles, which will be a bit odd, as there will be a lot of them, meaning small cells and potentially a lot of geometry.
In practice, it seems that corridor size is about 2.5 meters, so it makes sense to try 2.5x2.5 meter placeable cell, which is also a bit odd, as it doesn’t map properly on underlying 1x1 grid.
In other games I saw following approaches:
2.5 meter cells with 0.5 meter grid (space engineers), 10x10x3 meters cell (one cell → one room with furnishing and all), free placement with snapping (conan exiles, and crafting openworld games) and so on. Then there’s dwarf fortress approach with “one tile → one creature”.
You are overthinking it, I say that as someone who already overthink that exact same question. The real question is how does it feel visually and how well it serves the gameplay, so you should think in term of gameplay metrics.
Unless you are doing VR, real proportion is not what you want, you want thing that don’t look off and still allow proper game flow, even in AAA games where realism seems like a goal, most proportion are exaggerated when needed, like wider road. Heck even cinema cheat with proportion!
So do some mock up, have a generic character with the intended proportion, build scene to scale around him, define the proper grid from then. Personally I use 2x2m grid as a basis but I do space on 4x4m.
Is there any particular reason that your graphics need to be made as single cells?
Even if the answer is yes, consider that you can optimise both workflow and performance afterwards anyway. For instance, you can have prefabs which populate a whole bunch of cells at once with common configurations, and you can also have a tool which optimises geometry once you’re happy with your layout.
Cell/grid based approach allows great deal of freedom if you want procedurally generated content, dynamically generated content, or user-generated content.
For examples, see space engineers, dwarf fortress, minecraft, zomboid and so on.
This is a 3d grid:
I’m not entirely sure what counts as odd or not, but I think it really depends on the game. The way you size the cells will partly determine how things appear as well as the experience that players have playing your game.
Larger grid size will make things appear more blocky and simple, which can be compensated by adding small ‘hard-coded’ details to each item, and smaller grid sizes will make things appear to have a ‘higher fidelity’ but possibly make things more time-consuming to build. A sort of ‘prefab’ system might help bridge the gap between large and small details.
Perhaps there is a sort of universal balance point somewhere, but if so I wouldn’t know where it is.
“Odd” means that grid size is not uniform in all dimensions and does not map to 1 meter grid.
At the moment I’m starting to think that the right idea might be actually using a small cubic grid cell → 0.5 or even 0.25 meters, but allow placeables that take multiple cubes. This would allow 1x1x3 cells, 2.5x2.5x3 cells and pretty much anything else. And additionally things like walls can be represented by grid itself rather than placeables.
I already mentioned space engineers. Those guys screwed up with those. They have two grid sizes - 2.5x2.5x2.5 block and 0.5x0.5x0.5 block. They screwed up because those cannot be mixed, and connecting small grid to large grid requires voodoo. Another game - Medieval Engineers had ability to mix blocks, but they got absorbed by developers of space engineers, and the project has been pretty much abandoned since. That’s one of the reasons for thinking about small grid with large placeables…
Ah you mean in those games the object must be fully contained in the grid cell size? I thought it was just like a snapping feature. Haven’t really played those games…
If so that sounds much harder to figure out. I would just make the placeables any size as long as they don’t overlap any filled cells.
My procedural terrain is .96m x .96m x .08m
So 36" x 36" x 3" high When I do the side pieces I can use 3,6,9,12,24 and 36" high rectangles.
My guys are 5 foot tall.
The buildings will use an entire space for the wall, but the wall model will be 6, 12, 24, 36" wide.
So a building corner would have an “L” shape.
So a house would be 6" wide wall centered, with room for a bookcase in the remaining floor space.
A 36" wide wall would be a concrete bunker.
Floor to floor is 9 feet with a 12" floor. My game is top down view like Divinity so the floors come off.
I havent done the buildings yet but that is my plan for now anyway.
This is a view of the entire planet, pay no attention to that giant guy. I just use him so I know where world 0,0 is.
Plastic style Lego-ish
Edit:
I suppose I should mention I’m planning on the buildings to be sitting on the terrain as a separate system. Anything like rocks and trees are landscape items that sit on the terrain.
I use quad trees for everything. Yes, I’m probably doing it wrong
A part of what I was trying to ask is, is there any reason not to have a relatively small grid size and then allow items to be more than one cell in size?
If those reasons are “too much stuff to place” and “lots of separate objects” then both are solvable afterwards.
It is actually “bulkier data structure with higher memory usage”. It will also be less convenient to use if player is allowed to edit it…
Basically, as you know in situation when an object squarely fit into one cell, you have very simple data layout, even though you’ll still need some sort of sparse octree if the map is big.
When you, say, place a fridge into a grid that has 1 millimeter cell, then you’re no longer dealing with a grid, but with some sort of optimized voxel-like collision optimization structure, and cost of inserting such object would be quite high, as you’ll have to update a lot of cells and optimize them.
Personally, I’m looking for something that woudl allow me to have something like an isometric cityscape with interiors, without going bonkers with millimeter placement. Like I said, the right idea would likely be 0.5 or 0.25 millimeter cell.
