3D grid/voxel placement system with different block sizes

Help and Feedback Scripting Support
#1

I currently have a grid placement system that works with 1x1x1 blocks. I would like to modify it to work with other sizes, for instance I would like to be able to have a 1x2x1 block that functions as a door, or a 2x2x2 armor block. All of these blocks will be rectangular prisms so there is no need to account for odd geometry but I can’t figure out how to make it work.

here is what I have:

local mouseLocation = UserInputService:GetMouseLocation()
local unitRay = camera:ViewportPointToRay(mouseLocation.X, mouseLocation.Y)
local cast = workspace:Raycast(unitRay.Origin, unitRay.Direction * 1000, castParams)
if cast and preview then
	local snappedPosition = BuildFunctions.SnapToGrid(cast.Position)
	local normalCF = CFrame.lookAlong(cast.Position, cast.Normal)
	local relativeSnapped = normalCF:PointToObjectSpace(snappedPosition)
	local xVector = normalCF:VectorToWorldSpace(Vector3.xAxis * -math.sign(relativeSnapped.X))
	local yVector = normalCF:VectorToWorldSpace(Vector3.yAxis * -math.sign(relativeSnapped.Y))

	local cf = CFrame.fromMatrix(snappedPosition, xVector, yVector, cast.Normal)

	preview.Parent = workspace
	preview.Position = cf:PointToWorldSpace(block.Size/2)
end

as for the BuildFunctions.SnapToGrid() function being called, it looks like this (GRID_SIZE is just the number of studs per grid tile, in this case 2):

function BuildFunctions.SnapToGrid(pos)
	return Vector3.new(math.round(pos.X / GRID_SIZE), math.round(pos.Y/ GRID_SIZE), math.round(pos.Z / GRID_SIZE)) * GRID_SIZE
end

this system is pretty basic, all it does is find the nearest grid position to your mouse position and then use the normal of whatever you are hovering on to compensate for the preview of the block’s position so the preview isnt inside of whatever surface you are trying to place on. The issue is making this work for all cuboids, I have absolutely no clue where to even start. Compensating for cuboid size shouldn’t be terribly hard but I’m struggling on a number of things, namely keeping the snapping working (say you have a 1x2x1 cuboid, snapping its position to where a 1x1x1 would normally go will cause a 0.5 GRID_UNIT misalignment), checking to make sure the bigger cuboids don’t overlap with any other blocks, and also figuring out how to align faces.

has anyone done anything like this before? I’d love any advice or help I can get.

#2

Instead of snapping the grid to the constant you could loop through the size of the current block and get the side with the smallest size. You could then use that to snap your grid

I made something like this before but i cant access my computer as of now so i apologize for that :face_holding_back_tears::face_holding_back_tears:

#3

By current block do you mean the 1x1x1 I currently use or the hypothetical blocks I want to place? I still want all the blocks to snap along the GRID_SIZE constant even if they are multiple grid units long, for instance a 2x2x2 I would like to be able to snap along a 1x1x1 grid meaning that the block moves half its size each grid movement, a 3x3x3 would move a third of its size, etc. however I think using grid_size itself for this would be the most efficient as the value shouldn’t need to change from what I understand, correct me if I’m wrong

#4

solved, not gonna go super in depth on how it works because I am too lazy to do a writeup but the general idea is use raycast normals to decide which axis to displace and then combine that with the size and grid snap of the part to get a final position