CODEWITHSUNDEEP
swiftscenekitscnnodescnscenescnvector3
7ucci
7ucci

Reputation: 1

Finding a better method to identify points exist "inside" of a cube in swift

I want to find a way to check if points(vectors) in my scene are contained within a SCNBox I have displayed on screen. Currently I have an array of about 83000 SCNVector3's. So far, I do this by simply running a for loop on each point and checking against the SCNBox bounding box. If it falls within that bounding box I save the point to a separate array. My goal however is not 1 bounding box. I further subdivide the bounding box into equal sections. For each of those I then have to check each individual point again to see if they fall into each one of those individual bounding boxes. If they do, I save those boxes so that when I go to subdivide them again, I am not subdividing boxes that contain no points. This helps performance a little bit as large sections of boxes that don't have points are not needlessly checked. This works okay for a small amount of boxes however I need to subdivide the boxes into much larger amounts, sometimes in the hundreds to thousands of boxes. As you can imagine, it takes a long time to check all the boxes. Currently I am having to iterate through all the points every time for each box. Is there a faster approach to this?

Upvotes: 0

Views: 59

Answers (1)

Voltan
Voltan

Reputation: 1213

Your question says is there a better way, so using the physics engine would be a more 'standard' approach IMO, but performance wise I wouldn't know without trying it myself. It seems you either check manually, or let the physics engine check it for you.

Try this post - it's similar and there are some examples. [67575481].

The physics engine checks nodes for nodes and it may have changed since I did this, but I had to generate node names for my nodes and check it that way. There may be other options now.

My example is a moving 'missile' that collides with a moving node containing 'some monster', so I wasn't dealing with the size you are.

83000 vectors, dunno, that seems like a lot. You'll still have to do the checks you do now, just differently.

Hope that helps...

Upvotes: 0

Related Questions