Conversation

Replying to

@Johnee_B

and

@mmalex

This is working off the bounding box idea. But at some point a smaller box produces more cost?

0:46

57 views

Replying to

and

I think that's because the bounding box is dynamically resizing depending what's inside of it, but also trying to keep the fleck resolution consistent, so as it pops between different size boxes you get the changes you see.

Replying to

and

it is possible that at some point it reflects back on itself and creates a bigger box? Sort of like how if you try to compress one dimension too small that dimension goes completely flat?

Replying to

and

Hmmm, maybe, but I'm not the expert on this exact area as it's pretty in depth

@sjb3d

is the pro here as he coded it :)

Replying to

and

But the bounding box is the general principle, correct? I saw this result and it conflicted, so I wasn't sure if I'd misinterpreted something or what.

Replying to

and

The bounding box controls the underlying voxel/sdf res, a cube of say 10 X 10 X 10 etc. Which is aligned to the sculpt's grid, which is combined with the fleck density on the outside to create the cost of the object/

Replying to

And the bounding box jumps up and down by powers of 2 I believe. Pro tip is use the cut out tool, aligned to the local grid of the sculpt, using a cube shape to help optimise the sculpt to the max (technically just lets you add more the max tightness possible)

Replying to

and

makes sense then if you go below min grid size you get an unexpected result. min grid is 2^-5, something half that size would force to 2^-6, which doesnt exist, so maybe you get 2^-4 instead.

Replying to

and

Exactly!

Replying to

and

I'd always THOUGHT they conformed internally to grid based on experience from sculpting, but was never sure.

Replying to

and

Yep graphics cost should be roughly the number of flecks, but there are some rounding effects that can move that number around a bit. Grid size snaps to the nearest power of approx 1.19 (4 levels for each power of 2).

7:53 PM · Feb 29, 2020Twitter for Android

Replying to

and

Heres the experiment I've run: I lay down a 2 dot cube in standard grid. Then I go to successively smaller grid sizes and peel that amount off each dimension with stretch. 1/8 grid size is the most expensive.(15/8x15/8x15/8) Would the rounding be the thing accounting for that?

Replying to

and

Maybe! Grid-aligned cubes are kind of a worst case, I'd expect results that more closely follow the number of flecks for a sphere (or a sculpt with a more complex shape).