The limit for a single sculpty?

I have made one that has 90 solid pieces. Can I claim the record for now?
Picture at
No doubt this is similar to the 80, but it exploits the fact that only upward diagonals are connected to make a "unit cell" of only 10 vertices. Hard to fit together, and there should be scope for fitting in even more. This used a 32x32 bitmap with torus topology, which allows unit cells to wrap round. I think that may be more efficient than using the extra 33rd rows, at least in some cases. The unit cell here is this shape. Others are possible. Points 1 and 2 are on the lines and the 0s are the two other vertices of the tetrahedra. There are two lines sharing one unused point that fills in the spare 124 vertices!

......1 1
...2 0 1
2 0 1
2 2

I filled the values in by hand and made the bitmap using R (http://www.r-project.org).
So I must be mad.
PS. For now, this can be inspected at Parambikulam 115,205,91
PPS. It's gone now, supplanted by higher numbers (see below).

I think the limit is 1024 vertices, according to something I saw in the wiki about sculpted prims.

From the wiki faq:

Cool! I think you have the record. Your approach is actually quite different than mine; don't know about Cel's.

Congratulations!

I found a better tesselation of the same cell. Now got 96 tetrahedra, although they are all in one line, and even more useless. Only 64 wasted vertices left! It can be found in the same place, Parambikulam 113,207,91 ... now taken away, as there are biggger ones (see below)

Well, I haven't actually constructed it. But applying your idea to the 33x33 vertex plane, the straightforward tiling allows for 99 units. (Unless I counted wrong.)

Well! You provoked me into re-evaluating the planar topology. Using my original tile, I could only get 90, as the lack of wrapping left large blocks of unused pixels. But then I suddenly realised that you can use incomplete tiles at the edges, as long as they have all four required vertices. That means that 12 more could be fitted in the empty blocks! I can't wait to make it. Have we really cracked the 100 mark? I have drawn the plan, but it will have to wait till tomorrow. It's 3.30 am and I need sleep!

PS. Maybe the other tiles will give even more?
PPS. Can't sleep. I think the simpler tile may do 106!

Yes, with your new observation, I can add a column of 7 to my tiling and get 106.

There is room for a "shaved" column on each side, for a total of 17 columns. The number of unit cells in each column repeates the sequence 7 7 6 7 6. So by putting a shaved 7 on the each side, we get

7 7 6 7 6 7 7 6 7 6 7 7 6 7 6 7 7


or 113.

Now I am really tired! I made the bitmaps for the 102 (my tile) and the 106 (4x3 tile with ul and br corners out). BUT I had terrible problems getting the normals right for some of the partial edge tiles. I managed with the 102, but could not find the answer for the 106. This may be because the 102 tiles only have two vertices trimmed, while the 106 tiles (the ones that don't work, that is) get three trimmed. Maybe it's impossible? Si I reverted the 106 to it's parental 99 by removing the offending tiles. Both can now be seen in the garden north of my fractal gallery, Tiretta 87,140,51. I am satisfied having got past 100. So I will turn to more useful activities!

Omei - I am not certain from your messages whether you have made any of these? I guess your 106 is the same as mine. have you a solution to the normal problem (faces become invisible from the outside)? I don't understand how you can shave any more off the tiles without losing the off-line vertices.

Something went wrong with that ... now there are only 96 there. So I am still short of 100.
Ah... a programmers error "i" instead of "j" !! Now the 102 are back, so I can go to bed.

Drongle, I have not constructed any of these in toto. My construction technique so far has been limited to manual folding in Wings 3D, so I have only verified what seem to be all the the necessary foldings in isolation. If I've kept everything straight in my head, they'll all work together without conflict. But there's certainly plenty of room for me to have made a mistake.

I had been thinking that I was just making minor tweaks to things you had already worked out. In retrospect, my descriptions may not have made much sense if we weren't working along essentially the same path. But in any case, I've just been working with ideas that you initially posted.

So now I think I have constructed a tiling of the 33x33 plane to yield 119 pieces. There are 99 whole tiles (the 3x4. with two opposite corners missing) plus 20 partial tiles around the edge. The partial tiles come in two shapes, and I've folded each of these shapes into a tetrahedron with the proper normals, so I am pretty optimistic that the whole thing is constructible. Is it possible we could collaborate on its construction? I would propose starting with the competion of your 106 model. If you send me your current bitmap, I could try folding the edges manually and we could go from there.

This has been a lot of fun for me. I would like to see us push it through at least to the point we have the McMahon-Turnbull conjecture for the maximum number of separate pieces from a single sculpty.

how about smat usage of transparency to use the same plane for multiple isolated separate objects that have complanar faces?
that should allow for static sculpted 3d grids that can blend out some elements via single texture animation.

Can't find you in-world. Don't know how best to send you the bitmap. Just as a texture? It's 105 now, with the three at each end missing visible faces. I lost the extra six in the 102, but now I found them again.

Concerning your extras.... I still can't see how. To be separate, they are not allowed to connect to any of the existing visible vertices. In my tiling, there are only six vertices left in the edge pieces for which don't contact other visible ones, and these are all isolated so they could only make triangles. I must be missing something again. I should try looking at them in wings ... haven't really learned it yet.

Very interesting idea, but thinking about the possible consequences of the alpha sorting bug in such an object makes me tremble with anxiety.

I've found a couple of my mistakes. The last six shaved tiles I thought I had fit in, I could preserve the normals OK, but when combined with the gross unfolds that put all the tetrahedra in a line, forced 3 vertices of the tetrahedra to be colinear. Plus, the seven I added to get from 106 to 113 were the ill-begotten progeny of hope and an overly marked up diagram. So I am back to 106 -- still in my head, not actually constructed. Do you still think 106 is possible, or are you stuck at 105?

Assuming you can format the sculpty bitmap as a .PNG file, you can attach it to a post here, either in this public forum or in a private message. Or you can email it to me at the email address I left in-world.

Ollj, I think that's a fine technique. It probably has more practical application than the mathematical puzzle Cel, Drongle and I have been working over.

I actually own a copy of that stone path. If you flip it over its all attached. But the sculpties turn inside out on the bottom. I also have quite a few indipendant scuplties, they ususally use the trick of making the ligatures too small to render. But they always have them connecting the sections. So thats a good trick, just bend the ligatures inside out, they appear invisable.

In short, it all depends on the shape of the objects, their position relative to each other, and how they are constructed. You can make many more sharp-edged smooth-edged cubes than sharp-edged.

The most efficient shapes I've found so far for fractional prim modeling are shapes with hexagonal profiles, and 12-sided. These seem to fit the sculpt space very well with limited waste, and retain shape on lower LOD as well.

When non-square sculpts are added (16x64, 8x128 and 4x256) are added, fractional prim modeling will be able to approach the efficiency of professional game models. As of now, it can be a burden to make UV repeats work with such sculpts, and there is a lot of waste with the vertex overhead of connecting elements using degenerate faces.

I dropped a photo of my next door neighbors' cat onto a sculpty and it gave me 102301 shapes. and I'm still counting.

I couldn't resist trying this. I made a simple cube of four cubes. It works well, but as expected, the alpha sort bug is very bad. Really rules out this approach except for special effects, like trees maybe. Not so horrible if you make the objects translucent so that the bug isn't so bad. Examples at Tiretta 80,149 for anyone interested. I won't leave them there long because I definately don't like them!
Actually, with the texture reiterated and a little animation, it has become a rather interesting object after all!

Is the cat sick? I think that is a hairball.

Drongle, comparing your 105 (with missing faces) to my putative tiling, I found my error. When I drew my plane out on graph paper, I gave myself 35 instead of 33 rows. Somehow, that allowed me to squeeze in more tiles. Now, the best I can do with the 3x4-2 tile is 100. (100 rather than 99 because the columns can be shifted to put a column with 7 on each end.)

Looks to me like your 102 solution is likely to stand as the ultimate answer to this little corner of the universe.

Absolutely, The tetrahedra are simply the solid figure with the lowest number of vertices, suitable for cramming them in. The objects I have made, just tetrahedra in line, are not really useful other than as an exercise. They are horrible disasters at the first LOD reduction or before complete rezzing.

Yes indeed ... I should have seen that. They were optimal if the edge ones all worked, but taking those out changed that!....Made one and added to collection with appropriate acknowledegement.

Hello all first and congrats on all this study , now my question is simple how many pieces of separate cilinders that have a vertical subdivision of 7 rows or 6 from top to down I can achieve ? this for making a long rail of different shaped columns but that look separate and that also could look good from distance?

so far my limit is 4

If I can achieve more I could save more prims but how I do ? I use 3smax or wings and I found myself more confortable in making sparated stuff with wings-....