View Full Version : Having a brain fart

10-26-2009, 02:17 AM
I am normally pretty good at figuring how to make models but this one is stumping me. I want a star that is slightly rounded as if it were laying on a ball. I have tried several ways of making it but can't get it to look smooth.

The best version I made is shown below from several angles, this version was a star that I used boolean union with a ball to create it. The edges are made up of random sided polys and this looks bad.

I though the way to do it would involve sub patching but when I make a star with control edges (tried several ways) it ends up looking too rounded in the "V" parts.

Any ideas?

10-26-2009, 02:46 AM
I think what you have here might work if you separated the edge polygons from the front and back polygons. You can do this by giving them different surfaces (i.e., side, front, back,) or just cutting the front and back faces and pasting them and don't merge the points. You might then need to triple some polygons near the edges, and tweak the smoothing ange in the surfaces.

If you want beveled edges, however, you'll need to model this differently. Maybe construct your star with triangles so it can be properly deformed with the magnet tool? This might give you cleaner geometry to use one the beveling tools with.

There are also a few edge bevel shaders available that can fake a beveled edge during rendering. Happy Digital's EdgeFX and TB Edge Bevel Shader come to mind. The TB shader is free, but they do things differently when it comes to open edges and subpatches, so they're each useful in different situations. There is also a free nodal shader that does another variation of this idea. (I forget the name of it, but I think it's a DP tool.)

Hope this helps.


10-26-2009, 03:28 AM
start with a disc and make it point by point

10-26-2009, 03:44 AM
here is what i mean..... :thumbsup:

10-26-2009, 05:40 AM
Or you can create a tessellated ball with 2 segments....subdivide once and then go from there.


10-26-2009, 05:46 AM
It's a run... :D Different approach here


This can have some issues with the 6-edged vertices it has but its just a start :P

Layer 1: Star
Layer 2: Divided and cut for Subdivision quads
Layer 3: Some Fill Loops to make a smooth shape.
Layer 4: Three Boundary Loops, 1 small 2 big. (used Multishift)
Layer 5: Positioning and Scaling the Mesh for Spherize. Smoothing the 6-edges and points around.
Layer 6: Use Spherize. It always spherize to a 1 meter radius sphere so the projection over that sphere will be controlabe with the scale and distance to origin. This way you know how much will the object be wrapped around.
Layer 7: Shift 3 times, 2 small 1 big (used Multishift)
Layer 8: Put an aprox. 1m sphere and sized to intended final size.

Hope it helps, it was fun to do it :P

10-26-2009, 07:09 AM
Nice solutions there, ain't this community just cool! :thumbsup:

10-26-2009, 12:23 PM
Weee.... www.youtube.com/watch?v=QgPXL_9nbN0 ;)


10-26-2009, 12:40 PM
Yes, this is fun! Good ideas guys! :)

10-26-2009, 07:05 PM
Well like i said the other time the 6-edged poles can show up as you can see in the left star's specular. Preety ugly rings there =\

So i looked into Kuzey example, with the Catmull freeze and freeze the pentagram, Spherize it and Extrude it, but no Subdivision this time. Just Dpont's node for rounding the edges.

Looks quite ok to me. Well, i learned something today :P


PS: KO pointfit looks great for this kind of operation in non-sphere. Nice reference.



10-27-2009, 06:13 PM
Another super-simple way is to make the flat star, extrude it for depth, then use Julienne and divide it x times on each axis. Then use Bend to make the curve. This one took about 90-seconds. :-)

10-27-2009, 07:38 PM
an idea,
wrapping using dynamics,


10-28-2009, 05:00 AM
I would have just booleaned a star out of a sphere and then would have retopologized it.