This is a modification of neoarchaics "NSphere" http://www.thingiverse.com/thing:81449

It consists of twelve pieces that have these carvings on their surface and can be connected to each other to form a sphere of the size of a tennis-ball. No matter how the sphere is assembeled the carvings on its surface will always line up and make a pattern of lines.
If you turn one piece or switch positions with another the pattern on the sphere will change but no matter what you do it will always make sense.
This makes it a "antipuzzle": There are no wrong solutions, just different ones....

Make your own collection of modules, slice, print and assemble them and have fun... ;-)

P.S: Post a picture when you made one.

There are many different modules with different carvings. They all have different carvings on them but fit to each other. On the first print -the one that is shown in the picture above- i took the modules 2 to 7 and printed them out twice.

I have painted it in black by filling the carvings with toothpaste and spraying black acrylic paint on it. You can wash of the toothpaste after that, but there were flaws left on the surface. I got them by using a black sharpie and Tipp-Ex.

Slicer was Cura with 0.1mm layer height at 150 mm/s.
I had some problems with a non-closed surface. Perhaps that was because the arching of the surface so the lines were not broad enough to cover lines underneath it. Using a higher value (than 0.4mm) for the top/bottom could help.

The size of each module is about 5cm. The pins have about .7mm radius and a very small footprint on the bed, so it could happen that some of the first layers of the pins will be displaced during print.
This happened during my print. I did nothing about it and let the print finish. The parts work great as you see in the pics though the fitting is not perfect anymore.
On a really well configurated printer that gets the pins printed properly it will look even better!

Since the pins are the weak spot of the design i intend to come up with another model that uses magnets instead.

Also I will publish the .scad-files as soon as I have dealt with some scaling issues and cleaned/commented the code.