Just posted this over in my Renderosity gallery

In the end I used nine of the 37-hex grass figures scaled down. I started with just one figure using random new grass geometries and random 60 degree rotations but it looked too sparse. So I added a second in the same place and randomized in the same way. But some of hexes ended up with the same geometry (there are 5 new grass geometries to select from) and rotations (there are 6 sixty degree rotations) - what are the chances of that* ? - and it still looked a bit sparse. So I added a third and used 1 degree increments for the rotations on all three. Much better, and applying my ground texture to the ground of the Poser construct covered any remaining gaps, although with ray-traced lights the shadows of the grass mad it okay even with a white ground.

But the edge of my three overlaid hex figures was very distinct - grass where the figure is, just the brownish ground texture where it isn't. So I parented the second and third figures to the first and then duplicated the first figure, twice, moving one back and left, the other back and right. Better, but the far edge of those was now distinct. So I just used Poser's DepthCue to hide it - I think I used the basic one, although I may have used my own version of the bagginsbill one

from post #13 here.

With nine of my hex figures the render time slowed noticeably, but not drastcally, so I'm quite pleased.

It's a nice flat lawn

*Let's work it out just for fun (I had to do a refresher on... not probability, but how to add and multiply fractions !)

For a

**single** hex in the second figure to have the same geometry and rotation as the same hex in the first figure it must have the same geometry (1/5) and (*) the same rotation (1/6), so (1/5)*(1/6)=(1/30)

And for a

**single **hex in the second figure to

**not **match the geometry and rotation of the corresponding hex in the first figure, the geometry must be different (4/5)

**or** (the geometry's the same (1/5) and (*) the rotation's different (1/6)), so (4/5)+((1/5)*(5/6))=(29/30)

So for

**all 37** hexes in the second figure to

**not **match the geometry and rotation of the corresponding hexes in the first figure the first hex mustn't match (29/30) and (*) the second mustn't match (29/30) etc=(29/30)^37, which my calculator tells me is just over 28.5%.

So the chance of

**at least** one hex in the second figure being identical to the corresponding hex in the first figure is a bit over 70%.