Interesting presentation

Richard_Rayer · March 7, 2019, 3:39am

I enjoyed this lecture. It presented a simple situation with a complex decision to be made. I thought a lot about each shape. I originally ruled out the icosphere because it was setup as triangles and all I ever heard about 3d objects and triangles was DON’T use triangles, they are bad.

So, I spent my time pondering the UV Sphere and the cube. I thought that the cube would not be a good idea because even though you could use subdivision surface and the “to sphere” command it would never actually be a sphere and animating it and possibly using physics the Cube would still react as a Cube. So, I selected the UV Sphere because it was the most actually round looking object to start with.

After listening to the rest of the lecture, I found that the Ico-Sphere had points in its favor as well. At least, I found myself thinking and trying to make a decision base on the situation at hand.

Keith_Scholes · February 28, 2019, 2:41pm

I too was inclined towards the UV sphere, and wasn’t entirely convinced by the reasons for discarding it, true the geometry does vary, but the most extreme variation is at the ‘poles’ and at the point where the holes would be placed the geometry is fairly regular. Similarly the converted square could accommodate the holes, provided the finger holes were separated by on odd number of polys with the thumb hole between them. I noticed Mike tried the finger holes separated by 2 and 3 polys, and neither configuration worked, but did not try a separation of one poly, that may have worked better. I ruled out the icosphere because it was the least spherical of the three shapes and even increasing the subdivisions to three, as in the lecture, it was clearly not perfectly spherical. Of course, a lot depends on how the physics engine treats the object, if it treats an object as perfectly spherical irrespective of its actual shape then the icosphere is fine. I did feel though that since the finger/thumb holes in our Blender ball are largely cosmetic the way in which the physics engine handles the object should be the primary consideration.

Having said that, in the real world a football is pretty much an icosphere, and works pretty well in a game, so what do I know.

Manu_Scheller · February 28, 2019, 2:31pm

Even if the ico sphere is not perfectly spherical, you could use a sphere as parent for the physics and just turn it off in the render

Keith_Scholes · February 28, 2019, 2:39pm

Good point.

Manu_Scheller · February 28, 2019, 2:41pm

I remember in the 2.79 part of this section, Michael even made a lower poly collision mesh for the pin for Ben’s Unity course

Keith_Scholes · February 28, 2019, 2:43pm

Although, the exact shape of the pin is perhaps less important as it just has to stand there and be collided with, whereas the ball has to simulate rolling.

Manu_Scheller · February 28, 2019, 2:46pm

That’s true. My point was, even you use a cube, if in (for example) Unity you import it and give it a spherical collision mesh, it still behaves like a ball . So in this case, needn’t worry about it in Blender, if you anyway plan to take it to a game engine

Keith_Scholes · February 28, 2019, 2:55pm

Agreed, as I say it depends on the physics engine in use. I haven’t gotten to that part of the section yet but I thought the plan was to use the Blender physics engine, which I imagine is less sophisticated than the Unity one. I think that my error was to imagine that the physics would work like real world physics, whereas (of course) you can effectively use whatever rules you like when using a game engine (or something like Blender), for example wrapping objects in differently shaped colliders.