From: Fubax
Minuscule position differences mess up the modeling, boolean operations, planar surfaces, create invisible geometries that are too small to zoom in, All that could be fixed by using fixed point arithmetic instead of floating point. There's no need for extremely large positional numbers, like lightyears and extremely small. Floating point in CAD design is the worst possible use case. Fixed point would not only make the app faster, more precise, but also more robust and error-free. I'm writing this post after frustration while modelling in MoI. Someone sooner or later will make an app with fixed-point arithmetic and it will be better than anything before.
From: Michael Gibson
Hi Fubax, in general there aren't any CAD libraries available to use that are based on fixed point instead of floating point.
So what you describe would require implementing an entire new geometry kernel library. It's not feasible.
Also fixed point does not just magically solve all precision issues. Just fundamentally any representation of real numbers in digital form will have limits to precision because it needs to fit in limited storage space.
- Michael
From: Fubax
Ok, I understand. Let's speculate how it would work. Imagine a program, where you choose the project scale (maximum dimension) and machining precision (smallest size step, like 50 microns). From that the fixed point format could be chosen and it would work essentially as under-the-hood grid snapping. In my opinion it would be way better.
From: Michael Gibson
Hi Fubax, there are some programs that work like that such as 3D Coat's voxel sculpting mode.
It isn't common in CAD though.
- Michael
From: BurrMan
Minecraft!!!
