I noticed that the meshes on this slide were composed of quadrilaterals instead of triangles, which prompted me to look into the pros and cons of using one or the other. I found that subdividing quads is generally easier than subdividing triangles, as subdividing the latter can cause weird effects like pinching. This also makes sense considering that a lot more post-processing goes into subdividing triangles than subdividing quads, like edge flipping and centering vertices. Meanwhile, quads appear to divide really smoothly on their own, as shown on the slide. However, since all quads can be decomposed into two triangles anyway, is using quads more for the convenience of the human modeler than for the computer interpreting the mesh?
muuncakez
you bring up a great point and it also makes me wonder... does the computer inherently treat quads like two triangles but visually to a human modeler it's still a quad and the mesh is still rendered as such? that way the human can still have the convenience of a quad but the computer still interprets the mesh as triangles anyways since quads can be divided into triangles with one edge?
muuncakez
I also wonder if quads are only most convenient with subdividing and especially with more detailed meshes (detailed as in quantity of quads)? Looking back at the regularization and simplification mesh example(s) that use triangles, the overall geometry of an object seems to be better maintained in these processes when using triangles while quads seem chunky and less precise (with edges and contour especially) the simpler you get. So, then, while triangles require more "attention" from a human modeler would that time spent (always?) pay off since now you have more flexibility and "control" of your edges, counter, and, ultimately, your final render?
DestinyLuong
I know there is technically no use for this, but would it be possible to do the opposite and make a smooth or rounded object more geometric, or with harsher edges?
Alescontrela
I'm curious to hear about the rules people use to ensure that surfaces don't develop strange artifacts after repeated subdivision.
I noticed that the meshes on this slide were composed of quadrilaterals instead of triangles, which prompted me to look into the pros and cons of using one or the other. I found that subdividing quads is generally easier than subdividing triangles, as subdividing the latter can cause weird effects like pinching. This also makes sense considering that a lot more post-processing goes into subdividing triangles than subdividing quads, like edge flipping and centering vertices. Meanwhile, quads appear to divide really smoothly on their own, as shown on the slide. However, since all quads can be decomposed into two triangles anyway, is using quads more for the convenience of the human modeler than for the computer interpreting the mesh?
you bring up a great point and it also makes me wonder... does the computer inherently treat quads like two triangles but visually to a human modeler it's still a quad and the mesh is still rendered as such? that way the human can still have the convenience of a quad but the computer still interprets the mesh as triangles anyways since quads can be divided into triangles with one edge?
I also wonder if quads are only most convenient with subdividing and especially with more detailed meshes (detailed as in quantity of quads)? Looking back at the regularization and simplification mesh example(s) that use triangles, the overall geometry of an object seems to be better maintained in these processes when using triangles while quads seem chunky and less precise (with edges and contour especially) the simpler you get. So, then, while triangles require more "attention" from a human modeler would that time spent (always?) pay off since now you have more flexibility and "control" of your edges, counter, and, ultimately, your final render?
I know there is technically no use for this, but would it be possible to do the opposite and make a smooth or rounded object more geometric, or with harsher edges?
I'm curious to hear about the rules people use to ensure that surfaces don't develop strange artifacts after repeated subdivision.