Maybe I'm misunderstanding but how would this sample generate the same object appearance outcome as the subdivisioning example (aka the visually more detailed mesh)? It looks like I have lost a lot of the original geometry (e.i. the ears and horns are shorter, etc.) and my object now looks like a cow wrapped in wrapping paper instead of just looking like a cow. How simple of a mesh is too simple? Does the quantity have a direct relationship with the quality of my mesh? I feel I am inclined more is better... because how does distribution improve quality if visually it looks less like a cow?
(Is the goal to have my mesh look like a cow wrapped in wrapping paper hence why it's called a mesh?)
stang085
I think it's interesting how you can get so much more information even though it's the same amount of data, just having them placed in the right spot gives you a better idea of the structure. In a way, is modifying sample distribution an anti-aliasing technique?
Yeek2
Is there some kind of "golden rate of sampling" (Nyquist-theorem-esque) that preserves the shape/information of the mesh or that "looks the best" overall? Personally I find myself slightly more inclined towards the mesh on the right (Perhaps it's due to there being slightly more shading on underneath the nostrils as well as beneath the ears and horns), but much is say "too much"? I suppose this question asks something similar to munncakez of does quantity yielding more quality?
MillerHollinger
What exactly do we mean by "improved quality" here? I feel that the mesh on the left better expresses the shape of the cow, especially in the ear and eye area. I could imagine a practical use of mesh regularization being to prevent issues when rigging and animating a model, but the practicality in any other case seems limited to me.
ttalati
I remember in the beginning of the year we went over how we could have used used rectangles for example to make up our mesh, but in the end the use of triangles was overwhelmingly chosen as the desired basic polygons. Among the other properties that were discussed such as any polygon being able to be broken down into triangles, is having a triangular mesh representation beneficial in making these mesh processing algorithms and math easier? I imagined splitting rectangles into smaller ones and hence subdividing or simplifying may be less complex, but maybe it does not have desirable image effects?
Maybe I'm misunderstanding but how would this sample generate the same object appearance outcome as the subdivisioning example (aka the visually more detailed mesh)? It looks like I have lost a lot of the original geometry (e.i. the ears and horns are shorter, etc.) and my object now looks like a cow wrapped in wrapping paper instead of just looking like a cow. How simple of a mesh is too simple? Does the quantity have a direct relationship with the quality of my mesh? I feel I am inclined more is better... because how does distribution improve quality if visually it looks less like a cow? (Is the goal to have my mesh look like a cow wrapped in wrapping paper hence why it's called a mesh?)
I think it's interesting how you can get so much more information even though it's the same amount of data, just having them placed in the right spot gives you a better idea of the structure. In a way, is modifying sample distribution an anti-aliasing technique?
Is there some kind of "golden rate of sampling" (Nyquist-theorem-esque) that preserves the shape/information of the mesh or that "looks the best" overall? Personally I find myself slightly more inclined towards the mesh on the right (Perhaps it's due to there being slightly more shading on underneath the nostrils as well as beneath the ears and horns), but much is say "too much"? I suppose this question asks something similar to munncakez of does quantity yielding more quality?
What exactly do we mean by "improved quality" here? I feel that the mesh on the left better expresses the shape of the cow, especially in the ear and eye area. I could imagine a practical use of mesh regularization being to prevent issues when rigging and animating a model, but the practicality in any other case seems limited to me.
I remember in the beginning of the year we went over how we could have used used rectangles for example to make up our mesh, but in the end the use of triangles was overwhelmingly chosen as the desired basic polygons. Among the other properties that were discussed such as any polygon being able to be broken down into triangles, is having a triangular mesh representation beneficial in making these mesh processing algorithms and math easier? I imagined splitting rectangles into smaller ones and hence subdividing or simplifying may be less complex, but maybe it does not have desirable image effects?