Lecture 8: Mesh Representations and Geometry Processing (62)
CeHao1
It seems that more simplification will omit much information of the original object, even though we are trying to minimize the quadratic error.
Considering this, what are the criteria of this simplification, say we can still recognize the object, while there are only very few shapes.
Some places such as the eye and mouth position of the skull are more recognizable than other surfaces. So we should keep more shape at those more important positions.
xinwei-zhuang
As a supplement to CeHao's comment, the lecture mainly covers the geometrical way of mesh simplification, but it might be useful to take semantics into consideration.
micahtyong
Furthermore, how do we decide how many mesh elements to reduce our model by? Is there some relationship we can draw between human perception of the model and the mesh model itself?
greeknerd1
Is there a general rule of thumb for the number of elements required to maintain the geometric shape of an object? Or is it on a case by case basis?
It seems that more simplification will omit much information of the original object, even though we are trying to minimize the quadratic error. Considering this, what are the criteria of this simplification, say we can still recognize the object, while there are only very few shapes.
Some places such as the eye and mouth position of the skull are more recognizable than other surfaces. So we should keep more shape at those more important positions.
As a supplement to CeHao's comment, the lecture mainly covers the geometrical way of mesh simplification, but it might be useful to take semantics into consideration.
Furthermore, how do we decide how many mesh elements to reduce our model by? Is there some relationship we can draw between human perception of the model and the mesh model itself?
Is there a general rule of thumb for the number of elements required to maintain the geometric shape of an object? Or is it on a case by case basis?