I have to assume that even with this many small triangles, if we just build the model with them and do no processing it’ll still look slightly jagged. So the system has to do something after the mapping to smooth everything out. Is that process similar to the blurring of images that we covered before, but in 3d? Or is there something else?
Staffyirenng
@DanCubed Very good question! And yes, your intuition is right that we antialiasing techniques need to be applied.
The short answer is that antialiasing is accomplished with two (optional) rendering techniques: first, texture filtering (e.g. mipmapping), as described later in lecture; and second, supersampling as described in our lectures on rasterizing triangles.
But I think your question remains as a good question for all students to consider when studying this material: why do these two techniques work to produce the smooth and high quality results shown in the center image of the top row?
Some things to consider:
How will the result change if the triangles are coarser or finter than shown? E.g. what if the triangles are smaller than a framebuffer pixel, or large so they cover many pixels?
How will the result change if the texture image is higher or lower resolution? E.g. what if it was 4K x 4K, but the output rendering in the middle top was only 100x100 pixels on the screen? Or vice versa? (To answer this will require a look at the rest of this lecture.)
adham-elarabawy
There seems to be a high degree of variance in density of mesh triangles describing the surface. It seems as though it might make more sense to try and concentrate more triangles in areas of higher complexity, and less triangles to areas of lower complexity -- are there any automated ways to do this? Or alternatively, how is this balanced in the real world/industry?
I have to assume that even with this many small triangles, if we just build the model with them and do no processing it’ll still look slightly jagged. So the system has to do something after the mapping to smooth everything out. Is that process similar to the blurring of images that we covered before, but in 3d? Or is there something else?
@DanCubed Very good question! And yes, your intuition is right that we antialiasing techniques need to be applied.
The short answer is that antialiasing is accomplished with two (optional) rendering techniques: first, texture filtering (e.g. mipmapping), as described later in lecture; and second, supersampling as described in our lectures on rasterizing triangles.
But I think your question remains as a good question for all students to consider when studying this material: why do these two techniques work to produce the smooth and high quality results shown in the center image of the top row?
Some things to consider:
There seems to be a high degree of variance in density of mesh triangles describing the surface. It seems as though it might make more sense to try and concentrate more triangles in areas of higher complexity, and less triangles to areas of lower complexity -- are there any automated ways to do this? Or alternatively, how is this balanced in the real world/industry?