I was told that n-gons were bad, but they seem a necessary evil.
1.) There seems no substitute for 5-edged ngons, especially when intersecting 2/3 rows. Even splitting the 5-sided ngon makes things look bad. Usually, the triangle causes a seam over the roundness of something.
2.) There also seems no substitute for an an ngon that has a spit edge on one side to allow/cause a resolution reduction. This allows the resolution to grow/shrink.
Pretty smooth:
Now, I am wondering: would an untrimmed 5-edged NURB be helpful?
The problem is: triangles and wedge-shaped quads tend to cause cut depressions on convex surfaces.
Oddly, as in the first object, the 1 to 2 highlighted connecting polygon exists in many valid scientific meshes, as many have variable density zones, such as this Cfmesh meshing, albeit in 2 d.
I have experienced the same, sometimes, actually most times, Ngons look nicer than a perfect topology.
I have no problem with that honestly. And it also makes sense. Some seams just ruin the smoothness of the shape and having ngons average them all makes things look better.
There is a lot of beliefs floating around on the internet about modeling. Frequently there are valid in some contexts but not in others. Frequently the context is not apparent, not mentioned or perhaps not even understood by the person advocating the beliefs.
