Loft Problem

 From:  Michael Gibson
6464.2 In reply to 6464.1 
Hi bz, those stretch marks look like just a display artifact from slightly long and skinny triangles in the display mesh, it looks like your actual surface is fine.

To really see more of the true surface, try doing an export to a polygon mesh format like OBJ for example, then crank the slider over to the right and also if you have things that are long but only slightly curved, putting in a distance value for "Divide larger than" can also be good as well to make sure the polygons are dense enough that you're not seeing polygonalization artifacts alone.

Like in your case here I exported to OBJ and cranked the slider all the way to the right (the slider controls the "angle" parameter), and also put in "divide larger than" = 2, and that makes the polygon representation nice and highly diced up so you won't see what are called "mach banding" artifacts that are associated with trying to make somewhat rough polygon structure look as if they are smooth.

When I do the OBJ export (which you can just cancel if you were only wanting to examine the overall true surface quality) with those settings, in the mesh preview that you see on screen you should see those "stretch marks" are not there anymore:



If you do go to export this object to a rendering program, you will also probably want to put in some amount of "Divide larger than" at that time as well, when polygons are long and skinny it can tend to make shading artifacts spread across a sort of larger area and thus become more noticeable. When things are diced up only by angular measurements alone (which is the case for the in-viewport display mesh) then a semi-tubular shape like this that is only gradually curved in one direction but tightly curved in the other tends to get long skinny triangles since the long gradually curved region does not end up with very many divisions in it. That's what putting in "Divide larger than" can correct, it forces divisions based on length rather than just on angular deviation of surface normals.

- Michael