Problem with Minifig Head Pattern

[Christian Neumann]

hi,
i have run into problems while finishing a minifig head pattern. when i take a look at the result in ldview (quality mode), usually the edges with conditional lines are smoothed out, but not this time. the overall look is some kind of "ok", but there are several regions that look strange.

here is what i did:
1. created flat pattern
2. sliced with slicer pro, then optimized the result manually (e.g. re-triangulation with new vertices)
3. projected with slicer pro onto former shape
4. to get rid of some rounding errors (0.00000001), i used former shape as subpart in shaped pattern, and snapped on it with unificator
5. checked with planarcheck, found no problems
6. created conditional lines with edger2

however, the result is not satisfying me. i tried with several different thresholds and precision settings, but i don't know exactly at which point the problem occurs. maybe someone can give me a hint what to do?

here is a screenshot of what i'm talking about:



i discovered this in some other minifig heads, but there must be 1) a reason for this, and 2) a solution how to avoid it?
any help is appreciated.

greetings,
christian

[Philippe Hurbain]

Hi Christian,
It's very difficult to get smoothed ldview result for rounded patterned parts. There are many possible causes, the main beeing that to get a better result you would need to add condlines even in planar areas. Or it may be very thin triangle/quads that break smoothness. As I have not found satisfactory way of getting perfect, I simply copy condlines from former to get proper lies on edges of the shape, the initial sole purpose of condlines.

[Christian Neumann]

Ok Philo, thank you for that information.

[Travis Cobbs]

For reference, I consider this to be a bug in LDView. That doesn't mean I have a scheduled time for it to be fixed; it just means that I consider it to be a bug in my smoothing algorithm.

Before trying to fix bugs in my algorithm, I have to decide if it's worth it, since I have also considered replacing it with a different algorithm. This other algorithm is the one I used for smoothing in my POV export. I haven't really received any feedback about whether it's better than or worse than the current algorithm (although I also haven't asked for such feedback). However, the current algorithm is useful to part authors trying to not forget conditional lines, so it will likely always remain available as an option. (The algorithm used in POV exporting uses the angle between faces, but also detects the presence of non-conditional edge lines between faces to force a crease. Properly designed parts should always have an edge line between hard-edged facets, even if the facets have a small angle between them.)