Slic3r not printing solid walls at certain angles

At certain angles, walls that are not perpendicular to the plate, but also are not near 45 degrees, do poorly when I print them using Slic3r. They come out much thinner. Sometimes this leaves little gaps, sometimes it makes the walls so thin that they are semitransparent. As you can see by the picture I attached, my robot has polka dot underpants because you can see through the wall to the support structure underneath.

I upped the top and bottom layers to 5 solid layers. That didn't help.

I installed Slic3r 2.6 and changed the infill/perimeters overlap. Once I got up to 50% it helped a little bit, but not much, and it started causing other problems.

I didn't have this problem with Makerbot Desktop. The same pieces printed out with solid walls at all angles. However, Slic3r is better in many other ways, so I'd prefer to keep using it.

Any suggestions?

Edited 1 time(s). Last edit at 02/20/2015 12:49AM by JasonMBryant.

You should increase the number of perimeters, to make sure the perimeters above overlap the ones below. If the option to add perimeter as needed is checked, this would normally be accomplished automatically.

I'll try that, but it's still odd.

I've got "extra perimeters if needed" checked. It's currently set to 2 perimeters, which is the same as Makerbot Desktop. It seems that Slic3r just doesn't allow for enough overlap on these angles.

So this is still happening. I'm sure it's not an extrusion problem. Increasing the extrusion causes bad things to happen in other places, like putting out a mess of extra filament at the top of the part.

As I mentioned, "extra perimeters if needed" is checked.

I've upped the perimeters to 3. That made it happen a little less, but it still happens. Slicer has little holes in the walls at 3 perimeters when Makerbot Desktop has none at 2 perimeters.

Is anyone else seeing this? Could someone else try printing something that has walls at these angles and see what they get? To me it seems like slic3r isn't handling walls at these angles very well, but that only makes sense if it's happening to more people than just me.

You can use modifiers in Slic3r to modify the parameters in these zone.

That's a terrible, terrible solution.

I have a feeling that there is a bug in the code that calculates the amount of edge infill required under a sloping surface. In my experience it usually works OK, but some angles appear to cause Slic3r to lay down too little infill. I have however not conducted any formal tests or checks, and may therefore have misinterpreted what I have seen happening. It would not be difficult to make a few models with different slopes (or one model containing several different slopes) drop them all into plater and test how Slic3r copes with each of them. A large dome contains a continuously varying slope, and was where I first noticed the anomaly.

Dave