3
$\begingroup$

I'm trying to print a small box with 1 mm thick walls. The box has rounded corners.

Cura slices the rounded corners with straight internal lines instead of rounded lines:

Rounded Corners with infill

The reason for these straight lines is that Cura is using infill to fill the walls, because it determines that there's not enough space in a 1 mm wall for three 0.4 mm passes.

Rounded Corners without infill

This causes the walls to have small imperfections in the final print. I was able to solve this by increasing the model wall thickness to 1.2 mm (to allow for 3 0.4 mm lines), but even then it wasn't doing it as a single continuous line, but instead breaking it into short lines at the corner:

Rounded corners for 1.2 mm thick walls

Is there a way to avoid this issue without increasing the wall thickness, and have 3 continuous lines?

$\endgroup$
7
  • $\begingroup$ I migrated to a good slicer after having to deal with those things in Cura. $\endgroup$
    – towe
    Oct 16 '20 at 11:51
  • $\begingroup$ @towe What slicer would you recommend to use instead? $\endgroup$ Oct 16 '20 at 12:03
  • $\begingroup$ I'm currently using SuperSlicer (based on PrusaSlicer) and am quite happy with it. You do lose tree supports though. $\endgroup$
    – towe
    Oct 16 '20 at 12:11
  • 1
    $\begingroup$ @0scar actually it is a model problem - the walls in rounded corners get closer than in straight lines as they are pretty much attempting an infinitesimal solution, which puts the walls a tad closer together in the center between two walls in those areas. $\endgroup$
    – Trish
    Oct 18 '20 at 11:31
  • 1
    $\begingroup$ @0scar Not a geometry problem, at least not for proper slicers. Answer inbound :) $\endgroup$
    – towe
    Oct 19 '20 at 5:18
2
$\begingroup$

Fixing the model

My solution to this problem is, to increase the thickness in such corners by 0.05 mm by pushing the inner wall's radius towards its center. This allows some extra space as a round corner is not actually saved round but as a pattern of straight wall facets, and those can get squished together just enough to make the slicer believe the wall is too thin for a proper wall. As you see, the new corner rounding starts a little bit before the outer one - in this exaggerated model I use 1 millimeter extra - violet are the normal corners 10 mm apart, black is the new inner corner, the light blue area between them is the extra bulk that will massively reduce the problem.

enter image description here

origin of the error

As an example, let's put the number of facets to 1 and 2 when we form them from the rounded corner. The more facets there are, the closer the distance between the walls will get to the full wall thickness, but it will always stay a fraction below.

enter image description here

How does the model fix work?

Well, we shifted the inner corner away, and with 2 facets we get a fidelity to the intended wall thickness of 96.21 % instead of 92.39 % for the non-modified wall - and with even more facets the fidelity gets better more quickly. However, there is a point where your gaps will start to diverge in the corner, so just using a tiny alteration is advisable - in our model case, 9 facets (a 10°) is the point where the inner wall will start to become more distant to the outer wall. That's why I only need such a tiny bit of extra space if I export the model with the most fidelity in the .stl generation my CAD has available.

enter image description here

making it work in the slicer

Another thing I do when I can't alter the model is, knowing that I usually slice width a line width of 0.45 mm. In case such a corner appears, I can then alter my line width in the slicer down to 0.44[5] mm (or more) and get pretty much the same result at the cost of possibly needing to account for errors in other areas.

$\endgroup$
2
  • $\begingroup$ Changing the line width fixes some corners, but causes this issue to appear in layers that were previously OK. Can you elaborate on how to increase the thickness of the corners? Do you mean in the model itself? $\endgroup$ Oct 18 '20 at 11:38
  • 1
    $\begingroup$ @YoavKadosh Added a picture. In the model itself you alter the inner diameter a tiny bit. Technically you increase the inner diameter and push it's centerpoint away from the corner. This can be most easily be done by using a Push/Pull on the inside of the corner. $\endgroup$
    – Trish
    Oct 18 '20 at 11:45
2
$\begingroup$

Contrary to the other answer, this is not (exclusively) a model / geometry issue.

Frame challenge: There's no good / easy solution to this when using Cura, at least that I know of. Better slicers can easily handle this situation though.

I replicated all the tests in SuperSlicer, which is a PrusaSlicer branch. All tests were done with models with relatively high resolution:

enter image description here

As well as a model which I created to replicate a horrible quality .stl:

enter image description here

I checked the continuity of the added internal line with Simplify3D's g-code preview.

For a 0.4 mm extrusion width and a 1 mm wall thickness, PrusaSlicer simply adds a 0.2 mm line in the middle. This line is completely uninterrupted:

enter image description here

The same happens with the very rough .stl:

enter image description here

The same thing happens when using a 1.2 mm wall thickness: enter image description here enter image description here

Conclusion: You should not be adding esoteric "fixes" to your models to prop up terrible slicer performance.

$\endgroup$
2
  • $\begingroup$ Technically you use a Slic3r branch then. $\endgroup$
    – Trish
    Oct 20 '20 at 15:08
  • $\begingroup$ @Trish Correct, though Slic3r itself has become obsolete. $\endgroup$
    – towe
    Oct 21 '20 at 5:51

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.