Generated Circles in MatterSlice are horrible!?!

[AEK]

Can someone please help with this? Maybe I'm missing a setting somewhere??

Whenever I slice in MatterControl's MatterSlice I get very bad/weird circles/arcs... It's like it's making "large" straight lines to make up a circle. I understand that it won't use G02 & G03 (not very smart!) but the triangles created are so small that it should look descent as it does with Sli3r, Cura and S3D.

Using SolidWorks, exporting at the highest possible settings for .STL and setting them up with the same (and tried different) settings in MatterControl's MatterSlice, MatterControl's Slic3r, MatterControl's Cura and Simplify3D.

In Simplify3D I get beautiful circles/arcs.

1.00in Cube Test - Simplify3D - 0.2mm

In MatterControl's Slic3r I get beautiful circles/arcs.

1.00in Cube Test - MatterControl - Slic3r - 0.2mm

In MatterControl's Cura I get beautiful circles/arcs.

In MatterControl's MatterClice I get very bad circles/arcs!

1.00in Cube Test - MatterControl - MatterSlice - 0.2mm

I like MatterControl otherwise for it's great infill (grid & triangles), it's speed using MatterSlice is great too compared to Slic3r and especially Cura in MatterControl (very slow).

Am I missing a setting somewhere??

Attached see screenshots of the top countersink and side countersinks using Matterslice, Slic3r and Simplify3D.

A friend of mine drew up this test cube to test squareness at 1.000", holes at 0.500" with a countersink at 0.750" using SolidWorks.

Any suggestions/help would be greatly appreciated!

3D Printer: Rostock MAX v2 (stock)
Filament: ABS & PLA (for this, onscreen is the test)
Computer: Core i7 on Win 7 x64 (Win 8.1 x64 same results)

Andre

[Polygonhell]

Assuming you used the same STL exported from SolidWorks and you've looked to see that it has more faces than the matter slice output. I'd look at the STL file it has no concept of circles so the tesselation is based on the setting during export from solidworks. It's possible it looks like the Matter Slice image and the other slicers are trying to identify circles and re-tesselate them. S3D in particular has a horrible habbit of removing small details on curved objects, so it may be that it's "smoothing" the input in some way.

A lot of slicers change geometry, when they preprocess the geometry they basically pick a resolution and quantize to it. They them remove degeneracies, this fixes any small holes in the model or degenerate edges. Some Slicers appear to do much more than this.

Having said that I understood that MatterSlice was using Cura Engine and I haven't seen that level of introduced error in Cura.

[AEK]

Yes, for all the tests, using all the "slicers", the same .STL model was used.

Same settings for each "slicer" was used as well to keep things the same, ie: 0.200mm layer heights, etc...

Here are the screenshots of the part and the export settings to .STL from within SolidWorks.

SolidWorks File:

1.00in Cube Test - SolidWorks

SolidWorks Settings for File:

1.00in Cube Test - SolidWorks Settings

I've noticed before on other "round holes" come to think of it from downloaded things from Thingiverse, that there was this "basket weaving" action taking place only using MatterSlice withing MatterControl.

Is there a fix for this? Export at a much lower resolution??

Thanks again for your advice/help!

Andre

[Polygonhell]

Given your images of the export I'd say it's a MatterSlice issue, it's probably re-samplng the geometry, I don't use it so I can't comment on if there is a setting to fix it.

[JFettig]

I also run very high resolution exports from Solidworks but I haven't run MatterSlice. I'm guessing its a tolerance thing in the slicer, whether an option or hard coded.

A quick hint for STL export - click the 'do not translate...' check box, it'll make it easier to export multi-body parts.

[AEK]

Thanks for the help guys!

I'll try ticking off that box in SolidWorks as you suggested.

I don't "usually" export more than one .STL at a time though since Simplify3D CANNOT "un-group" or "group" multiple .STL's at a time, shame really (as is it's scaling feature, or lack of direct size input).

Seems there is no "perfect" slicer as of yet?

Again, thanks for all the help!

Andre

[JFettig]

Make an assembly in as and save it as a stl with that box checked allowing them to save as multiple files and turn auto arrange off in s3d.

[Captain Starfish]

If there was a perfect slicer we'd all be using it, no such thing yet. I still rate MatterSlice my favourite though, despite its quirks, and I haven't noticed problems with circles coming out of AutoCAD or FreeCAD.

I'd be using Slic3r for a few jobs and things like the hex fill if it didn't have such a propensity to bog down on the "writing G-code" step to the point where you think it's crashed and cancel it.

[kevinepope]

Quick Update - This issue have been identified and addressed in MatterSlice. The fix will be included in the next MatterControl update.

More info on the problem here:
https://github.com/MatterHackers/Matter ... issues/204

To quote:

The final analysis of this was that this model was exposing a very subtle bug in clipper. If there were a very large number of edge points on a relatively shallow curve, one can't just merge each as they walk the curve in order or they can collapse curves that have small enough inter point deltas but a large total curve.

Imagine a very tessellated circle. As we walk the edge, each triple we find can be collinear within our error and so we remove the middle point and move on to the next point. The new middle point is very close to the new third point and so we collapse and remove the second point. We can remove and create a large flat on the edge of the circle.