Feature Request: Spiralize with variable Layer Width

[TheTiEr]

Hey Mark,

i use your Version of Cura. I think your modifikations are really good and this Slicer benefits from these changes.

So i have the following Problem:

Cura won't generate a spiraliced contour with a varying Linewidth. The Linewidth is specified with a hollow stl file (shell).

Hopefully you are able to add this feature in the next version. The devs from the original Cura lethargic to add this (maybe very experimental) feature.

Greetings Timo

#Ultimaker#1058 (comment)

[smartavionics]

Hello Timo, I have read the comments in Ultimaker#1058 and I think I understand your request. I will look into what it may take to do. Will get back to you...

[smartavionics]

Hi Timo, I have something working, it's not perfect but if you want to try it I can build a release.

At the moment, I have two issues:

1 - the layer view looks broken with odd straight line segments appearing. However, the gcode prints OK.

2 - it's proving difficult to completely hide the z-seam so I will continue to work on that aspect.

[TheTiEr]

Hey Mark, that looks really great.

Yes a release would be great, so that i could test the feature and can give you some feedback.

The Layer View with the staight lines is nebensächlich as long as the GCode will work.

[smartavionics]

I have fixed the layer view problem but the z-seam is still difficult. I think that's going to take some time to fix.

I will build releases for Linux and Windows later today.

[smartavionics]

Hi Timo, I have uploaded Linux and Windows builds. At this time, no extra settings are required. It will recognize when a model is not solid and attempt to create spiralized walls with the correct widths.

At this time, the spirilzed line width is constrained to be between 50% and 200% of the nominal outer wall width. Maybe in the future people will want to change those limits and then I can add some settings to do that.

As I said, the z-seam is problematic and I don't know whether it will be possible to eradicate it completely. Anyway, please test and provide some feedback. Thanks.

[TimKrautwald]

Hi,

thank you for the fast response! I will test it on friday with my printer and will definitely give feedback on the result. Just to understand what I have to expect, is the problem with the z-seam a preview thing or something that will show on the printed part? I can imagine that the transition between various layers and layer width in a seamless way may prove difficult. Maybe smoothing with a lowpass could help as a workaround (limiting the rate of change, if there is a steep step).

The limits seem to be a good idea to keep the slicer from exploding when there is a flat plateau. Otherwise this could result in unprintable wide extrusion. With nominal outer wall width you mean the setting in Cura, right? So if I put 1 mm in there, the extrusion width can vary from 0,5 to 2 mm?

Again, thank you for your work!

[smartavionics]

The problem with the z-seam is that in some places it produces a visible mark on the part. It occurs in the places where the wall thickness is changing. In places where the wall thickness is constant, the z-seam isn't visible. I need to investigate what's causing that.

With nominal outer wall width you mean the setting in Cura, right? So if I put 1 mm in there, the extrusion width can vary from 0,5 to 2 mm?

Correct. I tried using 0.2 instead of 0.5 and holes appeared in the wall because the extrusion rate was so low. You will definitely need to use a high steps/mm value for your extruder if you hope to print really thin walls. I am using 830 steps/mm on that printer.

Looking forward to hear how you get on, cheers!

[smartavionics]

I've discovered a problem with my implementation. It only determines the wall line widths at the vertices in the outer boundary so if you have an object like this example, all the walls will be printed using the same width lines because the narrow regions are due to the inner boundary approaching the outer boundary and so are not detected.

It would be nice if it could detect the narrowing due to the inner boundary approaching the outer boundary at locations that are between the vertices in the outer boundary so I will put some thought into how that may be achieved.

[TheTiEr]

Hey, i have tested your implementation and it is working good for me. Only the mentioned is z-seam is visible if the print has a decent steep angle.
I think i have found a reason for that:
I use LinuxCNC for my Printer which has an inbuilt Oszilloskop . With that you can see the motion and velocity of every axes.

Here you can see the velocity of the x- and y- axis. Notice the spike of the red line in the middle of the image. For a very small time this axis turns backwards. This forces the printer in a full stop and the material has some time to ooze out of the nozzle.

I examined the G-Code and found the observed behaviour here too:
G1 X166.08 Y31.031 Z2.397 G1 X166.038 Y29.723 Z2.399 G1 X166.033 Y29.349 G1 X166.055 Y28.414 Z2.4 ;MESH:NONMESH G1 X166.052 Y28.413 F42000 <-- Travel Move in the opposite direction on the x_axis ;TIME_ELAPSED:61.629519 ;LAYERTIME:5.014128 ;LAYER:8 TYPE:WALL-OUTER G1 X166.284 Y28.311 F2757.6 G1 X166.322 Y27.575 Z2.401 G1 X166.36 Y27.097 Z2.402 G1 X166.413 Y26.544 G1 X166.558 Y25.499 Z2.403 G1 X166.611 Y25.169 Z2.404 G1 X166.703 Y24.7 G1 X166.805 Y24.26 Z2.405

Don't bother the missing extrusion axis, i am using velocity Extrusion on my Printer.

Here i found an individual travel in y and z, This will prevent a smooth blending. Maybe it is useful to combine these moves or better to get rid of them completly?
G1 X192.297 Y39.247 Z35.534 G1 X192.267 Y37.404 Z35.537 G1 X192.237 Y34.8 Z35.541 G1 X192.243 Y32.194 Z35.545 G1 X192.291 Y28.407 Z35.55 ;MESH:NONMESH G1 Y28.405 F42000 <----- travel move in y ;TIME_ELAPSED:598.694209 ;LAYERTIME:5.066477 ;LAYER:115 G0 Z35.45 <---- travel move in z TYPE:WALL-OUTER G1 X192.552 Y27.208 Z35.451 F2926.1 G1 X192.608 Y25.422 Z35.453 G1 X192.721 Y22.675 Z35.456

Pictures will follow!

[smartavionics]

That's all very interesting, thanks. One very odd thing is that in your last gcode example, the z heights on layer 115 are actually lower than the z values that come before??!!! How can that be?

[smartavionics]

I think the little backwards movement is because the nozzle position at the end of one layer is slightly different from the nozzle position at the start of the next layer. When printing a spiralized solid model (i.e. the wall lines are all exactly the same width) with the spiralized smoothing enabled, the coordinates of the last point in a layer should exactly match the coordinates of the first point in the next layer (that's what the smoothing achieves). But with variable line widths in action, the coordinates of the vertices are adjusted to account for the fact that the distance of the wall centre line from the wall's outer boundary varies as the wall width varies. The wall width is computed for each vertex and there's no guarantee that the wall width calculated for the last point in one layer will be exactly the same as the first point in the next layer and so the shift in the vertex coordinates can differ by a small amount. Of course, it only takes a small amount of backwards movement to screw the printer's motion planning, etc.

I think the above explanation accounts for the fact that the z-seam glitches occur in regions where the wall width is changing.

[smartavionics]

At this time, I am more concerned with the odd z values in your example above. Can you zip up and attach the whole gcode?

[TheTiEr]

It seems that some of those odd z-movements came from the Setting "adaptive Layer Heigths". I have overseen that this setting was still enabled.

Here is the gCode:
Test_Gcode.zip

[smartavionics]

Ah! Sorry, you can't use adaptive layer heights with spiralization.