Weird striping effect


#81

In Cura, turn on “spiralize outer contour”, let it slice, then View-> Layer View and look at the side of the model. You will see the exact same positive and negative lines shown in the test prints. (this happens with Cura 2.x and 3.4 - I don’t have other versions handy to test)
The same deviations are present in the normal mode gcode, but they are small – however, further math applied to the deviations could amplify the effect (like that done inside the printer). Since this appears to happen in multiple slicers (Cura, SImplify3D), I suspect that a common library is causing the deviations while merging the 3D meshes for slicing. The visual effect looks a lot like it may be caused by problems with floating point accumulation ( ((a + b) - b) != a, when the scale of a and b are very different).


#82

Changing orientation changes the size of the layers and the “path order” of the layers (the blue travel lines can change position), so a reorientation can change the result. But i’m not sure if my theory/observations are correct, i’ll make a print of the moai_gliches.stl.


#83

With “spiralize outer contour” i can see 1 glitch:

@matt3o @ccox Can you provide me a gcode file that contains this gliches? Non spiralized please, the blender addon can’t handle that.


#84

Sigh. Of course the forum won’t let me upload gcode.

Non-spiralized:
https://adobe.ly/2tYeJAv

spiralized (just amplifies the effect):
https://adobe.ly/2tYf5aj


#85

Look at this picture

look at the banding from one support to the other. It can’t be residual resin (It’s about 2cm wide).

Also on the circled spot you can see how the banding is always in the toolpath direction. Also the support to the left of it does not present the same problem.

I might be completely wrong but I still believe this is a slicer issue.


#86

hey guys, we are following this and testing as well. On the acceleration settings of Cura, Moai ignore those settings. Moai only reads speed, coordinates and when to fire laser as documented on Wiki from Gcode. Galvo accelerates much quicker than the common motors found on z-axis so you can pretty much ignore the acceleration and considered it as instantaneous.

We don’t think this is resin either and the it could be a combination of slicer, and hardware factors. We will be test printing as well and will be sharing what we found.


#87

thanks for this. It may be best to stay away from 3.x code base since they are more focused on adding features and even break a few existing functions. on Moai github you can find 2.6.x code base


#88

don’t enable that. Moai doesn’t read those codes and you would be making gcode file bigger.


#89

In Blender i cannot see any issue in the non-spiralized gcode. In Cura it’s hard to tell because there’s no orthographic view and zooming is hard because the gcode is outside of the moai build plate.

In general i’m having trouble reproducing the issue in Cura, at exactly 45 deg rotation i get only 1 glitch with “spiralized outer contour”.
If i rotate the part a little bit (42.478… deg) i can trigger the bug:

But when i turn off “spiralized outer contour” i get gcode without visual imperfections:


#90

I was reading the raw gcode values, and finding variations in where it thought the face should be. In the non-spiralized gcode, the variations are smaller, but again, could be amplified by further numerical processing. The spiralized example was just a quick way to see that it really does happen in the slicer (something about their spiralize operation amplifies the variations - making them much more visible). And the variations will change depending on what model printer you are targeting - I chose a generic Prusa i3, because I have seen the same defects fairly often while doing FDM prints for mechanical parts.

I’m trying to narrow down which library might be responsible for the glitch, while waiting for Cura compiles to fail for various reasons (then fixing each new problem).


#91

@matt3o I was basing this assumption on these three picture for example (from the FB group, not mine)



you can see the support contact points and the layers are basically flawless. I’m asking this because I hope I can get my Moai to this kind of quality and to still have hope. I have other problems too, but If even after they are fixed, this would be the end point then I have nothing more than a paper weight consisting of my savings.


#92

the tank is well positioned and angled, the skull is too big, we would need to see closer pictures. But --again-- I have pretty spotless prints myself, it’s just that sometimes you can’t really get rid of those glitches because of the morphology of the model


#93

I was able to print with the latest slic3r (v1.3.1-dev). The result is still not perfect but definitely better in regards to surface glitches.

Don’t ask for the slic3r profiles :slight_smile: I had to do a lot of post-processing to make it work :stuck_out_tongue: and some more work would be needed for production, but at least we are now pretty sure it’s a slicer issue.

@peopoly hurry up with that slicer! :wink:


#94

Thank you for doing that – I had just downloaded slic3r (and it’s source code) to do a comparison.


#95

speaking of spaghetti code…


#96

I’ve dealt with worse (sometimes much, much worse). See my linkedin profile for more details.


#97

Here are my tests with Formlabs castable resin (FLCABL02):
100 µm layer height
65 mm/s
Z motor speed 3
PM motor speed 10
~ 6s pause before tilt (patched gcode with slow G0 move before layer end)


The print is curling a bit, but i don’t see lines at the start or end of the holes/protrusions.

The following one has this settings changed:
Z motor speed 8
PM motor speed 30
no pause


A lot of layer lines and uncured resin, but no lines at the start/end of surface features.

@ castable resin users: I’ll post detailed settings for FLCABL02 once i’m finished, the top prints were done at 35 °C resin/chamber temperature.


#98

as far as I can tell the problem is more visible at layer height 60µ or lower, not so much at 100


#99

I’ll try some test prints when i have figured out the settings for smaller layer heights.


#100

I never used castable resin, it seems to do a lot of warping, maybe it’s not the best resin for benchmarking