GCode (and Print) Differs from Preview

[voss008 1]

Hello everyone,

 

I'm having an issue with Cura 4.4.1. When I slice a file to print, I typically check the preview to ensure there aren't any issues. This print looked fine upon initial slicing. However, I noticed midway through the print there are repeating artifacts across the surface. They appeared to be too regular and too evenly spaced to be attributed to a mechanical failure on either x or y axis.

 

Upon reloading the Gcode previously saved, it appears the Gcode itself displays these artifacts, demonstrating what I believe to be some sort of software issue. I am hoping it is a simple fix to a setting I am unaware of, although I worry something is amiss with Cura itself. I am uncertain.

 

I have uninstalled and reinstalled Cura and the issue persists. I have recently checked my belts, calibrated my e-steps, and my printer is mounted securely to my desk (which is bolted to the wall) so I'm confident it is not related to over-extrusion or ringing/vibrations. I am printing on an Ender 5 using Cura's standard quality (0.2mm) settings. I have tried adjusting a variety of settings, slicing, saving, and reloading the Gcode to no avail (the artifacts persist, even using a variety of STLs - so it is not simply a bad model). I am no expert, but it would appear these artifacts only appear at the vertices of the model itself.

 

In further attempts to narrow down the issue, I tried using Slic3r software to slice the model and then load the Gcode from Slic3r into Cura, following which the artifacts are gone. So it must be something on Cura's end, right? I apologize for my ignorance in this matter. I am still only a few months into 3d printing.

 

I have attached relevant screenshots and a picture of the completed print.

I was hoping this excellent community might have the answer. Any help would be greatly appreciated!

 

 

 

*Edited post to reorganize the order of the images.

Edited February 2, 2020 by voss008

[gr5 2,167]

This is interesting.  Please post the 3mf file (do file save) so other's can see exactly your model and settings and such.

[voss008 1]

Absolutely!

 

Attached are both the 3mf file and the sliced Gcode.

oddish.3mf oddishG.gcode

[burtoogle 513]

I can't say for sure what's causing the problem but I have two suggestions:

 

1 - increase the value of the Maximum Resolution setting to something like 0.25, that should remove the short line segments that are in the gcode. Sometimes, very short line segments can cause visible artifacts.

 

2 - either reduce the number of walls to 2 or completely turn off the Compensate Wall Overlaps setting. The overlap compensation is buggy and can introduce artifacts.

 

Hope this helps.

[voss008 1]

Hi there,

 

I just tried both suggestions, first separately and then together. Unfortunately, I am still seeing the same results.

 

Here are the results from reducing the maximum resolution:

 

0.25mm

 

0.5mm

 

It seems to smooth them out a bit, but the artifacts are still present.

 

Thank you for the suggestions!

 

 

Edited February 4, 2020 by voss008
Fixed image links.

[voss008 1]

Update:

 

Upon looking closely, I noticed there are strange variations in line thickness throughout that seem to be contributing. I hope this provides further insight.
 

Edited February 4, 2020 by voss008

[voss008 1]

Update:

 

I tried again on my Mac (running macOS Catalina, Version 10.15.2 (19C57)) using Cura version 4.4.0. It produces the same results as well.

 

Am I the only one able to reproduce these issues? Is it user error? I'm using the standard quality profile with no setting alterations. This is nuts.

[gr5 2,167]

@voss008 - your screen shots didn't show up in the forum.  Could you please try again?  Something about the process you did... well it didn't work.  It worked okay for your original post - maybe you can do whatever you did there?

 

Ideally you could go back and edit your posts and fix them.

 

[voss008 1]

Thanks for the heads up. I have edited the posts. Please let me know if the images are still not visible.

[gr5 2,167]

Okay, now that I see the image I'm intrigued.  Unfortunately, you did not post the project file (which is also .3mf).  In cura choose "file" "save...".  This file will contain your printer settings, your profile, your overridden settings, your STL file, how you scaled it and positioned it, everything I need to duplicate what you did and look at the options chosen.  Such as "coasting" and "wipe" which are the 2 I will be looking at initially.

[voss008 1]

Did you need the .3mf files from these 'Maximum Resolution' altered settings? I'm happy to do so, I just want to clarify.

 

The .3mf I provided in my second post (which I have also attached to this post) is the baseline file I'm using to produce these results (which, again, are just the 'Standard Quality - 0.2mm' settings in Cura).  Likewise, the screenshot in reference to the line thickness variation (where the model is blue) came directly from the .3mf file provided.

 

To be clear, my order of operations to initially produce these results were:

 

1) Open the .STL.

2) 'Slice' the file using the 'Standard Quality - 0.2mm' settings (from which I took the first screenshot)

3) 'Save to File' as .gcode.

4) Clear the build plate.

5) Navigate to File -> Open Files -> [select the saved .gcode from step 3]

6) Once it has loaded, inspect the results (the second screenshot I provided); here is where I see all the various defects and have provided

 

I apologize if I haven't answered your question appropriately, I hope this has clarified my process. 

 

 

 

oddish.3mf

[gr5 2,167]

There are too many different versions of cura.  I have a custom version that probably only 100 people use (the MB version, the burtoogle version).  Each version has different tweaks to the profiles.  Ultimaker makes changes.  Plus it matters what kind of printer you have.  I probably don't have your machine settings.  It's just much easier if you could save your project file please.  If I'm going to spend 30 minutes playing around I want to start with your exact settings.   The best way by far is to get a project file.  There are other people on this forum I can get interested in your issue if I can't figure it out but not without your project file (well I could give it a shot and create one but I'd rather you do it).

 

[voss008 1]

Great explanation. It makes perfect sense.

 

So... is what I've provided something different than what you're asking for? I've attached the .3mf project file to two of my posts so far; are those attachments not showing up?

[gr5 2,167]

I see the 3mf files but when I opened them in cura I just got your model.  No settings.  I just tried again.  

 

In cura in the top left choose "File".  Then choose "save...".  File type should mention "project".

[voss008 1]

I'm sorry. I don't know what else to say. That's exactly what I've been doing.  Could it be related to the fact I haven't changed any settings from the default? I'm really sorry. I'm not trying to be difficult. I'm following your directions to the letter. I open the model, apply the standard settings, and hit file -> save.

 

I've attached another .3mf, following your directions (and also tried the export setting too, which also appears to provide a .3mf file). I've attached both .3mf files; the 'saved' version as oddsave.3mf and the 'exported' version as oddexport.3mf.

 

When I open the oddsave.3mf, it asks if I want to open it as a project or import the models (which is what all the previous .3mf files have asked as well), so I assume that is what you're referring to regarding including the settings in addition to the model.

 

Edit: I took a screenshot of what my save confirmation page looks like, which, again, I've been doing since my first post. I assumed this would save the settings too, but I guess I could be wrong...

 

oddexport.3mf oddsave.3mf

Edited February 6, 2020 by voss008
Included screen grab of save process.

[gr5 2,167]

Well that's very strange.  I guess we have incompatible versions of cura.  I've done this with other people but never had this problem.  Maybe I don't have your machine settings?  I don't think that's it - I've done this with people who had all manner of printers (that I never even heard of) and it always seems to work.

 

Well I:

1) Sliced your model with default (0.2 layer) settings and I didn't see the same thing you saw when reloading the gcode file.  I saw it ever so slightly but not nearly to the extent you see it.

2) I loaded your gcode file into cura and I see exactly what you are talking about. My first thought is that this is just a rounding error as the amounts of extrusion on these tiny lengths are rounded to the nearest 5th decimal (1/100,000 of a mm) - here's a line from your gcode - the E value

G1 X65.119 Y64.052 E0.03559

 

But 0.00001mm should be good enough to get a good line width.

 

3) So I loaded your gcode into a different (better) viewer (repetier host) and found the exact gcodes you were talking about and loaded them into excel.  You can search your gcode file for "3317.777" to find the spot I mean.  Then I loaded these gcodes into excel and used simple math to find the distance the print head moved and the extruder moved and took the ratio and it was the same within 1% for every segment.

 

So the issue is with the crappy crua gcode viewer.  Not the gcode itself.

 

 

[gr5 2,167]

So I just reread your original post.  You mention a "repeating pattern".  Yet you circled something different in the gcode view above.  So which issue do you care about?  The "repeating pattern"?  Or something else?

 

The repeating pattern is related to the model - you can see it in your model in PREVIEW mode in cura.  So the actual STL file doesn't have the ability to represent rectangles - only triangles.  So each square in your model is at least 2 triangles.  Those triangles come to a point at the vertexes and you have very short line segments there.  So your printer is probably slowing down at those double vertexes where 4 squares come together.

 

I circled one of these "double vertexes" where you can have a VERY short line potentially a thousandth of a millimeter long.

 

Something is happening here - it could be the slicer but I'm guessing your firmware (marlin) on your printer doesn't handle very short line segments well.  Usually cutting the print speed in half will greatly improve things like this.  You could try in cura setting the skin speed half as fast as it is now - and keep the inner speeds not much higher (certainly no more than double).  Also it might help to print "outer before inner" (another cura setting).

 

 

[voss008 1]

53 minutes ago, gr5 said:

Well that's very strange.  I guess we have incompatible versions of cura.  I've done this with other people but never had this problem.  Maybe I don't have your machine settings?  I don't think that's it - I've done this with people who had all manner of printers (that I never even heard of) and it always seems to work.

 

Well I:

1) Sliced your model with default (0.2 layer) settings and I didn't see the same thing you saw when reloading the gcode file.  I saw it ever so slightly but not nearly to the extent you see it.

2) I loaded your gcode file into cura and I see exactly what you are talking about. My first thought is that this is just a rounding error as the amounts of extrusion on these tiny lengths are rounded to the nearest 5th decimal (1/100,000 of a mm) - here's a line from your gcode - the E value

G1 X65.119 Y64.052 E0.03559

 

But 0.00001mm should be good enough to get a good line width.

 

3) So I loaded your gcode into a different (better) viewer (repetier host) and found the exact gcodes you were talking about and loaded them into excel.  You can search your gcode file for "3317.777" to find the spot I mean.  Then I loaded these gcodes into excel and used simple math to find the distance the print head moved and the extruder moved and took the ratio and it was the same within 1% for every segment.

 

So the issue is with the crappy crua gcode viewer.  Not the gcode itself.

 

 

 

 

1) I'm not sure if I mentioned it earlier, but I did try uninstalling/reinstalling Cura (as well as both 4.4.0 and 4.3.0 versions) to verify if something on my end had been corrupted. No such luck, still saw the same results. Would it be worth trying another version of Cura? Is the version you're running modified or experimental? Is that accessible to the public?

2) Wow, what an in-depth analysis. Thank you so much for taking the time to do this. Very interesting results. I wonder why the .gcode appears fine in Cura's gcode viewer when sliced with another program but with the artifacts when sliced with Cura? Very strange. If you're interested in any further analysis, I can provide the .gcodes I tried when sliced with Slic3r and Prusaslicer; they appear with fewer artifacts when loaded in Cura's gcode viewer (despite being the same model), although that would make sense given the variations between slicers and settings. Also, given that I'm much more familiar with Cura, I'd much rather continue using it over switching to another slicer.

 

 

44 minutes ago, gr5 said:

So I just reread your original post.  You mention a "repeating pattern".  Yet you circled something different in the gcode view above.  So which issue do you care about?  The "repeating pattern"?  Or something else?

 

The repeating pattern is related to the model - you can see it in your model in PREVIEW mode in cura.  So the actual STL file doesn't have the ability to represent rectangles - only triangles.  So each square in your model is at least 2 triangles.  Those triangles come to a point at the vertexes and you have very short line segments there.  So your printer is probably slowing down at those double vertexes where 4 squares come together.

 

I circled one of these "double vertexes" where you can have a VERY short line potentially a thousandth of a millimeter long.

 

Something is happening here - it could be the slicer but I'm guessing your firmware (marlin) on your printer doesn't handle very short line segments well.  Usually cutting the print speed in half will greatly improve things like this.  You could try in cura setting the skin speed half as fast as it is now - and keep the inner speeds not much higher (certainly no more than double).  Also it might help to print "outer before inner" (another cura setting).

 

 

 

 

3) Ah, yes. I can see why you'd be confused. I am concerned about the repeating pattern, but the pattern itself seems to be linked to the region I've circled (in that the gcode appears to have variations in line width that contribute to those repeating pattern artifacts). I mention the vertices as it seemed like (to my untrained eyes) that these line width variations contributing to the artifacts were appearing at the vertices within the model. I'm happy to isolate another instance of the line width variation issue directly within the repeating pattern if that helps.

 

4) I'm not familiar enough with Marlin to be able to say empirically it was either slicer or firmware. You could be right, it could be on Marlin's end! - but again, I just found it suspect that it would be such a coincidence that the gcode would demonstrate the artifacts and the printer would reproduce them accurately. That - to me - seems like a slicer issue. But, again, I'm not super familiar with the firmware.

 

5) I will try to reproduce this vertex artifact issue on a model that a) isn't so large and therefore b) doesn't take 18 hours to print and verify if print speed directly affects the outcome. That might be able to narrow it down between slicer issues and firmware/mechanical issues. I will present those results when I've had a chance to do so.

 

6) Is there a way of adjusting slicer settings to round away these short line segments? Is that precisely what the 'maximum resolution' setting attempts to do? Can Cura reliably fix these issues or would this be an instance where loading the model into CAD/meshmixer/etc. be appropriate?

 

Also, I just wanted to take a moment to say thank you for taking the time to address this problem and answer my many questions. You have been extraordinarily helpful and patient and I can't tell you how much I appreciate it!

[gr5 2,167]

On 2/6/2020 at 12:41 PM, voss008 said:

 

 

 

 

5) I will try to reproduce this vertex artifact issue on a model that a) isn't so large and therefore b) doesn't take 18 hours to print and verify if print speed directly affects the outcome. That might be able to narrow it down between slicer issues and firmware/mechanical issues. I will present those results when I've had a chance to do so.

 

I like this attitude.  My conclusion was that the bumps you see in cura gcode view are a gcode view bug and so are unlikely to line up with the same bumps on the printed model.  If you can show 5 out of bumps line up *exactly* I'd start getting interested again.  But the data I looked at where you circled some line segments - that was definitely a bug in the gcode viewer and not in the gcode.

 

I've seen these bumps before.  They probably have more to do with printing speed and print speed changes and other settings in cura.  I don't get the bumps so much - or if I do I hardly notice them as I am more printing functional parts.  I know very many people get these and the universal solution is to print slower but I've seen many other people get rid of them but I've never tried any of the solutions so I forget what people did.

 

On 2/6/2020 at 12:41 PM, voss008 said:

6) Is there a way of adjusting slicer settings to round away these short line segments? Is that precisely what the 'maximum resolution' setting attempts to do? Can Cura reliably fix these issues or would this be an instance where loading the model into CAD/meshmixer/etc. be appropriate?

 

 

So Marlin has a 16 line segment (or gcode) buffer.   It looks ahead 16 moves.  Having 16 very very short segments all within 1mm or even 5mm can really mess things up and slow the printer down to a crawl where it overextrudes on the slow down because the head still has pressure.  But you only have 1 (one!) short move at the vertex and everyone has these - you can't break it up into triangles without having a few triangles come together at one point like this.  On every model ever.

 

But still - you seemed to have bumps in your print right near these vertices and that's all I can think of (that your marlin sucks, lol).

 

Yes you ask a good question about "maximum resolution".  Cura will eliminate extra moves and maybe it removed a point but I don't see how that would mess things up.  Or maybe the other slicers did a better job at removing the extra line segment.  One would have to look at the gcode carefully (I really recommend repetier host as you can see the 3d model and the gcodes side by side and when you highlight gcodes it highlights them in yellow in the 3D view).  Still it would take a few hours to compare slicers.

 

Maybe a better question is how is the quality of prints at the same print speed with different slicers and with actual photographs of the prints.