Cura tolerance/accuracy

[jd_3d 0]

I was playing around with Cura for the first time and slicing some of my STL files, and I noticed there are small perturbations layer to layer (as if the perimeters on each layer are shifted in x/y very slightly) even though they don't exist in the STL file. These don't show up when slicing the same STL in Simplify 3D or KISSslicer.

So, my question is, does anyone know if there is a tolerance you can adjust in Cura to keep the gcode very tightly to the original geometry? Something like KISSlicer's oversample resolution.

 

[gr5 2,167]

Interesting. I doubt it's serious but I could be wrong. Can you be more specific? Was it a cube? What was the difference exactly in mm?

Wait - was this infill that was different or the outermost shell?

 

[illuminarti 18]

Can you post an example STL and gcode file?

 

[jd_3d 0]

Hi Guys,

Yeah, I realize its a hard thing to describe. Here is a picture showing the difference between the gcode using Cura vs. Simplify 3D. Also, I have run my STL through netfab so I know it is clean.

http://i.imgur.com/y0deIDw.jpg

Is this typical with Cura? It is really the perimeters where the issue is (on a thin walled part).

 

[jd_3d 0]

I have an STL/Gcode file but I can't seem to find the upload attachment option. Is this because I am a brand new forum member, or am I just not looking in the right spot?

 

[illuminarti 18]

I think it may be because you're new... I can attach files by clicking the 'More Reply Options' button, which is next to the 'Post' button under the Reply box I'm typing this in.

That takes me to a screen that allows file attachments... but I'm not sure if everyone can do that, or just moderators. And even if everyone can, you probably won't be able to do it until you hit 5 posts, so make a couple more :-)

 

[illuminarti 18]

I think it may be because Cura isn't smoothing high poly-count parts, and instead faithfully reproduces lots of facets. I've seen something very similar on some of my high-poly parts, like 3D scans; the print ends up with a stucco texture, because the layers don't line up very well from one to the next, just as shown in your comparison photo.

 

[jd_3d 0]

Thanks, yes, I'm 2 posts away from restrictions hopefully.

 

[jd_3d 0]

I think it may be because Cura isn't smoothing high poly-count parts, and instead faithfully reproduces lots of facets. I've seen something very similar on some of my high-poly parts, like 3D scans; the print ends up with a stucco texture, because the layers don't line up very well from one to the next, just as shown in your comparison photo.

 

That is a very good theory, but I don't think it is the issue in this case. It is actually relatively low-poly. There is no variation in z, so it is a single triangle describing the z-height, so there should be zero variation layer to layer. Here are 2 more pictures of the simple model I will attach when I'm able.

http://i.imgur.com/l7uqZN8.png

http://i.imgur.com/5o7CqOD.png

 

[illuminarti 18]

Interesting ... It does sound rather like Cura may be introducing some noise.

[jd_3d 0]

I still can't seem to attach files, so instead I uploaded the files to Youmagine. Hopefully that works.

https://www.youmagine.com/designs/small-model-to-test-cura-slicing

I think you can get my settings from the gcode file, but let me know if you need an .ini.

 

[kenzu 0]

okay, this i kind of bad :-( Hope Daid will respond to this. But nice catch.

 

[Daid 305]

Photos?

The Repetier view says little. Due to the shading there, a tiny difference can look huge.

There is a slight rounding in Cura. To microns (0.001mm). But that's most likely not causing this. Cura also optimizes the path a bit by removing some points of the path that are almost in a straight line. Which could be causing it in this cause, due to the small curve you see.

More specifically, this code:

https://github.com/Ultimaker/CuraEngine/blob/master/polygonOptimizer.cpp#L25

I'm checking it out right now.

 

[jd_3d 0]

Thanks Daid, I appreciate you looking into this. Did you see this photo (it was earlier in the thread) comparing Cura to S3D?

http://i.imgur.com/y0deIDw.jpg

I'm using Repetier for visualization since I find its easier to see anomalies such as this. While the variations are small, they are definitely large enough to show on my actual prints. Let me know if you need any other photos or models.

 

[gr5 2,167]

I think it would be good to have a photo of your actual print in this same spot and compare it to the repetier host from the same angle and prove that the same lines are sticking in and out by the same amount.

I think Daid is skeptical as repetier host shows things that aren't visible once you go to print it.

 

[jd_3d 0]

Ok, to rule out Repetier as the culprit I opened them both in Simplify3D (which was developed completely independent of Repetier) and it shows basically the same thing. S3Ds gcode visualization is very good and exact so I trust it based on hundreds of prints I've done.

Here are the comparisons with 2 different angles shown:

http://i.imgur.com/HPlwjnn.jpg

http://i.imgur.com/xA89hKx.jpg

I have not printed this little test case (only my actual models, which have the same problem), but in either case I would probably need a good macro lens as its hard to show in photos what the eye can see easily. Plus, I don't want the discussion to detract into possible physical printer issues (i.e., if I had shown the actual prints first, people may blame settings, z-wobble, etc).

 

[Daid 305]

I can see the same in the Cura GCode viewer.

I know where it comes from, however, I haven't solved it yet. I might need to increase my resolution from 1 micron to 0.1 or 0.01 micron. Which is a bit complex. I've started on changing stuff for it. But I haven't reached the end of all problems yet.

 

[Daid 305]

I just found and fixed the problem, and the problem that arouse from fixing that problem. So now I have properly wavey lines that look good when printed.

Will be in the 13.03 release. Which will be the least buggy release ever! (that doesn't sound as good as it is... maybe I need to do better marketing...)

 

[Dim3nsioneer 557]

[...]

Will be in the 13.03 release. Which will be the least buggy release ever! (that doesn't sound as good as it is... maybe I need to do better marketing...)

 

I guess you know Murphy's law... It will destroy any least-buggy-marketing...

btw: what about the save/print icon (http://umforum.ultimaker.com/index.php?/topic/4487-bring-me-your-bugs-cura-1402-rc2-ready-for-testing/page-8&do=findComment&comment=41143) and the diameter2-issue? Any chance we get a bugfix for those in 134.03?

 

[illuminarti 18]

I added something to this thread, but it was nonsense, so I deleted it again.