Ultimaker / Cura

3D printer / slicing GUI built on top of the Uranium framework
GNU Lesser General Public License v3.0
6.18k stars 2.08k forks source link

Combing randomly deciding to travel over numerous gaps? #5121

Open Silver-Streak opened 5 years ago

Silver-Streak commented 5 years ago

Application Version 3.6.0

Platform Windows 10

Printer Anycubic i3 Mega AI3M_6-8_100.zip

Steps to Reproduce The issue SEEMS to randomly occur or doesn't occur depending on randomly toggling the following settings: Optimize Wall Print Order Infill Travel Optimization Combing (All/Not in Skin) Avoid printed parts

Actual Results Cura will travel across open gaps without retraction leading to stringing.

Expected results Cura should have the gcode set to follow the walls as it does on every other layer.

Additional Information I've been randomly running into bizarre issues with Cura and Combing. Sometimes it works perfectly as expected. Sometimes, slicing the exact same model (but with different temperature settings only) will have it add random travels in the middle of a layer that leads to heavy stringing.

I can't seem to pinpoint what exactly causes it, but I can usually get it to go away by disabling the 4 settings I mention above, then toggling one, reslicing, toggling the next one, reslicing, etc.

Here's an example layer, with the last 3 steps on the bar when you "play" the nozzle simulation. You can see it make to the end of the layer, but instead of traveling straight south to start the next layer (which it does on nearly every other layer), for some reason it travels to the middle of the bar, then travels south to start in the middle of the next layer.

6-8-100 last three steps layer 19

Here's screenshots from the previous gcode where I had untoggled and retoggled the settings. No weird mid-section travel. 6-8-101 last three steps layer 19

I'm trying to print models labeled "6" and "8" from this set: https://www.thingiverse.com/thing:2637201

I had also seen this in other prints but couldn't narrow it down (Printed a spool hub adapter, and for some reason the entire inside of the model was stringy to hell. Even with set to combing all, it was trying to randomly do travels over the middle hole).

Let me know if there's any other info I can provide.

Silver-Streak commented 5 years ago

I forgot to include an image of what the other layers look like. To see if it was the model itself, I reoriented both models, and the weird behavior did not move accordingly (but it did move).

You can see in this image how ever other layer is treated, then the errant layer 20 (instead of 19 now) 6-8-100 full model

smartavionics commented 5 years ago

I think there's two reasons why it's behaving like it is:

1 - you are using the sharpest corner z-seam alignment and that doesn't work well for models such as these. It's much better to set a user defined z-seam hint that forces the z-seam to be located at one end of the models as I show here:

screenshot_2019-01-06_07-51-15

2 - On layer 20, there must be a kink in the wall as it includes an extra vertex which gets used in the combing to jump from one part to another. Step through the lines on layer 20 and you will see it:

screenshot_2019-01-06_07-54-22

screenshot_2019-01-06_07-56-18

Silver-Streak commented 5 years ago

Thanks for the input. Any idea why the weird travel would move to completely different parts of the layer depending on when I loaded it/rotation? If it was an extra vertex issue, shouldn't the travel always be roughly in the same place?

smartavionics commented 5 years ago

Thanks for the input. Any idea why the weird travel would move to completely different parts of the layer depending on when I loaded it/rotation? If it was an extra vertex issue, shouldn't the travel always be roughly in the same place?

That's a good question. I'm not totally sure of how it decides to route when jumping from one part to another and my guess is that it maybe looks for the shortest distance. It could be that, depending on the position and rotation, sometimes the distances between parallel lines vary. They would only need to alter by a couple of microns and it could possibly affect the choice of which vertex to use as the start/end of a travel over air. Maybe.