Bridging causing unneccessary inner walls to be printed first (mid air)

[InstantMuffin]

Cura Version

5.12

Operating System

XUbuntu LTS

Printer

CR10SPro

Reproduction steps

See the file and video.

Actual results

Printing starts with the wrong wall order, printing the completely unsupported inside part of the outer shell first. Causing the entire bridge to fail later.

Expected results

Printing the outer shell from the outside first, where it is supported. Not printing any inside walls on this outer shell, and then bridging across.

Add your .zip and screenshots here ⬇️

BridgeTestWallOrder.3mf.zip

BridgeTestWallOrder.mp4

And this is what it looks like over-exposed. You can see the diagonal skin bridging successfully attaching to the walls of the inner circular structures, but then failing because the inner walls they should attach to were printed mid-air and so were not existant.

[HellAholic]

With the profile settings that outcome is not possible.
You can either use a cutting mesh to set the Wall Line Count to 1 and Extra Skin Wall to 0 for the initial layer (you would need to have a hole the size of the middle part to keep the 3 walls,
Or you can use a plugin to basically splice the model (Slice -> split -> join back)
Then you can get results like this:

[InstantMuffin]

With the profile settings that outcome is not possible.

But that's the thing. Shouldn't Cura be able to recognize this and produce usable output? Hence me filing this.
I previously filed an issue about slicing not completing but stalling indefinitely. The issue lied within the randomize infill setting. Would the reply there be also the same as yours? No.
It is not a bug, as Cura apparently does not have that functionality. But should it have it? Therefore the tag you chose is correct.
It is unwanted, wrong behavior, that can be currently mitigated with a lot of magic dust and trickery.

[HellAholic]

It gets into the depth of what is a bug and what is a feature gap or even expected behavior.
In terms of the randomized infill, that one is a bug, because it is behaving outside of the designed and intended parameters.
For bridging, I would have agreed with you that it is a bug, if your model and settings would allow for that bridging but Cura is behaving outside of those parameters. We have several improvement tickets for bridging on the backlog and I will take your model as input into those discussions regardless.
But in this case, we don't have the feature to treat the walls in the bridge layer separately from other skin layers (first feature gap), the model is extremely thin in the lower part which makes it hard to anchor, and you have certain settings changes like the wall count and extra skin wall which at the moment based on the software design take priority over bridging, and the application is following those the best way possible with the current capabilities.

So TLDR would be: Currently, with the implemented functionality of bridging, the expected behavior as you described it, is not possible; but we're working on improving the behavior in near future 🤞.

Hope that clear things up. Honest constructive feedback is always welcome.

[GregValiant]

Part of the problem I see is the model itself. The bridging starts right where the single wall lower portion becomes wider. At that point it picks up the second wall of the thicker upper "floor" portion (that's why it is green "Inner Wall" in your screenshot) which is not supported by any structure below it.

Here I've cut a section through the model. There is nothing but air under the green and yellow "extrusions" I drew in. If there was a chamfer in the corner, things would probably have worked out.
That's a tough little model to slice either with bridging or supports. If the lower portion had 0.80 walls then things would have worked out.

If the wall is 0.8 wide then the problem at the outside seems to go away, but the single wall holes (on the right in the image) are still a problem.

Another thing I noticed is that you have "Hole Horizontal Expansion" enabled but you did not limit the hole size with "Hole Horizontal Expansion Max Diameter". Consequently, there will be a shelf that will be even with the bottom of the side hole and another even with the top of the side hole. Set HHMD to 25mm and the shelf artifacts will go away.

[GregValiant]

This is using the "Banana Split" plugin and with the model -8.00 into the build plate. Super glue the two pieces together. No bridging required.
This looks really good with a 0.20 nozzle and that is the way I would print it. 1:30 print time as opposed to 45minute print time with the 0.40 nozzle. The 0.20 nozzle gives a much better form to the threaded post.

You can split the model in other editing software and print the two halves to get the same effect.

You can use the "HellaFusion" plugin from the Marketplace (as HellaHolic himself so delicately mentioned) to change the settings at a height.

[InstantMuffin]

Upon seeing your last screenshot I notice there is a misunderstanding. The outer, hollow rings, and the filleted rectangle shell are designed supports, meant to be broken off. This is also why they are so narrow.

Here is the original model along with the project file that can be inspected and played around with with the latest FreeCAD weeklies at the time of writing.

CameraHolder_CuraTest.FCStd.zip

This is supposed to be, in theory, an example of why with non-natural shapes designed supports are superior in terms of material and time cost, print reliability and visual result. If it were to be sliced "correctly".

[GregValiant]

This is with banana split and I removed the skirts.

They would need to be assembled but they would certainly print without issue.

If it were to be sliced "correctly".
I would take that as "If it were sliced like I want it.". It is slicing correctly based on Cura's design. I think that is why HellaHolic is on the fence between "bug" and "feature request".

I wrote a post-processor for models with large flat overhangs. It is similar to Pause at Height but it keys on the "support interface" as well as the layer. There is an initial pause and you change material, then, at the end of the "interface section", there is a second pause and you switch back to the model material. It works quite well, but you have to be there to service the printer. In addition, too many instances in a single print and the constant material changes get real annoying real fast.

"Support Interface Material Change" is not an "open it and use it" script. It's a more advanced process that takes practice. It does allow a single-extruder printer to be a dual-material printer for those models that qualify (large flat supported areas)

This is the lid of a plastic container. Because there is a handle on top, it needed to be printed with the inside facing the build plate. The model is PLA and the top layer of the support-interface had a 0.00 air gap, 100% zigzag, and was PETG. It came out really good. It's very hard to tell that this was the supported side.

I'll quit typing since I've finally gotten to insert the two nuts into my current print, and now it's off to celebrate St. Patrick's day at my local Irish Pub.

[InstantMuffin]

I'm aware of the possibilities of how I could possibly split, glue, or redesign the object. I could also print it with slicer generated supports. I found my approach to be the most reasonable from a design standpoint, and so do other sources focusing on designing to print for manufacturing (i.e. slant3d and many other). I recognize however this is not something Cura accounts for by design at the moment and it appears there is at least some hesitance to even want to adapt this behavior. I'm leaving this open for you to possibly close or keep around for discussion, the project is provided for anyone interested to tinker around with. Thank you for your time and insight.

[GregValiant]

It will get folks thinking that's for sure.
BTW, I tried slicing it with Orca and PrusaSlicer and I got the same results as with Cura. It's one of those situations that no one seems to have considered.