Bug Cura tree support generate in air

[wdq4587 0]

Cura  version:  4.11

Model: Porco Rosso

thingiverse ID: 4871593

Scale: 70%

Picture is layer 113 to 114

 

Tree support is great, but not work this time......

Thanks!

[GregValiant 1,243]

That's a nice model.  It leaves no question as to why they were called flying boats.  I have to wonder why the designer left it in one piece instead of making it a glue together assembly.  Unless you are able to use PVA it will be tough to get the support out without breaking the struts and stays.

 

You need to play with the settings for Tree Support.  I thought it looked better with the Tree Support Branch Distance at 0.5.  Maybe someone else has a better take on it.  I'll attach a 3mf anyway.

GV_porco_rosso.3mf

[wdq4587 0]

I hate glue. But this post is not about discuss this. My problem is some tree support generated floating in air with nothing below or around. A FDM 3D printer can not print this way, even in space station with zero G. I attach my 3mf file. But using your 3mf file rotate 75 degree I also get same result.

porco_rosso 2.3mf

[GregValiant 1,243]

With the Support at 0% density there is no internal structure for the upper portions to build on.  It is just an outer wall.

With the Support Density at 15% and the Support Branch Distance at .5 I don't see a problem.

You have the Support Interface Density at 100%.  I thought 60% looked better and will likely come off easier.

Here is todays try...

GVPoccoRosso2.3mf

[wdq4587 0]

Thank you very much for your guidance!  But what I report is I think the tree support generation algorithm has bug. It should not generate such floating structure anyway, because it is a support. Although I know the tree support function is not stable. That's why need we report bugs.

 

BTW, I found Cura also did not display warning for such slicer result of the normal model. I mean some parts are printed in the air with nothing below or around. Manually inspection takes a lot of time. I suggest at least use some color to display such areas.

[Odracir1978 0]

Cura do the same with me...

And i have no idea what to do...

 

[GregValiant 1,243]

@nallath is there anything you can add here?  I have seen this behavior myself but I've been able to get rid of the problem by playing with the settings.

Yes, close inspection of the preview can take time but not near as long as re-doing a failed print.

[Odracir1978 0]

what you need to see?

i try to play with the settings, but it's a big and complex model so every "try to change something" takes me 3 houres of slicing... right now i think too it's a bug in the tree support function...

i'm not sure if my pc is to old, in the task manager i see cur try to get as much memory as it can, it going up to 13GB just for the slicing engine, but after a while its going down an start again going up slow, maybe this happens because its less memory for the slicing engine???

 

[vavaroutsos 0]

I just upgraded to Cura 5.1 from 4.13.1 and now I'm seeing this problem. The same settings weren't generating this problem with 4.13.1. See layer 132.

UM2E PGV19.2_0.4 Generic ABS lw 0.6mm lh 0.2mm if 100% ext1 245.0C bed 105C.3mf

[GregValiant 1,243]

Cura 5.0 is considerably different than the 4.X versions.  The "Tree Support Branch Diameter Angle" is too large for that model.  Where the support comes out of the trigger hole to support the trigger guard there isn't enough room with the setting at 5.0.  Try it at 3.0.

You might also want to change the "Tree Support Collision Resolution to 0.2.

The image below is layer 132 with those changes.  You can see that the supports growing out of the trigger hole are much more robust and are now connected to the lower portion.

When I import only the model I get the "Not watertight" warning.  The model needs repairs.

For myself, I would prefer to use "Normal" supports "Everywhere" on this particular model.  I think they would do a better job.

 

 

 

[vavaroutsos 0]

Thank you for the very helpful reply. Is there anyplace where subtle changes like this between releases are summarized?

[hydrocannibal 0]

I know its an old post, but for whoevers reading.....put a support blocker on that spot? boom, done

[Slashee_the_Cow 262]

49 minutes ago, hydrocannibal said:

I know its an old post, but for whoevers reading.....put a support blocker on that spot? boom, done

The idea is to not generate the problem bits, not manually fix each one. I've also found that sometimes a support blocker will just make it move to next to the support blocker.

[hydrocannibal 0]

On 1/25/2024 at 10:00 PM, Slashee_the_Cow said:

The idea is to not generate the problem bits, not manually fix each one. I've also found that sometimes a support blocker will just make it move to next to the support blocker.

Ah, I see, to: treat the underlying disease (as opposed to the symptom). so-to-speak (which wouldnt be a viable solution really anyways it sounds like).  excellent, thank you.