nerys
Posts: 256
Joined: Tue Mar 29, 2016 8:47 pm

BUG report. repeatable vase mode inconsistency in pathing

OK this is an interesting one.

As you know I am vase mode. I love vase mode. but sometimes the slicers can do "strange" things to complex vase mode models. probably does not help that I am using tinkercad which can introduce micro flaws in models (that a slicer SHOULD ignore since we are talking issues smaller than 0.001mm most of the time. these printers can barely realized 0.1mm resolution anything literally the smallest radius you can form is 0.1mm with a 0.2mm nozzle.

Anyway I have this hollow ribbed fin. this is a "close" model (no inside island so no seam etc..) a much simpler model for the slicer to chew on for sure.

Anyway the TOP of the model refuses to slice properly. S3d is making "extrusions" that simply DO NOT EXIST anywhere in the model. what its doing "is not their" but its doing it anyway.

Here is the outside
001.jpg
See that hole on the side? that does NOT exist in any way shape or form in the model. there might be some defect but it does NOT represent "that" and here it is from the inside
002.jpg
See how it just erased part of the model and goes SIDEWAYS at an angle through the solid model. NOTHING is there. not like that at least.

it cropped the corner of one layer and totally bisected the next layer. this leaves either a gap or a compression in the model as it prints too high or too low as a result of the deviation from model geometry.

Now let me show you this.
003.jpg
PERFECTION. precisely what its supposed to do....... Seems only 3 images per post. next post
nerys
Posts: 256
Joined: Tue Mar 29, 2016 8:47 pm

Re: BUG report. repeatable vase mode inconsistency in pathing

Now here is that perfect slice from the inside
004.jpg
want to guess what I did? absolutely nothing. I changed NOTHING there is actually 3 of these fins on the build plate programmed to print sequentially. not 3 models. 3 identical models. I dropped one stl file and then did a copy and paste to get 3 and attached each to an identical duplicate process

originally all 3 had the defect. now 2 have the defect and ONE does not. want to know what I did to fix this? NOTHING. what I did should not have had any impact in any way shape or form in any part of the know universe.

THE ONLY thing I changed was to instruct it to "NOT" print the top 2 layers. I set a stop print at 101.6mm. that's it. I did that because the pointy tip tends to curl in and the model does not "seat" properly on the socket on the airframe as a result. so I figured to CLIP it in the slice instead of having to take nippers to it after the print.

For some reason doing that FIXED the irregularities in the slice for .... reasons.... I am hoping you guys can figure out the how and why (factory attached) note for some reason (again reasons..... I have no idea) the error is still their in the other 2

and yet the processes are IDENTICAL and the models are literally copy pastes of the same STL file dragged into S3D.
005.jpg
If anything about those models is not identical its a deviation S3D CREATED on its own for some reason. I dropped in a single STL file and used COPY PASTE. moved them into position on the platter and again did a COPY PAST on the processes and assigned one to each fin. that's it. that is all I did. and I can NOT figure out how to MAKE S3D slice the other two copies the same way. whats annoying the hell out of me is this gcode that fin is the LAST ONE. I wonder can I slice JUST that one and it still slice it right? let me try

Thoughts?

I would really love to see a dedicated vase mode slicing algo worked on. ie FORCE vase no matter what. orphan? delete it. Micro pinch or micro fraction or lesion in the model? FIX IT. jump the 0.001mm to the next "good part" of the model and just keep going (yes I know its not "that easy" from a coding POV) but I know its possible. PS does a pretty decent job of "fixing" errors it finds or invents on its own.

This is an error S3D can fix (it did) that should never have existed to begin with (in the slicing geometry ie whatever was causing it to make that error should not be causing it to make that error) and this example shows S3D can do it right. so WHY does it do it wrong sometimes? I don't know and I doubt the coders know either. I understand this is complex.

I am hoping this example might help them.... Please? :-)
Attachments
Ramjet Fins.factory
(236.08 KiB) Downloaded 28 times
nerys
Posts: 256
Joined: Tue Mar 29, 2016 8:47 pm

Re: BUG report. repeatable vase mode inconsistency in pathing

BTW that fin is the 3rd copy. I was able to remove the other 2 and it still sliced ok. I rotated the fin 45' and the error has returned.Z

Here is another factory file with JUST that fin and I can tell you how to TURN ON the error on and off like a light switch.

slice it? see the error? now exit back to platter and rotate the model -45 so its crosswise again instead of diagonal and now slice it again.

MAGIC the error is gone.....

Rotate 45' its back. -45 and its gone. so you can repeat this. I am really hoping this will help the coders fine the difference of what is going on.

NOW take the same orientation that works (crosswise) and flip it 180' and the error returns

45' the other way? still error -90 from cross? fail -135 from cross? fail but 45 from cross THE OTHER WAY so top of the fin now opposing the "x/y" start point and it works again. no defect.

These are not defects in the model. these are defect the slicing engine is introducing for some reason.

I tried all normal cardinal rotations. On X both ways on Y both ways and the 4 45' positions inbetween and ONLY 2 work. the rest fail.|
006.jpg
007.jpg
I have no idea why. I suspect if I "MOVE" the start point (which should have absolutely no impact on this but IT DOES) just like rotating the model should have no impact BUT IT DOES. will also change this.
Attachments
Ramjet Fins2.factory
(76.84 KiB) Downloaded 18 times
nerys
Posts: 256
Joined: Tue Mar 29, 2016 8:47 pm

Re: BUG report. repeatable vase mode inconsistency in pathing

I found other errors in the slicing of this model. where it seems S3D is taking shortcuts instead of following the shape of the model geometry. maybe these are a scale thing? size? every one of the tops of the ribs in the wing are wrong. and it "alternates" between the last "rib cut" layer taking a short cut (instead of coming back out and turning) and the layer below it "bulging" where there is no bulge. and it seems to alternate. cut bulge cut bulge from rib to rib. rotation does not appear to effect those they are a consistent geometry error. thankfully not fatal ie the model should print fine.
008.jpg
nerys
Posts: 256
Joined: Tue Mar 29, 2016 8:47 pm

Re: BUG report. repeatable vase mode inconsistency in pathing

Here is another one. where the top of the drawer starts to angle to the bottom of the drawer. then it "notches" in then out. that top edge was a point and for some reason it printed that fine but then screwed up somewhere else. I corrected that and fixing that fixed the somewhere else ???? why

But anyway this is another one where you can turn the error on and off. load the factory. slice it. perfect. not a single glitch.

NOW rotate the model 180 and slice it again. two glitches.

Rotate it back. no glitches.

Rotate it 180 and you get this
009.jpg
010.jpg
Attachments
Stackable Drawer.factory
(272.56 KiB) Downloaded 20 times
nerys
Posts: 256
Joined: Tue Mar 29, 2016 8:47 pm

Re: BUG report. repeatable vase mode inconsistency in pathing

Rotate it back 180 and its PERFECT
011.jpg
012.jpg
parallyze
Posts: 352
Joined: Fri Jun 05, 2015 4:18 am

Re: BUG report. repeatable vase mode inconsistency in pathing

nerys wrote: Tue Mar 07, 2023 3:21 am Rotate it back 180 and its PERFECT
Hmm... rotating an object in vase mode can cause various results, I'm not that surprised by this. Think about it, depending on
where you start the extrusion at the bottom in vase mode after possible solid layers it absolutely depends on where the
perimeter starts - so you will end up reaching certain points of a model at different heights, depending on the initial start of
the continuous extrusion.

I might have an idea what's causing the problems on the ramjet fin model from the first post, but I'm not sure if that's true but
convinced you won't like it. ;)

Need to verifiy this before posting more, here's a teaser:

Your model in 4.1.2 and vase mode:
m1_s3d412_thin.PNG
parallyze
Posts: 352
Joined: Fri Jun 05, 2015 4:18 am

Re: BUG report. repeatable vase mode inconsistency in pathing

Okay, this is gonna take some posts because of the pictures. Please note that I'm just saying what I've noticed, I'm not judging on any slicer/model files here. And also note I'm not using v5, so things might be very different from what I've seen using 3.0.2 - 4.1.2.
Other slicers used: PS2.5.0, Cura4.8.0, IdeaMaker 4.1.1

Assuming I've got the correct object orientation for the fin parts (remember, I can't open v5 .factory-files, they throw an error and leave the extracted model/process files behind) there's one thing I've noticed pretty soon:

Not all the cuts are reaching the top of the model. Therefore depending on layer height and starting point of the vase mode extrusion this will cause a problem:
m1_front1.PNG
Depending on layer height/start position (at what height this part will be reached) S3D will try to print that tiny tip which isn't connected
any longer:

Layer Height 0.20mm:
m1_s3d412_020.PNG
Layer Height 0.22mm:
m1_s3d412_022.PNG
You might not be able to replicate using the exact same layer heights. It's dependent on the starting position, model position, orientation and things like first layer height.

When slicing the model file each slicer has to do various checks. Imagine slicing a simple box with a hole in the front. At some point you will have slices of 0.20mm height, but the one where the hole does start might be open in the upper half, closed in the lower half. As far as I can tell slicers by default tend to check the center of the slice and decide based on that if the loop is closed or not at this point...

For the example above: The tiny tip might be at 0.08mm height of a 0.20mm sliced layer and get ignored - but rotating the part might translate that point to 0.11mm inside the layer, therefore making it relevant and printed, breaking the tool path for vase mode.

...continued in the next post...
parallyze
Posts: 352
Joined: Fri Jun 05, 2015 4:18 am

Re: BUG report. repeatable vase mode inconsistency in pathing

....continue...

Thing is, most of the cuts aren't very straight. They are very thin and running along curves. And right now I suspect some of the vertices
being at different heights causing these problems:

Here's a close up of the cut from the second "fin" from the left:
m1_slitol1.PNG
m1_slitol2.PNG
m1_slitol3.PNG
Because the vertices of the small 0.01mm cut aren't on the same z height I suspect the "decision making" while slicing causing trouble here.

There's another thing I've noticed. The distance on the inside from the cut to the outside wall is 0.9mm if I didn't measure this wrong. What I don't understand (this was already true for the pumpkin-thing) is why the perimeters are touchining each other using an extrusion width of 0.4mm. Even if those 0.4mm were printed at the outside of the model they should barely touch. Because the polygons/vertices in the model are almost everywhere on slightly different z heights this might have the same cause, but I'm not really sure about this....


(continued in next post)
parallyze
Posts: 352
Joined: Fri Jun 05, 2015 4:18 am

Re: BUG report. repeatable vase mode inconsistency in pathing

...so I tried to reduce the extrusion width. While this did have a dramatic effect in S3D it didn't really help
when using other slicers.

This is at 0.2mm extrusion width, 0.1mm layer height, s3d v4.1.2:
m1_s3d412_thin.PNG

Obviously printing like that isn't really an option. It does take much thinner layers to not cause gaps using such thin
extrusions, but it somehow confirms my guess about slicing tolerances within a single layer. In the end it can be seen
as a kind of "rounding error", one side of the gap being seen as "closed", the other one as "open".

Because model handling changed a bit in S3D over the years I looked at
it in 3.0.2, the oldest release I have at hand right now. At 0.48mm extrusion width
results did look better at first - but there's still some retracts to be found:
m1_s3dv302.PNG

Can you confirm this gap being around 0.9mm?
m1_cross1.PNG

If so, the perimeters shouldn't touch at 0.4mm width.... still they do. Reducing the extrusion width in other slicers did
"restore" the gap that should be expected, still they fail at reliably following the contours. Only PrusaSlicer doesn't seem
to care about all the little odds, for whatever reason... ^^

Return to “Troubleshooting and Bug Reports”