Missing Perimeter (1.1.0) #1950

Closed
4ndreas opened this Issue Apr 21, 2014 · 27 comments

Projects

None yet

5 participants

@4ndreas
4ndreas commented Apr 21, 2014

I have randomly missing perimeters at holes.
slicer-error

This happens in v1.1 and in the current master build.
In 1.0 all locks good.
The strange thing about is if I rotate the object by 90" or 180° I miss other perimeters.

System: Windows 8.1 x64
Slic3r 1.10 ex
strawberry perl current master.

https://www.dropbox.com/s/t54zs9l13elef5t/MotorHalter_0.stl
I don't beleave it's a config issue (send it a friend with the same results) but
https://dl.dropboxusercontent.com/u/1116168/Slic3r_config.ini

@alexrj alexrj added this to the 1.1.1 milestone Apr 21, 2014
@ledvinap
Collaborator

I can't reproduce it on github master on linux.
But there is config bundle attached - to fully reproduce I'll need exported config used tor slicing, not bundle.

@4ndreas
4ndreas commented Apr 21, 2014

Added (config)

@alexrj
Owner
alexrj commented Apr 21, 2014

I can't reproduce this one on current master and provided config file.

@alexrj alexrj modified the milestone: 1.1.2, 1.1.1 Apr 22, 2014
@magu
magu commented Apr 22, 2014

I have similar problems on the Mac (OS X Mavericks) with the 1.1.0 release from the webpage.

Using configs
https://www.dropbox.com/s/pt0xmgmyw9ajpfp/Baseline.ini
https://www.dropbox.com/s/1zfnjldugo1cxy5/Translucent%20PLA.ini
https://www.dropbox.com/s/fk0alpgqa20ecyv/K8200%20E3D%20auto%20level.ini
with the model
https://www.dropbox.com/s/1prk0s477miqytr/partA.stl
by running

$ /Applications/Slic3r.app/Contents/MacOS/slic3r --load Baseline.ini --load Translucent\ PLA.ini --load K8200\ E3D\ auto\ level.ini partA.stl
=> Processing triangulated mesh
=> Generating perimeters
=> Preparing infill
=> Infilling layers
=> Generating skirt/brim
=> Exporting G-code to ./partA.gcode
Done. Process took 2 minutes and 57.613 seconds
Filament required: 2756.5mm (19.5cm3)

produces
https://www.dropbox.com/s/o9g3d4xxc24e0n1/partA.gcode
skarmavbild 2014-04-22 kl 14 22 22

@magu
magu commented Apr 22, 2014

Haven't got the master to compile on the Mac yet. Will try and see if it's the same on that one.

@alexrj
Owner
alexrj commented Apr 22, 2014

In the mean time, 1.1.1 was released: can you test with that? Thank you!

@magu
magu commented Apr 22, 2014

Haha, I checked the slic3r.org page just before posting to make sure I didn't miss any new version ... =)

Re-sliced using the same command and input files as above with the following result:

$ /Applications/Slic3r.app/Contents/MacOS/slic3r --load Baseline.ini --load Translucent\ PLA.ini --load K8200\ E3D\ auto\ level.ini partA.stl
=> Processing triangulated mesh
=> Generating perimeters
=> Preparing infill
=> Infilling layers
=> Generating skirt/brim
=> Exporting G-code to ./partA.gcode
Done. Process took 2 minutes and 55.700 seconds
Filament required: 3012.1mm (21.3cm3)

(Interesting that it uses more filament ...)

Same missing perimeter in the gcode though
https://www.dropbox.com/s/v6jvjntcho64oqp/partA_1.1.1.gcode

@4ndreas
4ndreas commented Apr 23, 2014

I've still missing perimeters. Tested both x86 and x64 version of 1.1.1
Tested magu's part and got the same missing perimeter.
I use Autodesk Inventor for stl creation. This worked fine for the last 3 years.

@4ndreas
4ndreas commented Apr 24, 2014

vollbildaufzeichnung 24 04 2014 201108 bmp
vollbildaufzeichnung 24 04 2014 200801 bmp

I downloaded an old model of mine
http://www.thingiverse.com/thing:148054/#files (top.stl http://www.thingiverse.com/download:288175)

and there is the hole inner perimeter missing....

@justmyopinion

Not a solution to this bug, but turn your orientation 90/180 or any deg before slicing and problem magically disappear. I will leave any conclusion to the developer.

@4ndreas
4ndreas commented Apr 27, 2014

Rotation did't always help. I have models with a lot of holes and it does not matter what orientation (30,60,90,180 °) one hole at least is always missing.

@ledvinap
Collaborator

Using (nearly) master, I am partially able to replicate MotorHalter_0.stl - it produces strange G-code where it is retracting on each line:

G1 X74.284 Y82.411 F6720.000 E7.87859
G1 X74.169 Y82.779 F6720.000 E7.71530
G1 X73.995 Y83.122 F6720.000 E7.55208
G1 X73.767 Y83.433 F6720.000 E7.38877
G1 X73.491 Y83.701 F6720.000 E7.22552
G1 X73.196 Y83.905 F6720.000 E7.07329
G1 X72.827 Y84.085 F6720.000 E6.89916
G1 X72.456 Y84.189 F6720.000 E6.73591
G1 X72.074 Y84.232 F6720.000 E6.57264
G1 X71.689 Y84.210 F6720.000 E6.40936
G1 X71.314 Y84.126 F6720.000 E6.24612
G1 X70.957 Y83.981 F6720.000 E6.08284
G1 X70.629 Y83.780 F6720.000 E5.91961
G1 X70.339 Y83.527 F6720.000 E5.75635
G1 X70.094 Y83.230 F6720.000 E5.59306
G1 X69.902 Y82.897 F6720.000 E5.42984
G1 X69.767 Y82.536 F6720.000 E5.26656
G1 X69.693 Y82.158 F6720.000 E5.10329
G1 X69.682 Y81.773 F6720.000 E4.94006
G1 X69.735 Y81.392 F6720.000 E4.77678
G1 X69.842 Y81.049 F6720.000 E4.62453
G1 X70.024 Y80.681 F6720.000 E4.45043
G1 X70.252 Y80.371 F6720.000 E4.28712
G1 X70.528 Y80.103 F6720.000 E4.12390
G1 X70.823 Y79.898 F6720.000 E3.97163
G1 X71.192 Y79.719 F6720.000 E3.79751
G1 X71.563 Y79.614 F6720.000 E3.63425
G1 X71.945 Y79.572 F6720.000 E3.47100
G1 X72.330 Y79.593 F6720.000 E3.30771
G1 X72.334 Y79.594 F6720.000 E3.30606
@alexrj
Owner
alexrj commented Apr 27, 2014

Interesting, then there's some negative value being used in the flow math.

@alexrj
Owner
alexrj commented Apr 27, 2014

I tried to reproduce partA.stl on OS X using current master but I get correct results. I'll need to test it with 1.1.1.

@justmyopinion

downloaded recent ver 1.1.2-dev and tested ablove files, no success unfortunately with windows7/x64
PartA:
parta
Top:
top
DrrawBlnk:
drawblank

@alexrj
Owner
alexrj commented Apr 28, 2014

Then I'll need to test on Windows...

@alexrj alexrj added the Win7 label Apr 28, 2014
@ledvinap
Collaborator

MotorHalter_0.stl is triggering retraction is first part of each hole on (almost) master on Debian linux...

@ledvinap
Collaborator

Sorry, my fault ... the move is wipe going backwards for about 11mm ..

@alexrj
Owner
alexrj commented Apr 30, 2014

Okay, I'm finally able to reproduce it on Windows using MotorHalter_0.stl and current master.

@alexrj
Owner
alexrj commented Apr 30, 2014

It can be also reproduced using default settings.

@alexrj
Owner
alexrj commented Apr 30, 2014

Very interesting, this call fails even though the point is largely inside the polygon:
https://github.com/alexrj/Slic3r/blob/98e40d3fe462e7cf29d0bf61129bc283cd98833b/lib/Slic3r/Layer/Region.pm#L315

@alexrj
Owner
alexrj commented Apr 30, 2014

These coordinates fail:
POLYGON((207802834 -57084522,196528149 -37556190,173626821 -25420928,17128575
1 -21366123,118673592 -21366123,116332562 -25420928,93431208 -37556191,82156517
-57084523,129714478 -84542120,160244873 -84542120))
POINT(95706562 -57294774)

@alexrj
Owner
alexrj commented Apr 30, 2014

Fixed. Also covered by regression test.
One day we might want to use the PointInPolygon() function provided by Clipper for consistency...

I'll leave this issue open so that @magu can confirm his one is fixed as soon as 1.1.2 is out.

@justmyopinion

I have now today compiled latest master 1.1.2-dev windows version (W7/64) and tested all tree files as before (Drawblank-, Top- and PartA.STL) and they seem to be fixed for missing perimeters :-).
As a bonus I retested the issue with missing perimeters in #1859 and found this problem gone too.
Very good job and thankyou for your efficiency to go for the best. ;-)
justmyopinion

@4ndreas
4ndreas commented May 1, 2014

compiled the current build today, locks good.
Thank you very much.

@magu
magu commented May 2, 2014

We are golden on 1.1.2 on Mac OS X. Good job!
skarmavbild 2014-05-03 kl 00 32 31

@alexrj
Owner
alexrj commented May 3, 2014

Thank you everybody!

@alexrj alexrj closed this May 3, 2014
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment