Skip to content

HTTPS clone URL

Subversion checkout URL

You can clone with
or
.
Download ZIP

Loading…

Support severely broken (crossing into objects, missing regions, etc) #1032

Closed
cakeller98 opened this Issue · 30 comments

2 participants

@cakeller98

I noticed this a few times while examining files for Issue #1013

Sometimes support paths will just cross right over an object through perimeters and infill.

I added 2 extruders for easier viewing - 1 for infill, and 1 for support. in the following image you can see how there's an errant support prancing through the part :( heh.

image

woah - interestingly... as you can see there are layers apparently completely missing support altogether.

image

I'm sure it's possible THIS model may have issues - not likely, but possible.

However, I've seen this behavior on the simplest and tested models.

@cakeller98

same view and model as above, only change was rectilinear instead of rectilinear grid... easier to see the layers that are missing.

image

the support going through the object makes support removal a very difficult process
the missing support layers make support just not work very well at all.

@alexrj
Owner

Can you confirm this happens with same files as #1013 (comment) ?

@cakeller98

Seems to... but I had to revert to an older commit to avoid issue #1048 still working my way back up to see where #1048 comes in... but

this using commit: 46f35e3
image

strange missing perimeters, and there are layers with support coming through islands too, see here:
image

@cakeller98

OK this was the commit where things went really bad... 06a592f

image

@alexrj
Owner

btw, what previewer is that? Repetier-Host?

@cakeller98

yup... i've set slic3r for 3 extruders (not that i have them) but... much much easier to see overlapping geometry when they're different colors.

I wish it could a) show negative extrusions, and b) retractions... then, I'd not need my hack on pronterface at all. :)

@cakeller98

OK now with (latest at this time) commit: 6058384

image

@cakeller98

those gaps in perimeters, my guess is something in that layer is going wonky and it stopps evaluating things.

Another wierdness, which I can't explain are those blue traces which are obviously a rendering anomaly but why? what movement is causing that, and is it expected?

theres another one in red near the bottom of the back object.

FWIW this is an extremely fast to set of test objects. slices in seconds but has enough meat to actually see things in the support and infill ;)

@alexrj
Owner

Uhm, the test.stl file from #1013 seems to only contain the middle one.

Can you upload the G-code of the last picture? I'd like to check whether it's a rendering problem caused by the special Z coordinates for interfaces or an actual problem.

(btw, thank you for helping)

@cakeller98

probably interfaces. I have 5 layers interface on my support.

YES happy to upload the stl files ;) and gcode if you like.

@cakeller98
@cakeller98

Sorry for the mis-communication here. The blobs are not part of this issue.

only the missing portions of layers - which are mostly support, but actually are, as you can see in the images above, perimeters, and now I've noted some layers that have partial infill as well... like the exporter crashes on a layer or soemthing and then just continues. OR some loop is prematurely ending. <- just guesses.

@cakeller98

And, of course, the original reason I was looking at this was because support was crossing into islands (see the left side near the top of the internal diamond shape in the back image? that "support" is completely meshing with the object.

image

@alexrj
Owner

I have fixed this one! Can you confirm it works on your side too? :)

@alexrj alexrj referenced this issue from a commit
@alexrj Bugfix: simplification of support areas could lead to complex polygon…
…s with bad orientation, thus causing wrong pattern clipping. #1032
dc0f706
@cakeller98

Support Crossing into islands does appear fixed! :) But what should we do about the other issue of incomplete layers? Do you want a new and separate "Issue"

@cakeller98

As far as I can tell this issue is fixed, and the rest of the issues listed here appear to be related to Repetier Host visualization.

Thanks!

@cakeller98 cakeller98 closed this
@cakeller98

Support is much better... but should it be missing below the diamond?

image

@alexrj alexrj reopened this
@alexrj
Owner

So, @cakeller98, I just tried to slice issue_1013_thinner.stl with the config.ini you supplied above (the one generated Fri Feb 22 15:41:16 2013) and I get the following output which looks correct to me:
Schermata 2013-03-15 alle 21 34 01
How did you get your last picture?

@cakeller98

That model isn't the same one. that is "thinner", whereas I sliced "Bevel"

@cakeller98

It must be something to do with the bevels?

EDIT: is it possible that there's a loop that's not executing (IOW an array or something that's giving up after only 1 object? or the loop using the wrong number of elements, or hitting a null ... never mind I'm just throwing out random guesses.

image

@cakeller98

OH weird!?!?! I thought... maybe it's a loop - so let me repeat the objects and see if it has the same problem on every same object. interestingly... it appears to be a positional thing?

image

@cakeller98

I just tried it with them rotated 45 deg, and get infill more places. then I tried 30, 60, 120, and 150 deg rotations, and get stranger results again.

The config file I'm using is the same one you just used, except I, like you did, set the extruders to 1, 1, and 2 instead of 1, 2, and 3 (perim, infill, support)

image

@cakeller98

Here is another test case (new simple test case to test 4 types of unsupported danglies. :)

a line in X, a line in Y, a point, and a square... and repeated 4 times - what's wierd is ... some get support, some dont. is it a memory problem? or something else?

I think this case is somewhat interesting so I'll upload the stl and ini and gcode.

image

@cakeller98

OR maybe this could help

I drew in the red lines to make the point.... appears like the support "pattern" is being rotated, but the origin of rotation is off or it's getting translated... OR something like that.
(you don't want this gcode file (75MB) hah.... and took a few minutes to slice, but glad I did.
:)

image

@alexrj
Owner

Good catch, @cakeller98. :)

@cakeller98

Above test case (probably need to set threads to 1, or at least lower than 4 to complete the build.

however at 4 threads, perl crashed after having written enough of the gcode to see the above issue - so it's much faster, and loads into gcode viewer easier anyway. :) maybe you want to keep using 4 threads.

https://gist.github.com/cakeller98/5145878/raw/f62db1e0d8f88e5fdefd698da6ec2cf5a3d2f118/ToadStools.7z

@cakeller98

Self Supporting RATIO

image

@cakeller98 cakeller98 closed this
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Something went wrong with that request. Please try again.