Open VanessaE opened 8 years ago
I've noticed this in 1.2.9a too. It even does a weird thing where it skips the columns of support...
It is a good idea to improve anchoring of bridging perimeters. One has to be careful though about the appearance of the object's outer shell, if the "bridging perimeters" is used just for overhangs. Starting the bridging flow a bit earlier will likely cause marks on the object surface.
This has been left to lie fallow for far too long, can someone PLEASE fix this?
I am happy to support people trying to fix this; I have zero time myself to work on it.
It seems to me that it would occur less often with thicker perimeters, basically reducing the flow disparity between the two types of extrusion.
Version
git commit 223065221888e202a762bfeaff125f960896b009
Operating system type + version
Debian
Behavior
As it's laying out the perimeters of a bridge, Slic3r starts the thick parts too far from their anchor points. Each strand of the bridge starts with a very short length from the end of what looks like the normal extrusion rate for a perimeter, then it switches to the bridging flow ratio, and then eventually back to the perimeter flow rate just before reaching the other anchor point. Both lengths of perimeter-flow-rate-in-the-bridge are a few tenths of a millimeter long, so you get a thick strand with weak anchors, causing uneven stretching of the bridge, leading to drooping and often breaking at the starting end (at least in the case of PLA). No amount of cooling or speed control has been enough to completely work around this.
This only seems to affect perimeters, especially external perimeters, where the difference between bridging flow and perimeter flow is greatest. Solid infill seems to be fine in this regard. A thinner overall layer height seems to make the problem even worse, probably because that short anchor gets ever thinner, and causes an increasing disparity with the effective bridging flow rate.
STL/Config (.ZIP) where problem occurs
For a test model, use this bridging+retract/ooze test object, http://www.thingiverse.com/thing:15087 or you could use triffid_hunter's bridging torture test, http://www.thingiverse.com/thing:12925 (pictured above, with a bridge flow ratio of 0.8). Ironically, the bridge/ooze model is more of a torture test for this issue than is Triffid's model. Config files don't matter here, as it's consistent across all of my print, filament, and printer profiles, but here's my config bundle anyway: Prusa i3 - Slic3r 1.3.0-dev config bundle.ini.zip