It would be great to have direct access for the value in the distance from object line:
my $distance_from_object = VALUE * $flow->scaled_width;
Although I know have Perl installed on my computer because of this ;-)
Can you supply documentation (pictures, discussion) about the usage case(s) for a custom value?
No discussion found anywhere, but I shall make some test prints and pics to show the problem of the extrusion of the support adhering to much to the perimeter.
The balance is now too delicate i.m.o. according distance of the support and the needed tweaks for the wanted extrusion (according width and extrusion multiplier)
Of course the theoretical value of a halve extrusion width should have worked, but the 1.5 * flow (is that the extrusion width of the default or support material?) is often to less according to the extrusion width used. I tweaked a lot but never had a combination with enough distance AND good extrusion of support and perimeter.
Also, a bigger distance would prevent the generation of support material on even tiny overhanging spots that would be extruded fine without support (so, an overhang that's only 3 x 0.4 = 1.2mm and also (partially) bridged won't need support imo.
First I shall try to proof my point, but it would also be great to be able to restrict the height of the generation of support until a layer or a set Z-height. But this is maybe too much a specific setting, although I use it much.
BTW: ABSOLUTE GREAT!! That you managed the merge STL files, with ALSO clever support generation! Absolute unique! Some first test pics attached:
Here some tests with the support structure, showing the adhesion in XY direction from the support to the model:
In general it is enough to increase the value of 1.5 * $flow->scaled_width; to 3 * $flow->scaled_width to prevent adhering horizontally . Maybe for this is not an extra setting needed, but I think a default setting of 3 is giving better results, also with a little too 'fat' flowtweak.
Leaves the desire to have a little more control over the support structure generation.
-More control over the distance of the support structure. With a 'main' support still generated, this can prevent support in tiny places: Those to small (and print ruining) support structures of just a couple mm overhang.
-A max Z value for support max it possible for 'fragile' objects to make a basement of a couple of mm. This only of course when it's possible to print the object further without support (with low layer height and high widths)
Especially now with multi-printing there are too much corners having to less support (see the 'floating' volumes of the pics above). No matter how you tweak the settings, there is always a corner unsupported. Unless the intensity of the support lines is set to 0.2mm maybe.
Imo, there are these options:
I would like to add my request to the above request to get control over support distance from vertical walls so the support only touched the overhang, not the vertical wall. I like the honeycomb support for it's stability, but it attaches to walls too much, if I put chamfers in the upper corners of overhangs I can get the support to leave a gap against the wall which is much easier to remove and still works well for support.
Top image is model in repetier, bottom image as sliced image. Note clearance between support and model wall. I would like a setting to be able to do this without creating the chamfer.
Thank you both for providing pictures and discussion. Your input is valuable and precious.
I really want to work much on improving support material, and I'd like to make it smarter rather than more flexible. The problem with customizable distance is that while it makes sense for vertical walls, it creates problems with other angles (imagine a sloping wall, for example a vase): if the distance is too large, there will be no contact at all between support and object. It's not easy for the end-user to compute the right distance for a given angle, and Slic3r currently lacks a graphical preview environment, so there's no viable analytic or visual way to calibrate that value.
I think we could work on some logic to have Slic3r automatically choose the best distance according to the shape of the input model. Your test cases provide a good basis for such work.
I tried a lot with the settings and various (gently and steep sloped) models with the " distance" value. I think for now it hasn't to be a setting, but the value distance to object just has to change slightly from 1.5 to 2 . This solved the adhering at horizontal walls and left more than enough support for overhangs of any kind.
Corners not being supported because of the ' static' projected support structure are a bigger concern though.
Happy to test in the future if you want. I'm an architect, mainly focusing on printing models exported from CAD software.
An update about this. 4d8757b contains a change that reduces polygon simplification and causes less contact of support with object (honoring the distance_from_object variable better).
Closing, please see #1539 for further discussion (and see PR for workaround/fix)