overhangs

[velvet oracle]

Based on this idea (https://fullcontrol.xyz/#/models/b70938), I tried creating a model with different overhang angles. However, I'm having several problems with it.

1. Both bamboo and orca don't create a stem, but bamboo shows it as gap filled (if I set default and outer walls to 0.4). Cura slices it correctly

2. Orca starts laying out the overhang filament in the air with nothing below. I tried with and without extra perimeters, polyholes, and different order of walls and infill.

3. as for I was trying to do, figuring the overhang speeds, I used the model from fullcontrol, and found values that worked. However, I'm having trouble getting those values to be used with the non-classic mode

[velvet oracle]

as per 1) the polyhole margin was too high. Had it at 5% and was causing issues with the thin stem. Dropped it down to 1% and that fixed the lack of the wall

[velvet oracle]

Finally was able to print it, however, it isn't very useable that way. I had to use close holes for slicing method, since it is the only way orca would start laying filament over the previous layer and not in the middle of the air.

For some reason the shallower overhang (60°) doesn't print well

shown below are

60 75 80
45 90

[velvet oracle]

trying to figure the speeds, I needed to determine the transition for the different speeds. Seems that anything larger than 58° is treated as over 75% overhang. Those below are interpolated along the different options, so it isn't a hard speed depending on the range.

however, the non-classic mode really doesn't use the full 75%-100% range, as it never gets close to 50mm/s

The shown objects are
40 45 50
58 60 63
90

[tepid walrus]

I am also struggeling with overhangs in 1.7.0 release
Will try the nightlys then

Any particulat you'd recommend

?

[velvet oracle]

the tests where on 1.8

[velvet oracle]

@unreal crater not sure if this is the new ensure vertical thickness algorithm, but finally got a non noodle print of the test patern I'm using (60°, see above). Seems that for the parameters I'm using have a crtical angle at around 58-63°. The wall either is too thin or orca makes it a double wall. Can print 90° however lol

Also, why do I need to use close holes, to be able to print the overhangs connected to the stem, and not in the air?

before

now with the improve algo

[unreal crater]

the new ensure vertical thickness algorithm has not been merged to master I believe, are you using the version from the PR?

[velvet oracle]

i'm on the nightly dev actions 1.8

[unreal crater]

Which branch?

There are dev builds for different branches.

[velvet oracle]

I get lost, but used this one
https://github.com/SoftFever/OrcaSlicer/actions/runs/6497570320

the application should have a build number somewhere

[unreal crater]

I see, then you are using the version with the new algorithm.

[velvet oracle]

thought you might find this interesting

think you are the one doing the coding for the new algo

[unreal crater]

Yeah. I am interested in the overhang work you are working on because my printer always produces worse overhang result than what it's capable of.

[velvet oracle]

the new algo is adding a second wall to the 58-63 angles, which it didn't do before. Guess this is why the 60° kind of printed

as for not being able to print overhangs, some parts print with nothing below.

[unreal crater]

I'll give the file a try this weekend and see if anything comes to my mind.

[velvet oracle]

thanks, this is the latest version. I been having trouble with bed adhesion, my pei sheet died, so I'm stuck with the smooth until the weekend

[unreal crater]

😲

Interesting.... Not sure what's going on with the 6th one...

Ah that's the only one with 2 walls.

[velvet oracle]

wow, very nice

On the 90°  I had to slow it down to 15-20 mm/s 😧 Maybe there should be a seperate speed for 90° overhangs, different from bridges. These 90° overhangs don't have a support on both ends which makes them a little different

Not sure if you have a higher angle that messes the walls. On my printer around 60° it starts making weird walls. Either leaving gaps or double walls. Though your 63° seems ok (certainly 1000x better than mine, no surprise there 😉 ), but does have 2 walls. My guess is that the angle causes the the wall to be too thin.

Is there any way to print these without having to use the close hole method?

[velvet oracle]

not as good as yours, but better than what I had before. Still can't get a decent 63° piece, which has an external overhang and an internal solid infill. The other 2 pieces have an outer wall and internal solid infill

I had to change several settings, grouped into 3 different reasons

1. Thin walls

Increased the line width for outer walls and internal solid infills to 130%
flow ratio to 1.2
reduced the print temp from 215 down to 205
disabled z-hop and wipe

2. Allow overhangs to print next to an existing wall, rather than into the air
   Changed slicing mode to close holes (for 90° overhang)
   wall order to infill/inner/outer (the 63° laid the external wall in the air)

3. Speed
   Reduced outer wall to 100 mm/s
   Internal solid infill down to 25 mm/s (the ~60° have this infill)
   Set external bridge at 20 mm/s

From this think I can conclude the following:
Probably the speed for internal solid infill and external bridge should be handled separately when the wall is inclined. Otherwise to print highly inclined walls, you would slow when printing horizontal surfaces.

Also no overhang should print away from an existing wall

[velvet oracle]

by the way, the new algo is doing a much better job albeit those 2 issues listed above.

Downloaded the new version, but didn't include your new algo

https://github.com/SoftFever/OrcaSlicer/actions/runs/6515168946

[unreal crater]

Meanwhile I'm also doing some overhang testing by adding the "alternating wall direction" from Cura & SS and the result looks promising

[velvet oracle]

wow, those look great. All the changes you been dong will allow to print much better, thanks 🙂

[velvet oracle]

just realize you already pushed a build. Will try it later, thanks

[velvet oracle]

my results are never even close to yours(, obviously 🥴 ), but saw that cura released 5.5b1 and gave it a try. Didn't play much with it, and dled the profile from elegoo discord.

Still their result was better than with yours (https://github.com/SoftFever/OrcaSlicer/actions/runs/6517595365). Not sure if I messed something and the comparison isn't valid. Cura had trouble with the 90° (each 10mm is 5° more starting from 45°). Even the underneath seems better

Used their concentric (connected polygons) for the top/bottom surfaces, since I hadn't used it and sounded interesting. The monotonic base on orca looks better because of that. The speed was a little slower on cura, and some settings don't seem to correlate directly

[velvet oracle]

just for completeness, I tried cura for the overhang pieces

[unreal crater]

Interesting... I wonder if Cura lifts z on overhangs... Can you see anything like that in Cura's gcode?

[velvet oracle]

don't see any z changes other than the layer

[unreal crater]

😤 Even more interesting then... Not sure how it achieved better overhang wall bounding then...

[velvet oracle]

Thought I had sliced the model with the same heights, but cura is doing a lot more steps

For the overhang cones, I'm not sure what I'm seeing. However, orca does no seams on those

[tepid walrus]

Just a Quick user question thrown in.
Could you tell me please what is the difference between the overhand algorithm used in 1.7 Beta and 1.7.0 release and respective 1.8.0-dev ?

I do experience quite some issues with overhangs and would be open to support with tests altough i cannot contribute on the coding side of things.

[velvet oracle]

noisyfox has been working on some improvements for 1.8 (you have to see that it shows his name on the build, not sure when these are integrated into the main dev code). Two changes would allow you to print overhangs (see the cones above). His latest work is on getting printing done in alternate directions every layer(?), so the print will be better (#1160899252340392006 message)

[tepid walrus]

Ok. So can i support this actions by printing test parts an reporting back. Or do you prefer to continue without additional user testing ?
(Still will need to figure out how to download the https://github.com/SoftFever/OrcaSlicer/actions/runs/6563445979 files for MS)

[velvet oracle]

You would need to ask one of the devs. I assume they want feedback, and tend to flood them with comments (not sure if they appreciate it lol)

If you find a build you are interested (there should really be a reference in the application), you click and if they build the binaries, you will find them in the bottom of the page

[tepid walrus]

Thank you 👍

[velvet oracle]

@unreal crater I just tried the latest alternate walls, and still laying filament without any support. The cone overhangs fails on the 90°, unless I choose close holes. Got a long print, so will try to print after it is done

[unreal crater]

Yeah I guess this won't help in this case. I'll see if there is some easy way to force extrusion start point.

[velvet oracle]

thanks 🙂

Either I'm doing something wrong, or there is a limit on overhangs of about 60°

The initial section is at 45° and each bump is 5° until you reach 90°

this was with a reverse threshold of 50% (default)

also coincides where orca starts doing overhangs

[velvet oracle]

just got the latest build 🙂 Not perfect, but couldn't print the middle row without the new algo

[velvet oracle]

I reprinted the with the new build, but still can't print beyond 60°