Adaptive resolution slicing - shell sub resolution

[pm_dude 27]

Hi all,

I was wondering if anyone ever considered making an adaptive resolution of the slicing of an object.

Lets say you slice the object at 0.2mm layer heights. On could consider making the shell with slices of 0.05mm making it 4 times more precise on the shell than on the infill.

If could cut down print time considerably and gain great quality?

Does that sound like a retarded idea?

 

[anon4321 16]

I don't think it's a retarded idea but I don't think it will work either.

If I understand the suggestion, you propose printing the shell at 0.05mm while printing the infill at 0.2mm.

The problem is that the infill must meet up with the shell. Think of the edge of the infill as a right angle that exactly vertical and is 0.2mm high. This right angle edge must be next to the shell so that the shell sticks to the infill.

Given the infill's right angle and given that it must be next to the shell, the shell is basically constrained to be exactly the same for each layer that makes up the shell.

In your example, the 4 layers that make up the one infill layer need to be exactly the same.

The result is that you don't really get the increased resolution.

One way you could work would be to allow only some layers meet the infill. The layers that don't can be different from the others that make up the shell. If in your example, the 2nd and 4th layers met the infill, the 1st and 3rd could be slightly different but not by much.

You could have just one layer meet the infill allowing the others more deviation from the shape of the infill.

In both of these examples, you would get more resolution in the shell at the cost of the strength of the bond between the shell and infill.

 

[Dim3nsioneer 557]

Daid has done some tests on that topic. Although the setup was not ideal (not exactly his words...) it seems that it worked. However, I think he postponed the implementation in Cura...

 

[pm_dude 27]

anon4321 I though of that and what I had in mind is basically extend the shell tickness to meet the smallest aligned infill.

lets say you have this slice

the bottom part is thinner than the to part

 

so shell on top would be extended to have a common infill border to attach to.

The shell thickness would be considered as a minimum shell thickness

 

[anon4321 16]

Ahhh, I see. So the infill shape is set by say the lowest shell layer in your attached example but continues upward at 90 degrees.

However, to meet up with the infill, the layers higher up are thicker or could "bend" in to close the gap...

Interesting....

Quick, get a patent!

Daid from this forum is the best to speak to this.

Not sure how much time you would save or how much increased resolution you would get. It seems like the slicer is unexpected complex to handle all the variations and it might be too hard to implement when compared to the benefit.

 

[Daid 306]

You mean... like this?

https://github.com/Ultimaker/CuraEngine/commit/21494ffc6c334385d1ee8f5cd37d3418b44d3459

 

[pm_dude 27]

Quick, get a patent!

HeHe! if only

Lets say you go for maximum quality. On UM2 that 0.02m layer height. If you object is 200mm height your going to spend a lot of time on doing those infill while they bring nothing more to the print quality in the end.

If you can infill at 0.2 and shell at 0.02 then that would be 10 X more precise surely not at 10X the time for the entire print.

Side note:

The same could apply with dual extrusion you could have a nozzle at 0.3mm and one at 0.8mm and do the infill at 0.8mm. That would allow for much faster infill speed and very precise shell.

 

[pm_dude 27]

You mean... like this?

https://github.com/Ultimaker/CuraEngine/commit/21494ffc6c334385d1ee8f5cd37d3418b44d3459

 

Its my first look at the code but yes that looks alot like it.

It would need some adjustment for extending the shell to have a common infill border. Otherwise you could create a too small or too big infill. unless I didnt see it in there.

 

[IRobertI 520]

This was one of the features I really liked about Netfabb back in the day. It would print the exterior walls at half of the infill height. It worked well and often produced a nicer surface. I think KISSlicer does it as well in the Pro version but I could be wrong on that. Skeinforge did a lazy variant of this where it simply stacked two half height layers on top of each other but it didn't follow the counter of the object so it was kind of pointless.

 

[pm_dude 27]

Cool. I'm happy to see this as been done in the past. Now with 20 microns precision it would be even better

 

[eldrick 65]

Kisslicer and Slic3r do this by printing the infill every N layers as an option.

 

[3dcase 37]

Look for a copy of cura 13.04 ?

It had the option to print the outlines with halve the layerheight (on the same position).

But it helped for overhangs and looks, ( as I recall it :cool: )

 

[pm_dude 27]

Really? why is this out in Cura 14?

 

[3dcase 37]

Really? why is this out in Cura 14?

 

first of all, it was not adaptive as mentioned in the topic title.

it just halves the outline layer height and prints the second halve exactly on top.

Why its out? No idea, my best guess is Speed and simplicity. A balancing act between Speed and Quality

I still like it (and its options, different infill's, sequence in infill and outlines etc..) for small and/or tricky prints.

 

[Daid 306]

Really? why is this out in Cura 14?

 

13.04 was the last version that used the old and slow Skeinforge engine. This was a feature in this engine, which never really worked properly (as said, it just used the same outline twice, so it really only worked for mechanical stuff)

With the introduction of my new CuraEngine (codename Steamengine) a lot of features went missing. Some are making it back right now. But with the new engine we got a 100x (on average) improvement in preparation speed, and much greater control on anything. As the skeinforge engine code was a huge mess.

Note that I'm only combining multiple sparse infill layers into a single layer. This had little to no effect on the outside skin quality in my tests, while combining the top/bottom infill has a lot of effect on the overall quality, and actually didn't save a whole lot of time.

Now, this sparse infill layerer combining feature is only prepared for the next code-name release. Which is called PinkUnicorn. Project PinkUnicorn is a whole UI overhaul. The UI code from Cura 14.07 is still based on the UI code of Cura 12.06, it has grown into a mess, and cannot support the advanced and complex features I want to build.

I have a lot of plans for the PinkUnicorn, but they won't make it all in the first release. (Stuff like select-able support material, one object with sections of that use different settings)

 

[codemaven 0]

Nice to hear that Cura is continuing to progress. The feature mentioned in this thread is one of the few big Netfabb features that I miss which are missing in Cura... In fact, now that you have seperate speeds for outer shell vs inner shells I think the different layer heights for infill vs shells is the only significant feature that I find missing in Cura compared to NetFabb... Although the ability to do contour type top/bottom skins would be nice too.

Cheers,

Troy.

 

[Daid 306]

Although the ability to do contour type top/bottom skins would be nice too.

 

https://github.com/Ultimaker/CuraEngine/commit/e1a11d07414ce0d5687dc014e1ee9c8328f3f97a#diff-7888f314216c9766280f546d5058697fR625

 

[codemaven 0]

Cool! I assume that is planned for release with PinkUnicorn too?

 

[pm_dude 27]

@Daid Thanks for the detailed explanations.

I see the future of Cura is bright!