Cura vs. slic3r

[alaris2 1]

thought I'd see how far the competition had gotten so tried slic3r again today.

some interesting observations I thought I'd feed back for Daid -

1) Cura is definitely much easier to use, and the visual feedback (model window) makes a world of difference.

2) slic3r does have some options which I would describe as advanced, but which are extremely useful and it would be nice to see them in Cura (list in a moment)

3) the print quality, trying several different out of the box variants and trying to copy my Cura settings as closely as possible, were all inferior to that produced by Cura

4) however the time it takes to slice is impressive. 33mins for a test model under Cura, 3mins 30.28seconds for the same under slic3r.

5) default start/end gcode for slic3r is pants compared with Cura. first thing it did was ram the nozzle into the bed with cold plastic still on the hotend which made a nice snapping noise. then heat the end up whilst still buried in the bed, peeing pla all over the place and going all over the fan shroud..

6) stringing under slic3r is as bad as repg. print a hollow cylinder and it tries to take short cuts through the center leaving a nice mess to clean up afterward. Cura just printed a perfect cylinder for me with no clean up required by moving along the perimeter instead of taking short cuts.

list of options that would be really useful in Cura: (in no particular order)

1) skirt in Cura joins with the beginning of the print proper. whereas there's a retraction between end of skirt and start of model in slic3r. actually this is really important since sometimes the skirt doesn't completely work and loose bits work their way onto the print head (or more usually get caught by the fan shroud) and make a mess.

I therefore usually remove the skirt just after printing (ie. while printing) but I can't do that if it's attached to the model..

this should be easy to implement but would make a big difference.

2) retraction has a 'lift z' option. i can see the logical point for this, but haven't decided if it's really useful or not. mention it more out of interest than anything. under the same category is 'infill every N layers'. i'm not sure what the implications for this are (collapsing infill? structural problems?) but it can be used to make the print much faster of course.

3) can print without walls. this is quite useful for artistic prints as i mentioned in another post. (I was making a lampshade)

4) i didn't try all the fill patterns, but 'concentric' looked interesting. it might be the same as the 'circle' option already in Cura but I think it starts at the outer edge and spirals inward when filling which looks good for top surfaces. it also has a infill and top surface fill as separate options. using lines for infill and concentric for top surface would be my choosing probably.

5) first layer can have a different temp from the others. you mentioned this yourself Daid recently.

6) print speed differentiates between : perimeters, small perimeters, infill, solid infill and bridges. that's really useful when your model is anything more complex than a cube and you've gotten past the noob stage.

what do others think? so far those 6 items set slic3r apart, but not far enough ahead that i'd choose it over Cura i'm afraid. that's good news for the Cura folk!

[joergen 2]

mention it more out of interest than anything. under the same category is 'infill every N layers'. i'm not sure what the implications for this are (collapsing infill? structural problems?) but it can be used to make the print much faster of course.

This is a very fundamental difference between cura and slic3r (and NF for that matter): Skin (in cura) is just printing the outer loop twice with half the layer height... nothing gained in terms of resolutions, steps are exactly the same steps. "skin" in NF (half layer) is increasing the slicing resolution, and the number of layers double up.... "infill every N layers" is supposedly the smartest solution, since you get real resolution, while things that can be printed thicker (infill) just does

[alaris2 1]

I keep meaning to try the 'skin' option in Cura and not getting around to it. so you're saying that if I print a 0.2mm layer object with skin, the outer surface will look as if it had been printed at 0.1mm with only a slight degradation in printing time?

I think I misunderstood the difference between that and infill every N layers then - are they not equivalent?

Cura - print at 0.2mm/layer with skin = slic3r - print at 0.1mm/layer infill every 2 layers?

[joergen 2]

I keep meaning to try the 'skin' option in Cura and not getting around to it. so you're saying that if I print a 0.2mm layer object with skin, the outer surface will look as if it had been printed at 0.1mm with only a slight degradation in printing time?

I think I misunderstood the difference between that and infill every N layers then - are they not equivalent?

Cura - print at 0.2mm/layer with skin = slic3r - print at 0.1mm/layer infill every 2 layers?

As I have said, cura does slice the object at i.e. 0.2mm, and then doubles the outer 'skin" as 2x0.1mm, changing the surface appearance... if you take a 45deg slant, it will still have 0.2mm 'steps', although each step has 2 fine perimeters.

slic3r and netfabb on the other hand, would slice the object at 0.1mm, and then make a infill with 0.2 or 0.4mm (slic3r: infill every 2 or 4 layers, netfabb: every layer). this increases the resolution, since a 45deg slant would have real 0.1mm steps. the netfabb solution is cumbersome because of the gross miscalculations how many layers the object has (i.e. if you want half-layer skin, you need to double the up/down skin layers to get the same result you were expecting before).

[Daid 306]

It's a silly compare. Really it is. Because I've always said "when Slic3r has the stability and features that I need compared to SF, then I will use Slic3r as back-end instead of SF." So far every time I've tried Slic3r I've found it producing some kind of wrong GCode in cases where SF worked fine. And I personally rather spend 10 minutes slicing a project which takes 8 hours to print and know it's sliced right, then slicing for 10 seconds and needing to inspect the full result.

Cura is trying to be a complete solution, Slic3r is just a slicer. Cura-RC2 actually has the option to use Slic3r as backend, but I disabled this in the current development version, because it lacked to many features to get the project planner and dual extrusion working. I also never got Slic3r to build from source.

(If you want to use this in RC2, you need to copy slic3r into the Cura installation, on the same level as Python/PyPy/Cura, and then select Slic3r from the cura-preferences. But note that not all options will work then, but it will allow you to compare the Slic3r results to Cura. This was only tested with Slic3r-0.7.1)

The main difference between speed in Cura-SF and Slic3r is because of the way infill is done. Skeinforge get slower if the surface area of the fills get bigger, Slic3r gets slower when the outer contour gets more complex. This means a 100x100x100 box will slice shitloads faster in Slic3r, but for example the Owl statue that I tested was only 33% faster in Slic3r.

[destroyer2012 0]

It's a silly compare. Really it is.

I disagree. Both slic3r and cura try to simplify the process of turning stls into gcodes, so they are comparable from a user standpoint. You may have lofty ideals as a developer but I don't care when I'm just trying to slice an stl, slic3r and cura seem the same in my eyes.

 

So far every time I've tried Slic3r I've found it producing some kind of wrong GCode in cases where SF worked fine.

well that's funny because I've had the opposite experience a lot of the time. That's why we are having this discussion. You say slic3r lacks features, well I think cura lacks a lot of features too in terms of parameters we aren't given access to.

[Daid 306]

It's a silly compare. Really it is.

I disagree. Both slic3r and cura try to simplify the process of turning stls into gcodes, so they are comparable from a user standpoint.

Then you simply don't see the added value of Cura. Cura is not just a Slicer. It's a 3D printing software solution. They are comparable from your standpoint because you only look at "turning STL files into GCode". While the goal of Cura is, "how to improve the user experience of 3D printing".

Slic3r is trying to be the best possible GCode slicing software. And I say, GOOD! Go for it. I would even say, drop the GUI. Just concentrate on the slicing part.

Cura is trying to improve 3D printer user experience. Especially for none-techy users. Good working installers, easy to use GUI, feedback to users. Extra tools like the project planner.

 

well that's funny because I've had the opposite experience a lot of the time. That's why we are having this discussion. You say slic3r lacks features, well I think cura lacks a lot of features too in terms of parameters we aren't given access to.

I'm not targeting the power users. But I am wondering which parameters you mis. There should be about 220 Skeinforge options that indeed are unaccessible.