How "smart" is netfabb?

[mm_build 1]

Don't know how else to ask this. When using Skeinforge, the biggest bottleneck seems to be the unnecessary jumps it makes while creating infill. It will start to fill in a layer, then jump randomly to the middle, do a few lines, then jump ahead and do a few lines, then jump back... basically it's jumping all over the place all the time when creating infill and it creates a lot of opportunity for inconsistent adhesion... especially at lower layer heights.

So my question is, is it worth it to upgrade to netfabb? this is my number one issue and greatest frustration with skeinforge at the moment, and if I shelled out dollars to netfabb only to find the same thing, I would not be very happy, because for the most part I have tweaked skeinforge enough that I am otherwise happy with it. I know there are settings I can change in skeinforge that will change the amount of jumps and where they occur, but it seems to be totally random. If I was paying skeinforge to mow my lawn, it would be constantly turning off the mower and dragging it to another area for no reason and restarting it... and I would have to fire it. :(

[owen 19]

The fills are very uniform with Netfabb. Generally only jumps where necessary to fill another area. I don't think you would be disappointed with Netfabb.

Owen

[Daid 306]

Have you tried changing the "Start from choice" setting in SF? It might help (not sure, might just be the starting point for the fill)

[somerwil 0]

Hi,

I'm very new to 3D printing and the Ultimaker but I bought Netfabb last friday. The results Netfabb calculates is very quick and detailed. The quickness of calculation was the main issue for me to give it a try.

About the downsides on Netfabb I found this far (after ONLY two days of playing around)

1) I also noticed a few strange jumps throughout my print of the OK-hand (

http://www.thingiverse.com/thing:10017

).

It's pretty obvious the fingers can't be printed without jumps but I found some string in the palm on the hand as well. That's not making any sense to me...

2) The same thing came up with a church (

http://www.thingiverse.com/thing:11727

). On both sides of this model three short pillars are drawn. Netfabb calculates these correctly on one side of the church but comes up with a lot of stringing on the other side of the church.

Also I noticed Netfabb doesn't get the strings inside the object where this could have been possible. It goed stringing on the outside of the object making the strings visible. I hoped it was clever enough to just make string inside the object where no one can see them.

3) When the Spiro-coasters (

http://www.thingiverse.com/thing:8928

) are calculated through Netfabb the outcome is terrible. It comes up with an awfull lot of unnecessary strings while SF35 did that one pretty well.

4) Quick build profiles are not available yet. I'm not able to get some really quick print code from Netfabb because it simply makes the Gcode too precise which is not a good thing for every sort of object. I really like Joris' vases but quick style for these objects is more than good enough.

Perhaps these issues can be solved by using the software in a different way or configuring it yourself in the good interface. With the standard, beginners knowledge I have it has some points of attention. I'll try to post some pictures of the problems I had later today. For now SF finally got my huge tornado calculated (approx. 1 hour) ;-) Ready to print now!

On the good side are calculation time, graphical UI (really nice) and user-friendly build and material styles. This compensates more than enough for me to make my investment worth every single Euro I spent on it. I love the Netfabb software and I think with the direct 3D print option to come and some more trial and error I'll never go back.

Michael

[cyclone 0]

In my experience Netfabb is only slightly smarter, if at all, about jumps vs Skeinforge.

Seems like something with access to more processing power like NF would you some advanced

shortest path algorithms to cut down on stupid fills. But it doesn't. It even resorts to jumps

when doing hollow objects that don't need a jump at all.

Don't get me wrong I like NF, there are just a new set of stupid things it does. But it does things

a whole lot faster than SF. (A lot faster) so there's the trade off I suppose.

-b

[somerwil 0]

OK, here are my experiences with stringing in Netfabb

Netfabb's output of the Spiro-coaster:

https://lh6.googleusercontent.com/-Zj2x ... oaster.jpg

Netfabb's output of the vanderBilt building:

https://lh5.googleusercontent.com/-XYXz ... erBilt.jpg

The Spiro shows an afwull lot of unnecessary strings where the vanderBilt is calculated with a rather unclever path for the strings. In the vanderBilt calculation the strings are very long while shorter paths are available in mostly all cases.

I really hope this can be optimized for future releases.

[Daid 306]

Note that calculating the most optimal path to print a layer is a Travelling salesman problem Which is hard to solve.

Seeing those pictures I come to the conclusion that Netfabb does not use something like the SF "comb" module. Which should prevent all strings in the spiro-coaster.

[somerwil 0]

I did some more testing with Netfabb today and yesterday and came to the conclusion that Netfabb makes some extra strings when the STL file is scaled through Netfabb.

Tried to print a Random Vase V2 by Chopmeister last week. I downloaded the STL from Thingiverse, opened it in Netfabb, decided it was a little too big for me and scaled it down to 50% of the original STL with the Netfabb rescaling tool. When the toolpath calculation was complete, some red strings were there. Today I noticed that when I just throw in the 100% STL and do the toolpath calculation on this 100% model it does not show these extra strings.

I can't think of a reason why rescaling should result in different outputs for the toolpath and I need to do some more testing.

[ddurant 0]

Based on the pictures you posted above, I suspected it was related to the thickness of walls between where the head is and where it wants the head to be..

An interesting test might be to save a 50% STL then reopen it.. Does it still produce the same number of strings? If you then scale it 200% (or is it 100% ? whatever gets it back to the original size), does it still have the strings or is it back to the way the original STL was?

[somerwil 0]

Did some testing with the stringing this morning. I came to the conclusion that rescaling is not the problem.

I tried:

1) OK-hand in both Lowpoly and original STL and all of them gave strange toolpath results, particularly in the palm of the hand. Rescaling did have some effect on the output of the toolpath but not what I saw with a different model last week...

https://lh6.googleusercontent.com/-g0jO ... K-hand.jpg

https://lh6.googleusercontent.com/-MbM2 ... owpoly.jpg

2) RandomVase V2 model (solid model) by chopmeister in different Netfabb Print Quality settings. This gave good toolpaths without stringing. The non-solid model (3mm walls) gave some extra strings in the toolpath.

https://lh3.googleusercontent.com/-zt2s ... 2520V2.jpg

3) vanderBilt in different scales and profiles. The toolpath outcome was basically the same with only minor differences as shown on the OK-hand tests.

4) SpinnyjiggyVase on different sizes and print qualities. Now problems really

https://lh3.googleusercontent.com/-TUxa ... gyvase.jpg

New conclusion: Rescaling only affects the output of the toolpath on a detail level and is not responsible for some extra stringing. The STL however is the source of unnecessary strings as can be seen on the RandomVase outputs.

I still don't know why the OK-hand is not calculated properly. I really think the strings in the palm of hand up to where the thumbs splits up from the rest of the model are unnecessary.

Even with the stringing issues I still love Netfabb! The fact I can make these print screens within an hour and a half (including opening different files, do the toolpath calculation, copy-pasting, editing and placing them on Picasa) makes me love it. The presentation of the toolpath, the slices and the fill in full-time, full-speed 3D is worth buying this awesome software!

PS: I tried to edit my previous post because my conclusion not good. I couldn't find an edit button though...

[chopmeister 0]

Hi!

I'm glad you tried to print my vase

As for the strings in the thick wall object, that, in my experience, is due to fill layers. Whenever I disabled infill on complex models, the jumps would go away completely. I think netfabb has some problems with calculating fill layers because

sometimes I too saw jumps where there was no real reason to have them.