PDA

View Full Version : stl export problems



Tim21
08-26-2015, 08:02 AM
I'm experiencing some problems with exporting an object into stl to try out a 3D print.

in a render everything looks ok:

129492

but when I export it into stl problems arises:

129493

I'm not experiencing this problem with the obj file.

anyone know what to do with this issue?
thanks a lot in advance! :thumbsup:

lertola2
08-26-2015, 08:39 AM
I would try tripling the model, merging points, unifying polygons and deleting any one and two point polygons and aligning polygons. Then use the select boarder edges command to see if there are any holes in the object.

Tim21
08-26-2015, 08:52 AM
thanks, I'll try it.

I also kept searched for a possible solution and found meshlab (http://meshlab.sourceforge.net/) to be usefull since the obj file was ok.

with this I opened the obj file and converted it to stl and everything seems ok now ...hopefully.

Tim21
09-03-2015, 12:09 AM
Hi,

I keep getting these stl convertion issues. One issue that is solved is the scaling though.
What I'm doing at the moment is delete all 1 and 2 point poly's (trough the statistic menu), merge points and do tripling. But some poly's keep causing problems + I can't seem to find the border edges command to search for holes yet.

And are maybe there other things to do before converting it to stl to get a good file?

any help is more then welcome.

spherical
09-03-2015, 01:29 AM
Well, it all comes down to good modeling practice in the first place. Many times it is overlapping polys that cause these issues. IOW, two or more polys are nearly in the same space OR not in the same space by a big margin. Automatic tools like Unify Polys will not identify these. The tool to identify their existence would be Edges on One or More Than One poly. Edges should ALWAYS be on two polys. No more, no less. If that doesn't immediately show anomalies, it's still a good place to start.

Tim21
09-03-2015, 02:42 AM
Thanks spherical, I'll look into it.

ps my modeling skills needs plenty of work to say the least.

spherical
09-03-2015, 03:36 AM
Well, we all do start somewhere... :) It's taken some time for me to recognize the actuality of where different problems lie as far as creating a manifold mesh that is sound and robust for 3D printing or milling. You'll get there.

The various manufacturers of machines would have everyone believe that it is a simple point and click operation but it has a significantly long road to go before it ever gets to that status. I pity the librarians who are suddenly tasked with nursemaiding a 3D printer that the committee thought would be a good idea.

Tim21
09-03-2015, 04:36 AM
lol you got that right! :lol:

my saying in this is that is not like using a senseo coffee machine just yet. :)

I'm exploring the field for a bit and found out that I need to rethink how I 3D model/design almost completely. For a nice render everything is quite ok, but for a 3D print ...well that's another thing.


ps despite some serious issues I did get a 3D print out of this model, with a lot of help from the firm that printed it. Mucho kudos to that firm for their help and patience!

129584

spherical
09-03-2015, 03:40 PM
If you can make the original model available I can take a look and report on what I find and then you'll have an idea on what to not do :D and/or how to fix it. The model isn't complex, at least in comparison to what we do here, so whatever's Borked should show up easily and we can all learn from it.

Tim21
09-04-2015, 12:03 AM
Thanks spherical! much appreciated.

Here is the latest version (facet.lwo).

the edge statistics show some elements with 1 polygon, I think that's where the problem is. But I have no idea on how to repair it, or maybe even better, to avoid it in the first place.

Even with a simplier test (1.lwo) I still get these issues. (the model in layer 2)

my aim at the moment is to have a clean stl export out really for this and future models. I don't want to further test out the patience of this firm really with sending them messed up stl files :)

spherical
09-04-2015, 01:16 AM
OK. Thanks. BUT, I don't want the "latest version". I want the original that gave you the problem in the first place. That is where to start learning. I'll look at these, but the one that caused this thread is perhaps of more use; not something that has been fixed.

OK, looked at both. First things to correct:

Anything that is going to slice a model will not understand LightWave layers. Go ahead and build your model as multi-layer if you like, but get it all into one layer and merged into one mesh in another file that is then converted to STL. The Facet model is on layer 3. Delete layers 1 and 2 then convert.

Use the Statistics panel to identify cruft in the model:
Edges can ONLY be on two polys, no more, no less; UNLESS it is an edge that leads from a perimeter to a hole.

In the Facet model there are points on only two polys. While some instances of this are legal, most times they are not; and some of the ones that may be legal could be eliminated if they do not provide a function. Select the points on two polys and locate where they are. Deselect and then select one location that was identified. You'll find that there are two or more points there that need to be welded. This will likely take care of some of the Edges issue.

More later. Dinner calls....

Tim21
09-04-2015, 02:04 AM
Hi there.
I think it was this one (spoon 24 teeth) but I think it doesn't really matter at this point. Every stl convertion I try to make at the moment is getting an error when analizing it. I must be doing something completely wrong.

spherical
09-04-2015, 03:06 AM
Correct the above and you'll be a lot closer to right. Bear in mind that the paramount objective is to have a model that is manifold. Manifold means "water tight". That means no holes (openings through which you could see inside at the back of the polys) in the mesh, no disconnected points. If any of those exist, there is no "inside"/"outside" relationship. It's all "outside". The slicer needs to understand what is the outside, so it can correctly do the myriad cross sections that can then be stacked upon each other to reassemble the shape.

OK, 24-teeth has:
68 edges on only one poly
2 edges on more than two polys
1 point on only one poly
46 points on only two polys

Of the points on only two polys that I sampled, all of them have two points in the same location. Select each set of them individually and Weld. BUT, the first thing to do to clear the low hanging fruit is to hit "M" and choose Automatic, turning off Keep 1-point polygons. That immediately brought the 46 down to 8. Then, you can either Merge again using Fixed with a very small distance to identify points that are close but not completely coincident; increasing the distance to a value where the mesh changes undesirably and then Undo. After that, it's all manual finding and welding.

Until those are corrected, there's no point in analyzing the rest of the readouts as to points on 3, 4 or greater than 4 polys, as those will change when the above are corrected.

Good modeling practice will prevent the huge majority of this. Learn from what you find in these example models and things will go more as expected in future projects.

Tim21
09-04-2015, 04:06 AM
Many thanks for the feedback :)

I'll try the correction, and what you said about good modeling practice ...I think there lies the real problem that needs plenty of work.

I had the "easy" preassumption that the computer would outline the outside sorta speak in an stl, just like a visual render is doing in a way. I guess I was wrong.

spherical
09-04-2015, 02:30 PM
Heh. No, it is a conversion from one format to another. All 3D files store the mesh information in some way. It's code that is understood then translated and written to the destination format. They're just like languages.

A bit of clarification on the "holes" that cannot exist in a mesh to be sliced: A hole in a mesh that is fully closed, as in the letter O, where the hole in the center of the letter has sides that join the front and back, is not what I'm talking about. If properly constructed, that is a manifold shape that has a surface all around.

A "hole" in slicer terms would be a poly that is missing. You can then see inside. If you can see inside, then the outside "wraps" around through the hole to the inside and you have only one surface that is inside itself, so to speak. But, it doesn't take a missing poly to create this. An unwelded point will do the same thing. It isn't water-tight at that vertex. There's a leak that will confuse the slicer.

A slicer first looks at the entire shape to analyze the construction and determine if it is viable. Actually, some slicers don't make this determination and will happily slice any ol' buggy mesh you throw at them and do a crappy job of it. Good ones do perform this step and users of the dumb ones complain about how the one that is actually doing more work to obtain a better result are "broken" and a pain to use. But, I digress.... Once the initial analysis is complete and the mesh is found to be viable, the slicer then only looks at the shape in thin cross sections and plots a path for the head to follow that describes the shape at that elevation from the bed. It is only concerned with that one elevation. Then it moves on to the next highest again and again until it reaches the top. It is the succession of these discreet paths stacked upon each other that reproduces the virtual shape that we can see as being the desired object. But, if we looked only at one layer at at time, we would not be able to determine what the overall shape would be. It is only a meandering line that returns to where it started.

The point, not to make a pun, of all this is that if only one of those extremely thin paths encounters what would amount to a chasm as it goes around the part trying to get back to the beginning, it falls in and is lost. If your traveler is to be kept safe, you must provide a good road upon which to walk.

For the most part, a renderer doesn't care about any of this. It just fires rays that bounce off of surfaces until they expire in one way or another; either my being recorded as a pixel color or extinguished by Ray Limit. When you are dealing with 3D printing or CNC milling, you are bringing a virtual object into reality. It cannot just "look good", it has to BE good or it will fail.

Tim21
09-04-2015, 04:09 PM
For the most part, a renderer doesn't care about any of this. It just fires rays that bounce off of surfaces until they expire in one way or another; either my being recorded as a pixel color or extinguished by Ray Limit. When you are dealing with 3D printing or CNC milling, you are bringing a virtual object into reality. It cannot just "look good", it has to BE good or it will fail.

That's what I was afraid of in a way. :lol:
but it confirmed my thought that I still got more then a few things to learn modeling wise for 3D printing.

spherical
09-04-2015, 05:03 PM
The Statistics panel, combined with Merge and Weld, and sometimes deleting/recreating polys will go a long way to cleaning things up that... er... don't go as planned. :D

You want to eliminate:

Points on no polys
Points on one poly
Some points on only two polys (there are legal ones, depending)
Edges on only one poly (unless going to a modeled hole)
Edges on more than two polys
1-point polys
2-point polys


As you become better at modeling, these will be found less and less.

Tim21
09-04-2015, 05:48 PM
many thanks, I'm expecting many of such moments :D

I think I got most things covered now, except for one thing: the edges.
Could you by any chance explain for a bit on how to deal with them?

spherical
09-05-2015, 03:39 AM
Well, in rare cases they can be deleted. Most times, however, you have to delete the poly(s) that they are on and manually recreate the polys by selecting points and hit "P". Many times, deleting 2-point polys will take care of the Edges that are causing problems. It all depends upon the situation.

Sometimes, Edges on only one poly will be simply a poly that isn't there. IOW, there is no adjacent poly for them to attach to; so all of the Edges around that perimeter (hole) are only on one poly. Use Sketch view to see if polys are missing. If any "appear" to be missing, select in the apparent hole to see if there is actually a poly there that simply needs flipping.

Tim21
09-05-2015, 03:59 AM
I see, delete poly's and recreate them...I'll check it out.

Meanwhile I try some easy modeling from scratch in an attempt to have a better insight to these things (and hopefully evolve into cleaner modeling habits). See attachment for my first abstract print without stl issues :-)

Many thanks for the help!:thumbsup:

spherical
09-05-2015, 05:39 AM
Alright! Nice forms. What are their dimensions?

Tim21
09-05-2015, 07:06 AM
the tiny one is around 3cm in height, the biggest one has a 13 cm height.

It's just a very basic form that I subpatched and altered a bit. This to test out subpatching in 3D printing in several settings.

btw had a look at your website, pretty fascinating things you make!

spherical
09-05-2015, 04:06 PM
Thanks! We're working on a new award right now for the National Aviation Hall of Fame. It's the first Neil A. Armstrong Award for Outstanding Achievement. The recipient is Jim Lovell, Commander Apollo 13. Each award will be unique to the person's achievement and we have a really great one in development that speaks to his. I'll post a render once the design is finalized and the embargo is lifted. Everything is fully modeled in LightWave before we go into production. Often, we create a flyaround movie, so the client can see all sides.

This is the second award that we have created for Jim. The first was the Space Foundation General James E. Hill Lifetime Space Achievement Award. For his individual sculpture, we made a blown glass moon as the sphere that is attached to the spiral.
129602 129603

We made 20 years of these, all different but similar, not knowing at the time (with the exception of Jim) who the recipients would be. So, we created planets, the Sun and a star ball; silver leaf specks on a black glass sphere. We chose the Mars for Dr. Charles Elachi, Director JPL.
129604

As one would figure, five of the fifteen awarded so far are Astronauts. It is a great honor for us to have the opportunity to create these and become part of their legacy.

Tim21
09-05-2015, 04:36 PM
nice!

so everything is modeled in lightwave and then transformed into a glass object?
pretty impressive.

spherical
09-06-2015, 07:21 AM
Yes. Working in glass is demanding. It requires intense concentration to arrive at the intended goal. We need to know before going in not only how it will look but the size and shape of everything, so that it all fits together. Working up a high fidelity 3D model allows an intimate experience with the design from start to finish. The huge majority of kinks are worked out and different methods to achieve a particular goal are discussed and hammered out before we get into production. Many times we develop an entirely new technique when we run into an issue during the modeling phase. Saves a lot of trial and error in the Hot Shop. Same goes for the carbon fiber or wood constructions and mechanisms like the programmable orrery. Once the design is the way we want it to look and everything fits, the model is then the blueprint from which we extract the metric data to build with. I don't like surprises; especially on a big commission.

Tim21
09-06-2015, 09:02 AM
I can imagine. Glass isn't the most easy material to work with I think. But the results are really impressive!

So lightwave is quite essential in your workflow it seems?
Can I pm you with some more questions about this?

spherical
09-06-2015, 06:02 PM
Very essential. I'm an engineer first. Artist second. May have been the other way around if my high school guidance counselor hadn't thrown me off the path, saying: "You can't make money in art--choose something else." So, I chose engineering drawing and Descriptive Geometry. They were the closest to art that I could get and be in an "acceptable" field. I found that I had the ability to "understand" the machine from its point of view. I could feel the forces as they presented themselves.

Of course, if I had followed my original choice, I may not have become an engineer; which led me to Penske Racing and winning the Indianapolis 500 and Can-Am Championship. SO, art projects are approached from a scientific point of view. It allows far greater things to be realized; especially when doing things in Real 3D. I can fake a lot with a brush or airbrush. Working in 3D applications is the same, only more so. Nothing has to follow the laws of physics, unless you choose to apply gravity and wind to scenes. In the Real World, things have to actually work and not fall down. There is no fakery. Understanding the physicality of an assembly or especially a working apparatus is paramount. Using 3D models is the best path to that understanding. It allows you to be intimate with the design and you can also take the design further because of Save As and Undo. If you are approaching something that no one has ever done before without having a way to go through the process beforehand, then the decisions are much more difficult. More than once I've ruined an acrylic painting because I went too far. That is, however, one of the things that makes an original valuable. There is only one.

To give a bit of perspective on how important we consider this to be, that 7' tall, 345 lb. (2.13m, 156.5kg) master award sculpture seen above, which has to be movable between locations--so is also disassembleable, was our first glass sculpture. It was fully modeled in 3D first. That gave us the confidence to proceed into uncharted waters. Obviously, we learned a lot along the way; developing concepts and methods we still use today. At the heart of it all is good design using good tools.

Tim21
09-07-2015, 09:32 PM
same here in a way. I'm raised as a (interior) designer. I thought it was the best option to take. Creative but still with some sense of functionality, the best of both worlds :)


Meanwhile I'm focussing on just playing around with box shaping instead of detail modeling. This to reduce mistakes and to give me more insight into cleaner modeling. And it is working so far thanks to your advise. fewer mistakes so far, and if I make a mistake I got more insight now to what is wrong and how to fix it.

You can say that the artist in me is having a great time at the moment :).
The designer in me is already focussing on reducing material and improving strenght of the model at the same time, in a more abstract way.
Hence the progress already from pure solid forms towards forms with holes.

Here are some examples of the progress I made so far:

129638

129639

spherical
09-08-2015, 05:07 PM
Those are some nice looking, intriguing shapes.

Interesting parallel on the job front. When getting out of engineering for a time as my primary focus, I did ArchViz illustrations in traditional media for a well known designer. Prismacolor pencil and brush/airbrush watercolor and India ink, mostly. Rarely was the space built, so I got the architect's plans to be able to understand the building and all of the swatches used in the interior design. The first ones were fairly remedial, but the latter became more demanding as the designer learned what I could do. I was a bit unsatisfied with the accuracy and wanted to do illustrations that, as much as possible, reproduced what a camera would record from a given location. What was and was not visible and how large. That's when the engineer came back.

I knew there had to be some way of determining a perspective projection of a 3D world and transforming it into a 2D image. I developed my own system of plotting points where they intersected a cutting plane; using the field of view of a chosen 35mm camera lens. A blizzard of points were recorded in pairs on two planes. I could then scale the plane point locations up to match the aspect ratio of the illustration and plot where the pairs intersected. Then it was a session of connect the dots to reveal a correct perspective image, as seen from a chosen location in the room that didn't exist. This all had to be done in one sitting; keeping the relationships in my head. If I came back to it later, I'd be totally lost.

All went well for a number of illustrations until I was tasked to do a modular office landscape in close quarters. That required a very wide angle lens to get it all in. I set to work and got to the pairing up stage and the system was falling apart. Verticals weren't lining up where they should and I thought that I had missed something of my own system. Knowing that it had to work but not knowing why it didn't, I went further into it; plotting intermediate points in order to prove what was going on. It turned out that the system was so good that it was actually reproducing the barrel distortion of a severe wide angle lens. IT WORKS!

One in particular was a nightmare. Board room with 16 Vecta Contract chairs. Each had to be drawn the same only smaller/larger with slightly changing perspective. I iterated through those chairs for more than a week. The ornate Oriental style carpet didn't help. I got into a discussion about that when she said that my carpet in the shadows was too dark. To prove it, she dropped the swatch in the shadows and placed the illustration right beside it for comparison. I was gobsmacked. Uhhh, the shadow in the illustration is built in or it wouldn't look like a shadow. She didn't get it. I finally walked over and pulled the illustration out into the light and the shadows matched. Zing....

It was like being paid to go to school. I gotta dig those 4x5s out, scan them and get them up on my website. Lots of work represented in them.

Now, of course, we use LightWave, set things up, push a button, walk away and come back when the fans slow down. :)

SO, back on topic, how are you generating these model shapes? With them being compound curves, some of which are tight, I'd think that they would be more ripe for errors than the faceted spork.

Tim21
09-08-2015, 10:26 PM
I would like to see those illustrations, sounds very interesting.
When I started with my studies in high school the computer was not mainstream, we still had to use pens and sheets of paper/blueprints (ooh man that was a nightmare compared to now :-)
But it had it's charm as well.


These models are just playing with some subpatched boxes, afterwards I make holes in them, connect the dots again to create poly's in the holes, tamper with some points, and thats about it for now.

It was also a surprise to see that the spork design gives alot more errors then these shapes. I deliberately chose that facet look in that cutlery concept design, partly because I thought it was more easy to get a 3D print out (it is my first dive within 3D printing).

Yesterday I made this (see attachment) and clearly noticed the strength of this thing and at the same time reducing material big time. It's pretty fascinating to personally experience Neri Oxman has a point.
(In case you have around 21 minutes of spare time: https://www.youtube.com/watch?v=txl4QR0GDnU)

In time I maybe hope to be able to shape these kind of things into a more functional form. But I'm not there yet, and I'm having fun at the moment just experimenting with these more abstract shapes anyway :) :boogiedow

spherical
09-08-2015, 11:53 PM
When I started with my studies in high school the computer was not mainstream, we still had to use pens and sheets of paper/blueprints (ooh man that was a nightmare compared to now :-)
But it had it's charm as well.

I still have two 2 meter hydraulic drafting boards, each with a channel drafting machine. Will never give them up. There's something that is "right" about them. I find that I think differently about a project on a basic level when working with them and don't use them often enough.

Tim21
09-09-2015, 12:26 AM
yeah, I found it hard to let go of those things too (I managed to survive graduation with nothing but a manual approach with pen and paper, even though the computer was heading towards better results I kept with it rather stubbornly).

Would you by any chance be interested in some experiment?
I see this rather spontanious abstract evolution towards more strenght and reduced material trough a more organic and computed (strangely so) approach in these models, but maybe an engineer can confirm these thoughts?
Maybe these kind of things are very difficult to calculate from an engineering point of view though, but pm/mail me in case you might be interested.

spherical
09-09-2015, 03:16 AM
Well, in considering the concept, my mind went immediately to what we did in determining the most efficient roll cage for our Trans-Am cars. Generate a representation of the passenger compartment that includes points to which welding can be done, apply forces upon various vectors and watch what buckles, place a tube along that vector, rinse, repeat, eliminating extraneous tubes until you get to a strong, cross-supported structure that is as minimal as it can be.

In the slicers that I have used, there are a number of sparse infills of differing configurations that work very well in maintaining strength and rigidity while reducing mass to a significantly low level. In some situations, a 20% sparse infill will be nearly as strong as a solid.

Tim21
09-09-2015, 03:32 AM
Interesting. Was it a more mechanical (rectangular) end result or more organic in nature?

spherical
09-09-2015, 02:47 PM
Well, it was one of the first applications of computer tech to design a rigid structure. Given the general volume of a '68/'69 Camaro, the outer tubes formed the top and four sides of a cubic shape; the bottom being the subframe of the tub to which everything was welded. Then there were four intersecting tubes of varying length at angular vectors within that outer structure. Most teams build roll cages as safety systems to keep the compartment from crushing if it went wheels up through the air. We took it a few steps further in making it be integral with the entire tub; extending it in key vectors past the norm and outside the compartment to the corners. The result was a far stiffer chassis that could then be a stable platform for the suspension to work against. If the body has compliance, you don't know where the forces are going or being absorbed. That makes it a nightmare to apply adjustments to correct for course conditions. That was working with what we called "taxi cabs". You have to work with the stock chassis and can only modify as far as the homologation allows, plus a few exceptions granted by the sanctioning body. Indy Car or F1 are another story entirely, as they do not in any way conform to a street vehicle shape. There, the possibilities open up.

Tim21
09-09-2015, 11:44 PM
ok, do you maybe have some pictures of it?
a 20% decrease of material, sounds pretty good!

If you managed to have a look at the presentation of Neri, she looks at the possibility to have a 50% decrease in material this way.

I'm still wondering on how to calculate these organic shapes in terms of strenght to comfirm or dismiss these initial thoughts. Meanwhile I'm trying out some finer pieces and strength wise it seems to be more then ok, despite the reduction in material (see attachment).

spherical
09-10-2015, 03:28 AM
No, a 20% sparse infill is 80% nothing. It's an internal honeycomb structure.

Tim21
09-10-2015, 04:26 AM
aah ok :)

spherical
09-10-2015, 04:29 PM
In one approach, the strength is maximized by the shape itself. Density is ignored. Culling down to a best shape is always beneficial.

Then, the density can be brought into the equation to further optimize the strength to weight ratio of that optimized shape. The density and strength in this scenario have a relationship that trades off; but not in a totally inverse manner. IOW, choosing the correct honeycomb pattern for a given external shape can result in an overall strength equal to the solid object, but with a savings of mass. Anything more is a waste.

Many times a completely rigid structure is decidedly what you do not want. Aircraft wings are a great example. If they didn't flex, they'd break off.

Tim21
09-10-2015, 09:49 PM
yeah, a bit like the infill of some of these 3D models to further improve it's strenght.

Meanwhile I'm starting to develop some finer pieces, and finally it's starting to show some flex as well, especially on the "weak" spots where the shape develops into more of a thin line without structure/shape.

I'm pretty amazed that the printer can still print it though. And it is great to see those weak spots in action in a 3D model. It clearly shows where a model needs some more shaping if needed.


Btw talking about cutlery design again, I saw this concept the other day: http://www.shapeways.com/blog/archives/1640-3d-printing-food-in-the-future,-pffft,-3d-printed-cutlery-is-now.html

pretty amazing concept

Tim21
09-11-2015, 03:33 AM
damn, tried to create a marriage between a sharp box shaped piece and an organic one, and once again I generated these errors.

This time I used boolean/union, converted it to stl, but it gave me errors once again. Now I remember it, I used boolean as well in that cutlery design. :bangwall:

spherical
09-11-2015, 10:07 AM
Native Boolean isn't all that robust. That which usually is the issue is that the cutter will try to divide a dimensionless entity. This will either crash it or leave a bunch of cruft. That the organic part is probably pretty dense, odds are that the cutter went through a point or along a line. Points have no dimension, so cannot be divided at all. Lines have only one dimension, so can only be divided across the line; not split in half lengthwise to be half as thick--because they have no thickness. 3rd Powers BooleanTool does a much better job. Still leaves some edges and overlapping polys to manually clean up, but far fewer than the native tool.

JoePoe
09-11-2015, 11:48 AM
damn, tried to create a marriage between a sharp box shaped piece and an organic one, and once again I generated these errors.

This time I used boolean/union, converted it to stl, but it gave me errors once again. Now I remember it, I used boolean as well in that cutlery design. :bangwall:

Did you Merge after Boolean?
(and after merge just check for 2pt polys.... I doubt you'll have any tho.)

spherical
09-11-2015, 11:55 AM
No doubt, he did. We've been through that on the first project.

Tim21
09-11-2015, 01:03 PM
Yeah, I think I did (tried so many things really).
I did get a shape out of the printer though, the organic shape that I managed to print vertically.

It was the intention to put this shape into a box as some kind of support tool during printing, and afterwards you can cut the box away from the piece.


I'll try again tomorrow with a more simple object.

many thnx for the feedback :thumbsup:

spherical
09-11-2015, 01:10 PM
If your printer isn't a dual extruder, so cannot use an alternate material for support generated by the slicer, you can instead pull small polys from the base mesh down to the "floor" as legs. Saves the Bool operation.

JoePoe
09-11-2015, 01:34 PM
Yeah, I think I did...

Do, or do not. There is no "think". :yoda:

(I so rarely get to use the Yoda emo :D)

But seriously. How about throwing up one of the troublesome .lwo's for a hands on inspection?

jeric_synergy
09-11-2015, 02:42 PM
But seriously. How about throwing up one of the troublesome .lwo's for a hands on inspection?
I've never understood the reluctance to provide samples. It's the best, fastest, most complete way to get useful feedback.

spherical
09-11-2015, 05:17 PM
I got two samples, three actually, one was slightly different in a second layer. He's doing just fine.

JoePoe
09-11-2015, 07:07 PM
Well... I'm a little embarrassed I missed the file posts :D.

WOW.... a lot going on with those models. Trial by fire ;).

Tim21
09-11-2015, 08:54 PM
hihi, trial by many many errors :D but I got to learn it one day.

spherical already explained me a lot, at the moment I'm trying to clean up my act and try to model with no or way less mistakes, but they are hard baked into my system it seems. 8~

single extruder here, many thanks for the advice for using poly's, I think I'll try that one to save some filament as well.

spherical
09-11-2015, 10:42 PM
Heh. Saving filament and time is good. You'll get there. We all start somewhere. I just happened to have jumped in earlier, spent a significant amount on a fairly large build volume dual extruder machine and took the hard knocks. I'm happy to help out where I may.

Tim21
09-12-2015, 02:00 AM
Heh. Saving filament and time is good. You'll get there. We all start somewhere. I just happened to have jumped in earlier, spent a significant amount on a fairly large build volume dual extruder machine and took the hard knocks. I'm happy to help out where I may.

thanks spherical. I really appreciate it. :) Nothing better then a 3D printer to put some modeling to the test I guess :D

JoePoe
09-12-2015, 10:29 AM
It would be interesting to know your construction process to help you understand why so many holes are popping up.
For starters, I can suggest, at least, to make sure your centerline is zeroed out on an axis before mirroring (press "v" for set value).

Other than that.... one more thing for the project at hand:
You can have "legal" geometry still give you problems.
Case in point: on your first model, clean manifold geometry is folded in and crisscrosses/intersects itself.
This will NOT show up in statistics panel. Be careful. Happy modeling :).

129702 129703 129704

Edit: Seems to be cleared up on 24 teeth version :thumbsup:

spherical
09-12-2015, 05:09 PM
You can have "legal" geometry still give you problems.
Case in point: on your first model, clean manifold geometry is folded in and crisscrosses/intersects itself.{/quote]

That's not "legal" or "clean". Might be manifold, in the basic definition, but overlapping geometry will choke any reputable slicer. Dumb ones won't even see it, though.

[QUOTE=JoePoe;1444145]This will NOT show up in statistics panel.

Sometimes it will. Totally depends upon the situation. Statistics won't identify an overlap, but many times there are indications. You have to know how to interpret what the statistics panels is telling you. Combine that with Sketch view at close examination and you can find and fix most of it. There'll be edges that submerge beneath the surface and the points aren't selectable, except in Wireframe. After that, it's up to a mesh analyzer to flag Borked geo that you've missed.

JoePoe
09-12-2015, 06:37 PM
That's not "legal" or "clean". Might be manifold, in the basic definition, but overlapping geometry will choke any reputable slicer. Dumb ones won't even see it, though.


:D Uh.... that's my point. Hence legal in "".


Sometimes it will. Totally depends upon the situation. Statistics won't identify an overlap, but many times there are indications. You have to know how to interpret what the statistics panels is telling you. Combine that with Sketch view at close examination and you can find and fix most of it. There'll be edges that submerge beneath the surface and the points aren't selectable, except in Wireframe. After that, it's up to a mesh analyzer to flag Borked geo that you've missed.

What I was pointing out? No. Never. At least not in LW.
Indications of an overlap (in otherwise clean geometry) in statistics panel?? Like what?
Again, that's what I sayin'....Can only be found by a walk around.
(and yes, I know how to interpret the panel ;D)

yes... of course, polys and points are fine too.... statistically :hey:.
129717

spherical
09-12-2015, 06:45 PM
It was the intention to put this shape into a box as some kind of support tool during printing, and afterwards you can cut the box away from the piece.

Forgot to comment on this. If I read correctly what you are attempting, another approach would be to add the box not by using Boolean; which you can see has its problems.


Create the box you want to add, put the box temporarily in a background layer and position it on axis.
Sight along the axis where the two are lined up or view from the side to see where they intersect.
Select points on the near surface of the object that are within the box and delete them. Nuke the 2-point polys left behind.
Select the points around the hole just created and hit "P to create a polygon.
Cut/paste the box into the layer with the object, with the two separated from each other by a bit.
Select the two mutually facing polygons, one on the box, the other the poly just created on the object.
Use the Bridge tool to join the two meshes together. A multi-polygon adapter between the two is created that will join any number of sides on one to any other number of sides on the other. No cruft to manually fix, because there is no cutting through geometry involved.

Tim21
09-12-2015, 09:22 PM
Hi there Joe,

phoe, that is indeed messy.

about the construction process:
First i created a box,
knifed it to have some more points
I created half a spoon, that I mirrored later on.
afterwards I used boolean to create the little knife elements and fork elements in it. And I think now that is where the real problems started.


thanks again for the tip spherical, I'll try that one out next week. I am looking for options to avoid boolean now.
I'm also looking towards the option on how to merge a object into another, for instance putting a logo on a surface (or imprinting it) without stl issues and boolean.


If I can offer you a possible tip in return: do you know netfabb?
http://www.netfabb.com/
http://www.netfabb.com/downloadcenter.php?basic=1

I got this tip from someone. It can show you the problems and even fix some of them.
But in the long run I want to just model more cleanly anyway.

In attachment the latest result. I used that facet look again for this one.
I also tampered with the layer height to reduce printing time and so far so good really :)

spherical
09-13-2015, 02:02 AM
Netfabb more often than not creates more problems while it's "fixing" what it thinks it's found. Tested it many times and it "fixed" stuff that was absolutely sound, then generated garbage where the mesh was clean. There is no automatic solution to find and repair discreet mesh errors. MeshLab sometimes does a good job. Netfabb does sometimes, too, but not often enough for me to bother with running anything through it.

If you want a good mesh analyzer, get a copy of KISSlicer. It will slap your wrist proper when you load a Borked mesh, show you where the errors are and identify what type of error it is. Then, open your model in Modeler and drive around to the location indicated to find the error and manually fix it. KS started out following the KISS principle but has grown so much that it really does need renaming. FAR from simple anymore but, IMO, it's the best slicer out there. Even if you don't use it for slicing, the analyzer is indispensable.

Good results on the new models! Keep goin'.

Tim21
09-13-2015, 07:40 AM
Netfabb more often than not creates more problems while it's "fixing" what it thinks it's found. Tested it many times and it "fixed" stuff that was absolutely sound, then generated garbage where the mesh was clean. There is no automatic solution to find and repair discreet mesh errors. MeshLab sometimes does a good job. Netfabb does sometimes, too, but not often enough for me to bother with running anything through it.

If you want a good mesh analyzer, get a copy of KISSlicer. It will slap your wrist proper when you load a Borked mesh, show you where the errors are and identify what type of error it is. Then, open your model in Modeler and drive around to the location indicated to find the error and manually fix it. KS started out following the KISS principle but has grown so much that it really does need renaming. FAR from simple anymore but, IMO, it's the best slicer out there. Even if you don't use it for slicing, the analyzer is indispensable.

Good results on the new models! Keep goin'.

will look into it and will do. I'm getting more and more these very peculiar results :)

spherical
09-13-2015, 02:05 PM
Are you meaning the surface quality? The banding indicates a hardware issue. Something is binding, slipping or just plain loose. If that happens, the steppers don't really know where the nozzle is, so reproducible indexing is impossible. Check that the bands oscillate between sides. IOW a bump on one side is a depression on the other. If it is protruding all the way around on the same layer, then that is a slicer issue. I seriously doubt it's the latter.

Tim21
09-13-2015, 11:28 PM
np with the printer. I'm playing around with some settings and made the layer building 0.2 instead of the usual 0.1. It prints faster but the layers are indeed a bit more visible.

This with some more "facet" playing that is starting to have a very peculiar look to things :)

ps have you seen this? a violin created trough 3D printing:
https://www.youtube.com/watch?v=gF0pOUBS3sg
https://www.youtube.com/watch?v=Gr-Vu4w35RM
I just love that fragmented, more organic shape, like you are playing with a bone structure of some sort.

ps2: https://www.youtube.com/watch?v=bafbvJOGiDI