stl format, how smooth?

[VetteMan]

*.stl is polygon. If you try stp or iges you will get better results. Here is a stp example:
http://www.boatdesign.net/gallery/showphoto.php/photo/2989/cat/500/ppuser/276
http://www.boatdesign.net/gallery/data/3054/2763blade.stp

[albynism]

Thank you for sending me the links to the files. I see that iges and stp are nurbs based. I will try that :)

Another question is if i have two watertight mesh/nurbs intersecting. will that be printed out as one single object? I figured I ask this question here cause the printing service is so slow at responding, poor customer service I say. If someone can recommend me a better, cheap 3d printing service (preferably in australia) that would be great.

Thanks again. :)

[JimCarruthers]

You will want your object to be a single shell, don't give the rapid prototyper a jumbled mess of objects and hope they come out right.

STLs are mesh files and are what you want to output. The accuracy you need depends on the accuracy of the technology being used to make it. If you're looking for a low-cost "concept model," then the mesh probably doesn't have to be any more accurate(in Rhino you'd use the "max dist edge to srf" setting)than, oh, 0.010" or 0.005".

[albynism]

Okay they are now single shells. I have multiple objects that I want to assemble together later. Do I give them a single .stl file with all the objects in the file. or give them a separate file for each separate object.

I just boolean them to form one shell, this causes some n-sides polygons. does it matter?

Thanks :)

[JimCarruthers]

I don't know anything about "n-gons," but STL files are tris only, so they'll have to be converted if it's automatically or what.

[albynism]

yup silly me. its going to be tris anyway so i guess it wont really matter. thanks.

[saiko]

STEP is the best if your rapid prototyping machine suports!

[albynism]

the printing place only support stl and will charge me processing fee if i submit it in other formats... hmm :/

Also the model is made in maya out of several polygons. I can easily boolean union them together to form a single mesh and export the mesh as stl in rhino. to convert it to nurbs for igs and step may be a little tricky though. I can convert the polygon to subd and from subd to nurbs. but im not sure how to EASILY join multiple pieces of INTERSECTING nurbs properly to form a single nurbs surface. the nurbs trimming feature in maya leaves much to be desired. :/

[JimCarruthers]

Don't try to convert a mesh model into NURBS to export as an STEP that will just get turned back into triangles for the STL, that's...well, nuts.

The service bureau will specify STL only so that it's not their fault if they tesselate it and the result turns out too rough for your taste.

[albynism]

Jim that makes sense. Do you know how smooth does the stl have to be to get a smooth surface? If you can direct me to some images of stl wireframe and the actual 3d printed output that would be really helpful.

"in Rhino you'd use the "max dist edge to srf" setting)than, oh, 0.010" or 0.005". how smooth is that?

thanks!

[JimCarruthers]

Well, it's as smooth as you want without the file size going nuts, it totally depends on the scale of your object and how it's being built. The RP technology itself isn't perfectly smooth so the mesh doesn't have to be either, the tolerances may only be +/- 0.010" or greater. If you have no way of actually specifiying a number for the deviation of the mesh then you'll want to look at it without the 'smoothing' that gets applied in the display, and know that the edges will be softened a bit.

[Ls3D]

Well with new UV cured bubble jet like output (polyjet) parts CAN be smooth, check the tolerances claimed for one vendor:

Polyjet modeling offers outstanding surface features and fine details. With the highest resolution of horizontal layers in the market, 16 microns (0.0006 inches), RPDG can build thin walls down to 0.6 mm (0.024 inches) and eliminate any stair effect on complicated curved surfaces. Polyjet is most suitable for small detailed parts.

-Shea

[albynism]

Thanks for the reply. I already send my file for prototyping. I have a feeling its going to have the stair stepping effect. although the file is like 10 mb each and its nuts having to wait while uploading them to each individual vendor for quote purposes. I guess I just have to manually sand it after I receive it. cheers!

[albynism]

Hi. just thought Id give you guys a bit of an update. I just received my 3d printed sculpture. First impression, WOW! The whole thing was smooth and not faceted as I thought. I thought it would look like polygon on the screen with all the hard edges showing. but everything was smooth EXCEPT there were annoying small horizontal lines on my model that I assume is called "Stair stepping" effect...

So.. Is this how advanced 3d printing is at the moment? necessitating to manually sand your work after it is printed? Ive seen many websites advertising accurate, smooth printing etc and I used one of the most advanced process (SLS) using somos resin instead of brittle plaster like material but I'm still not getting 100% smoothness. Is it my file or is it the technology or is it a cheap beadblasting? Some areas are smoother than others and I dont think its caused by low polygon count because some of the areas that have high poly count still have stair stepping.

Opinion is welcome. Thanks all.

edit: I will post photos if you are interested but to give you an idea it is smoother than the stuff you see at 3darttopart but still not 100% smooth.

[JimCarruthers]

Basically, yes you do have to finish your models if you're trying to make some presentation piece out of them.

The highest-resolution methods that will give absolutely no 'stairstepping'(layer thicknesses on the order of of 0.001" or less)use a delicate wax that you then investment cast a metal part from. SLS is one of the "rougher" methods, I've been doing some stuff with it lately, but it makes for a tougher part than SLA, that you would use to test how something is going to work when molded out of plastic.

[albynism]

Sorry I was meant to say SLA instead of SLS. I had mine done using SLA technology. not SLS

I'm researching on what Ls3D was saying about objet. I'm seeing some pretty impressive high-detailed stuff using object with the veroblue material. I will still need to obtain a sample of it though just to be sure. Its also a less expensive option too.

Check this out http://redeyerpm.com/images/mouth.jpg

I will keep you updated. :)

[billrobertson42]

Where are you getting your quotes from for the objet?

BTW, if the layer thickness really is what they say you won't notice any stair stepping.

[albynism]

i got the quotes from www.arrq.com.au (http://www.arrq.com.au) they are based in australia.

[rootoftwo]

Susantio:

Hi. I ran an SLA 5000 for a number of years making prototypes for Evenflo baby products. Jim Carruthers is right. You do have to finish your models if you're trying to make a presentation piece out of them. This is one of the many misconceptions about rapid prototyping - they still require hand finishing. As an example here are 2 prototype nipples (for baby bottles):

http://i108.photobucket.com/albums/n11/rootoftwo/nipples.jpg

The one on the left is straight out of the SLA (I hope you can see the stair-stepping - I got lazy and threw them on my scanner). The one on the right has been sanded and polished. The green colour is because it is filled with urethane - it was used as a master pattern to make a silicone mould for flexible urethane reproductions.

You can use parts straight off machines - but only if you don't mind a certain amount of stair-stepping - it is not really exhibition quality and there is no substitute for sanding. Believe me, I spent 5 years sanding. In the same way that you can't ever get away from pixels with a raster image you can't get away from a certain amount of stair-stepping with RP as it is today.

It really comes down to what you want to do with the part.

Cheers.
http://designedobjects.blogspot.com/ - my research blog

[albynism]

Artcore:

Thanks for scanning those models! I can see the stair steppings and they look exactly like my SLA model. and the material are the same too.

Thanks for clarifying the misconceptions about rapid prototyping. The models Im doing are fairly complex with interlocking parts so sanding each parts to be pattern ready takes a lot of my time.

Have you tried objet yet? They are relatively new technology and only a few bureau have them. they look smooth and the costs are relatively cheaper than a beadblasted finished SLA model. I have read that the finish is pattern ready for mould making but I have yet to receive the sample...

[rootoftwo]

Susantio:

Yes, I've seen some Objet parts. A few years' ago when they beat out 3D Systems patent suit (2000? or there abouts) I was really excited that the entire RP industry was about to go through the sort of price meltdown that software went through around the time that Maya changed its prices from $15K - $6K. My money would have been on Objet to lift the crown. It hasn't worked out that way. 3D Systems merged (bought) DTM and now owns the SLA and SLS processes. The major problem is materials and maintenance. There are only a few companies making materials for these machines (regardless of process) so they can just about charge what they like - $800 a gallon for the resin I used to build the nipples in the previous post (bear in mind that you need to put 67 gallons in the machine before you turn it on). Maintenance on a machine is another huge burden. I don't know what the current price on a solid state laser for an SLA 5000 is but it was $50K in 2003 - you need to replace it every year (which is a huge difference from replacing a couple of $75 UV lightbulbs in an Objet). Basically, if your machine isn't running you are losing money. All this and the company I worked for didn't get much change from $500K to buy the machine and build a room and post-processing facility (you need to put parts in a solvent bath and then a UV 'oven' before you can handle them) for it in the first place. That said, I paid off this initial investment within the first year in unmade tooling errors and enabling the company to go from 12 to 36 new products a year.

If you think about the nature of these processes it is a process of layer manufacture - it is micro lamination (even if there is an interpenetration between layers) it essentially reduces a 3D object to a stack of 2D profiles - no matter how thin each layer is there will be a step between. I'm now finishing my PhD on how the use of these technologies is transforming the practice of certain artists, architects and designers. If you are interested in the various types of process out there have a look at this page on my wiki: http://designedobjects.pbwiki.com/RESOURCES for links to the sites of most of the main companies. Here are some other links that might be of interest:

http://home.att.net/~castleisland/home.htm
http://www.cc.utah.edu/~asn8200/rapid.html
http://www.rapidprototyping.net/

Cheers.

[albynism]

Thanks for the great post! very informative. I did a little research on my own and I think Objet is the best option for what I do and what I am trying to achieve.

The envelope that suppose to contain the Vero Blue material arrived last night and to my dismay, it wasn't there! :sad: I'm suspectiing it might have been lost in transit as the envelope was a little torn when I received it. Nonetheless I wrote an email and received a reply this morning.

"Sorry to hear the sample was not in the post, very disappointed with
Australia Post.

There is stair stepping within Objet parts (vero blue is one of the Objet's
materials available), but it is a lot finer compared to an SLA part (16
micron v's 75 micron steps)"


I guess I'm just going to get a prototype done using Objet and compare the differences.

[rootoftwo]

Susantio:

I'm glad to be of help. If you get carried away with this you might want to build your own RP machine. Full instructions here: http://128.253.249.235/wiki/index.php?title=Main_Page

I find the list of materials that can be used in it particularly interesting:

Gypsum

Plaster

Playdough

HotGlue

Putty

Silicone

Rubber Cement

Conductive paste

CircuitWriter Conductive Ink

Solder It-Silver Bearing Solder Paste

Low melting point alloys

Icing

Peanut Butter

Cookie Dough

Chocolate

Caramel

The last five on this list conjure up all manner of weird and tasty possibilities :)

Also, I can highly recommend this book: Gershenfeld, Neil, 2005. Fab: the coming revolution on your desktop - from personal computers to personal fabrication. (Basic Books: New York). An short article about Gershenfeld is here: http://www.wired.com/news/technology/0,1282,69495,00.html

[albynism]

LOL interesting website! I wish i have the money to afford an RP machine next to my desk. for the time being I'll stick to getting them outsourced.

hmm i wonder if they can make SLA out of sintered chocolate powder. that would be edible. but makes a very expensive chocolate treat. :)

[CGTalk Moderation]

This thread has been automatically closed as it remained inactive for 12 months. If you wish to continue the discussion, please create a new thread in the appropriate forum.