3D STEP File Compatible: For supports export and import file


I still fail to understand how Autodesk was somehow responsible for the design of Formlabs Preform. Autodesk did not acquire NetFabb until late 2016, and the software was redesigned in mid to late 2017. By then Preform was in full swing, and had gone through nearly 30 revisions.

So please tell me how was Autodesk involved in the development of PreForm?

Last but not least, what is it that makes NetFabb a red herring? Is there something deceiving about it (or about its use) that we should all be aware of?

Edit: Preform uses the NetFabb engine for fixing defective models when they are loaded. Even if there is more to the NetFabb core in Preform, this is NOT the Autodesk Netfabb, it’s the original German NetFabb, as this functionality inside Preform predates the Autodesk/Netfabb merger by more than 3 years.


500,000 polygons is quite a lot. and the ability to see the individual facets is dependent entirely on the size of the object.

Here is a 37K triangle object. one at normal size, the other at 5x the size. Obviously you can see the individual faces at the larger size, but the normal size of the object is 20mm in length, there’s no way you’ll be able to see any thing, simply because the printer cannot resolve faces that have edges smaller than 0.2mm

Additionally, many programs like Solidworks do smart tessellation. They assign (break down the curves) into more triangles to achieve a smoother curve, while the flat surface have very few triangles. So a properly tessellated object can still have less than 500K triangles, yet look perfectly smooth.


I’ve spoken to Autodesk to enquire about the purchase of Fusion360 or Netfabb, because Pre-Form isn’t full proof when it comes to basic STL repairs.

Which is annoying in a professional environment as I don’t have all day doing tweaks, because time is money.

Also, if Preform isn’t an integrated solution it isn’t a long term solution it not a real solution.

I suggest you speak to them too, because there are some similarities between some of the Autodesk UI architecture e.g. Recap Pro & Print Studio. We already know that Netfabb involved with Pre-Form repair tools, but to what extent is something I couldn’t get a straight answer. Netfabb was developed in Germany and said direct there “was” involvement, but other Autodesk members said Formlabs only acquire the their OEM version for development.

While Preform have told me about their software development process and couldn’t tell me about the lead-time or scheduling or even an approval process when & how to treat a reported bugs or add improvement features.

In another word, if there is a software issue with Preform, there doesn’t seem to be anyone I could speak to directly anywhere in the world. I’ve tried.

One positive note is that Formlab is still very young and very willing to listen to communities concerns, unlike the older and more professional companies, which doesn’t tell you anything for product improvement. Unless you show them $$$



Not for me, it common to have professional 3d models with intricate details for print (especially injection moulding parts) and it usually runs into few millions of polygons or else end up with a pixelated prints.


If it isn’t an Integrated solution, it isn’t a long term solutions, especially in a professional environment, because time is money.

Most graphic designers I know (directly) prefer working with raw data e.g. Vector vs Jpeg similarly when capturing in raw than JPEG in cameras.

Are you saying jpeg is more important than raw or adobe illustrator? Eventhough Jpeg are designed to remove workable data, unlike raw.

STL file where never intended to be modified (invented before windows) and the only reason 3d printing industry uses it because it free, not because it easier to work with compared to raw data.


Not for me, it common to have professional 3d models with intricate details for print (especially injection moulding parts) and it usually runs into few millions of polygons or else end up with a pixelated prints.

That sounds like it is a memory limitation. Though I’m not sure I understand how a laser tool path could result in a pixilated print.


Does Solidworks have any control over tesselation on export? MOI has outstanding controls on curvature while minimizing faces on flat surfaces. FormZ has more of a global approach but also does a very nice job. I made some threaded parts and fit was tricky but I was able to do it. I actually had more of an issue with an acme thread drooping rather than the resolution of the model. You really need to not go courser than 50 micron when doing threads from my experience unless your making super course threads. Finest thread I was able to do was a 28 pitch screw and that was difficult. As far as dimensional accuracy the prints seem pretty good and I haven’t had any issues with that.
I do agree injection molded parts you may need to tweak the files to handle the better tolerances. I assume most if not all resin based prints are subject to shrinkage as they cure to some degree and 3d printing in general you have to contend with the stepping of the layer thickness that affect surface quality even if not easily noticeable. Just the nature of the beast. This is probably why subtractive prototyping is still more accurate for FFF but not always practical or possible for all shapes.


I think Kevin is not referring to pixels, but Voxels.

STL is triangulated geometry. The triangles describing curved surfaces are planar; where large enough, those planes can become visible in the print when using a printer with the resolution of the Form2.

MOST models you can download online will show planar geometry because they are modeled to render good, rather than be printed.

I routinely model at 80 million polygons- but because of preform’s 32 but memory limitations, I have to decimate the file down to under 140 mb or preform won’t input it.

It also has a bug importing several STL files in a row… claiming subsequent imports are “bad” geometry when I know they are perfect.

However- the problem with STEP or other formats that have spline surface capability is that surface modeling has real problems with solidity.

It doesn’t matter in MILLING because the mill software is only deriving a TOOL PATH in space… the cutting head doesn’t care which side of the surface is solid… it will trace the surface in free air if there is nothing to cut in its way.

But all Print technologies REQUIRE the model to qualify as a topological solid. The printer is not CUTTING a preexiuating material to form- it is deposting solid material where the model is solid… so the file need to know what parts of the model ( all the way thru ) are solid- not just the surface acutting head will trace.

STL- by eliminating all spline data- stitch data- and not having to deal with coons patches versus other types of UV geometry- has far less issues maintaining solidity, and is far easier to rectify lack of solidity.

Its no big issue to convert subD or surface data to STL- unless your software is not capable of resolving or detecting solidity issues.

Also- the idea that STEP files or IGES files are smaller is a myth.

They are generally much smaller when modeling simple shapes like engineered parts composed of cubes and cylinders… but to model a highly detailed organic figure in nurbs with any level of accuracy requires so many fit points per patch and so many patches that the file size quickly equals or exceed an STL model that can capture the same level of detail.

And the only way to reduce the file size of a surface model is to LOSE overall fine resolution- whereas STL models Can be intelligently decimated down to a target size and preserve fine details.

Used to be, we only ever created nurbs models because many milling apps would ONLY accept surface data… tho for 15 years now, most modern milling apps can mill directly from STL as well.

Parametrics and surface modelers are fine for engineering… but for anything organic, Voxel stl modelers and subD apps reign supreme.

And Preform accepts both STL and OBJ… so-

The last figure I did in nurbs ended up being a 1.2 GB STEP file… Preform can’t import that much data even if it could accept IGES or STEP.


I found that you can import files with about 6M polys (about 300MB). Anything larger than that generates an error. The cowboy STL file is about 288MB and 5.89M triangles.

However, that only applies to any single model. You can import multiples for a higher combined total. Each one of the Space Marines is a little over 2M polys. The combined total is 8.35M, the combined FORM file without supports is 718MB


I’ve been having a different issue.

I often decimate and export multiple models out of Freeform. They export as separate files.
And I know from 18 years experience that Freeform exports perfect qualified solids every time.

When I import one model into Preform- it comes in fine- but when I try to import a SECOND model… it ALWAYS returns a File Broken error.

However- if I then save the file and CLOSE preform… and reopen the file… it will import the next model without a problem.

But here’s the really weird part… this only works if I re-open the file by clicking on the ICON for the specific file on the desktop or in its folder.

If I reopen the file from the taskbar recent files list… or if I relaunch preform and then try to open the file from inside Preform… I still can’t import additional models.

I was hoping the recent update would solve the problem, but it hasn’t.


After 12 months of complaining, 64 bit is finally here:

and after some initial testing it seem to process both files (.stl and .form) twice (almost) as fast than before.

Let just hope they will add STEP import option as well, because others already starting to do this for the sake of file handling.




You can keep your “High School” mathematics & your arguments about PreForm “STL is more than sufficient to meet/exceed the resolution of the Form2.” Go get some structured light data or inspection measurement data to support you claims. Because all I have read is an opinion & my opinion is different based on my experience during the conversion process & my requirements.


I’m an Engineer 20+ years with CAD/CAM. Many don’t understand fully what you are saying because they probably have never done it in a real manufacturing environment and don not understand the processes used. I’ll put it Basic Terms: STL sucks for Engineers. The way Engineers use NEW Products is also dependent on what other industrial products stipulate (i.e. what are the current industrial standards) . I have read Kelvin’s many examples and he has given numerous industrial examples that I agree with and support his argument. That tells me FormLabs may not understand my industry and thus my needs as a Design/Development Engineer well as they think they do.


Depends on which G code they are using… Is it 100% industry standard? I have found many are not… lol


What’s the latest on this. We are now running into file size issues with STLs as well and I’m typically only exporting 100k-300k triangle models. It’s making it very difficult to send these files over email to service bureaus. During our revision control implementation we’d likely be looking at alternative file formats. We would also need support for these file formats from formlabs to make printing a bit easier.


Summary & Current issues:

From an Engineering perspective do you know…

1 - Anyone would store SLT file (modified or not) into PDM workflow as part of ISO9001 traceability?

2 - Form 2 machine uses 32 bit? Which makes file transfer very slow when importing into the machine, even though this machine was designed during Win 7 era.

3 - Professional engineer who uses Solidworks, Inventor, PTC and others cannot generate high polygon count due to limitation of the “solid” 3D CAD packages itself? Which unlike surface CAD packages (or Art CAD).

4 - How many R&D firms uses cloud based storage without questioning the validity of it online security?

5 - How many injection tool makers, production engineers or CNC machinists you know of would prefer STL file over STEP?
I’ve seen a number of Carbon 3D customers would question the use of STL file.

6 - Form2 machine needed to send to another country for Z height calibration?

7 - I own the 3D CAD data, but I cannot delete my own data & log from my online dashboard? European GDPR law will have an issue with this.

8 - Many firms want to make use of “weekend or overnight build” to maximize productivity? e.g. no z height model stacking on top another to make full use with the available printer volume (more supports removal though) Also, print quality is difference when finished print left ‘hanging’ for too hang.

9 - Resin expire date display on software (even if it just estimation) would greatly improve productivity?

10 - Any professional Engineers who uses 3D “solid” CAD packages would prefer to modify using STL files over STEP / Parasolid?

11 - STL polygon file repair is not a thing for FDM machines? Not supposed to anyway e.g. CatalystEX, unless the polygon are very corrupted or incomplete.

12 - Optical window cleaning is a continual concern? Cleaning is an issue and window seal can come off and hard to replace, I even supplied a harder screw type to prevent less chance of screw head wear to their maintenance staff, but was skipped over during repair.

13 - The cost of resin (per litre) is actually higher than professional SLA machines 10 years before?

14 - Tray are heated? But slow to reach up temperature.

15 - The machine is noisy enough during printing? It best not sit right next to your office desk. Also require vibration free / clean room environment.

16 - The support structure produced are too thick (even at smallest settings)? Making “post process” more difficult compared to commercial SLA models.

17 - ‘No pause option’ during print from power outage (Protection) or from swapping out low cartridge to survive long weekend print?

18 - Orange cover easily get fingerprint marks (even with gloves) and are not easy to clean?

19 - Digital spirit level seem off compared to real one?

20 - Difficult to remove models from build platform without damaging the platform surface?

21 - Orange tank ghost up optical window after few litre of use? LT tank much longer

22 - Formlab recommend Orange or LT tank to remove it resin back into cartridge after use? To minimize seal damage on tank optical window to prevent leaks.

23 - Model require orientation before slicing? But this will increase Z height build time & more build supports attached to model (more ‘post work’).

24 - Commercial Machines never recommended for using mesh repair programs such as Netfabb, Mesh mixer & etc.? Never heard the need for it, until the rise of these enthusiast desktop machines begins in 2008-2011.

25 - It doesn’t have print failure detection e.g. auto pause + camera + ‘load cell’? e.g Stratasys F123 series

26 - It would’ve been helpful to give further design support for Engineers?
e.g. “Prototyping” tests: for Visuals, Tactile shapes, Functional housing and mechanical inc. Jig & Fixture and rapid tooling or “Production” parts that replaces: vac casting, vac / pressure forming, low pressure over-molding and other low quantity manufacturing.

27 - Power consumption / operational cost (inc. decibel) is a dirty word within 3D printing industry? Including (perhaps especially) with commercial machines. Particularly damning for high end FDM or SLS printers and admittedly less so with Laser SLA.

28 - You have to be aware of hidden cost (beside power consumption)? e.g. “consumables” has 1 year expire date, so you cannot overstock with wide ranging materials unless you have plans to use it and “maintenance cost”.

29 - Engineers are not “printing technicians”? Engineers prefer spending time designing and solving real world problems rather than messing with 3D printers (software & redesign models) whole day/s just so that what you “wanted” are also what you “get” after prints.

30 - It software doesn’t account for “refillable options” to show true resin amount in cartridge? Note commercial machines treat you like consumables like cost of Kodak / Polaroid films vs hardware, but I doubt Formlab had the same intention.

31 - Formlabs Tech Supports are more engaging than most commercial machines (positive feedback)? Even though it still taken a year to resolve things like the 32 bit to 64 bit software. At Least they are trying.

32 - Built estimation slicer (Preform 64bit) are still slower than a 10 year older commercial FDM software?

33 - Used cartridges, gluepy IPA and Cleaning tissues disposal are still an issue? No network, no reclaim (unlike Stratasys Cartridges)

34 - There are no machine rental or better still ‘trial’ scheme? Once you buy it you are locked into their eco system e.g. consumables & support. It hard for any small medium business to folk out 10k and not knowing what they are buying into especially with ‘resin’ based machines. (though Form2 cost half that). Cost argument is a nightmare of owning a machine v outsourcing e.g. post processing, consumables, maintenance, running cost & others are major factors to consider.

35 - Print / slicing software doesn’t include mesh editor? For those who use STL files (which is everyone).

36 - Having “upside down printers” (build platform move upward) are great for printing small scale, making materials more easily swapping? For quantity run, it much better to have “right way up model printers” (built platform moving down), because of “diminishing returns” benefits (even though the machine has to accommodate bigger volume resin tank).

37 - Formlabs tries to introduce maintenance program? Which is great, but the problem is you have to send it, rather than having the ‘option’ for (even 3rd party) service engineers coming over fixing it. Much less disruptive to businesses.

38 - All resins has a shelf life? Which should be store in a controlled environment (like those commercial resins) to extend the shelf life (temperature & moisture), but Formlab ‘seem’ to formulate their resins to have a shorter shelf life for the sake of having wide storage range, sunny cooked up dusty warehouse to indoor office environment. I really wish people would treat any resin with a bit of respect.

39 - Doesn’t have 3D mouse option? Though admittedly medical student are not engineers (and 3d art designers) who used to manipulate 3D object on 2D screens. Medical sector seem to use ‘AR’ input instead, also engineers doesn’t need 3D printing for themselves as they can test it “visually” (visual test) on 2d screen, they only printed in 3d for those who are incapable of seeing 3D objects on 2D screens.

40 - Print estimation time are not very accurate? As Form2 doesn’t take into account with the machines startup / heating time before printing.

41 - Resin tank walls are too shallow & wiper require firmer design? At best of times wipers tend to splatters to the inside wall of the orange cover after some use. At worst, the wiper (single point attachment like a ‘pivot’) and can flicker violently when caught up solid resin stuck to the tank optical window (messy).

42 - Tank optical window can get cloudy? Formlab uses software to ‘estimate’ this cloudiness, which are useless in practice. Because in real life, it always better to inspect the optical window before ‘every print’ and it would be far better to have a camera inside looking from underneath the optical window to inspect cloudiness remotely without the need of opening this UV protective orange cover and exposing the resin to outside condition to the minimum. You then then automate the wiping action during visual inspection (remotely) if it a non-transparent resign. Less resin mess are generally good practice, avoid touching/ handling resin directly,

43 - The direction of the resin level sensor are questionable? As it can work only in the best of circumstances i.e. if the resin tank/ tray where changed from the machine, the resin remain sticky to the side wall (after being sloshed around during the change), interfering this level sensor, even though the resin itself is at it “max” level.

44 - Unreliable & questionable ‘valve’ design underneath the resin cartridge? Which basically just a rubber plug with a ‘perforated cut’ in the middle. I had to "cut a small incisions " myself rather than relying the machine to “squeeze” open this perforation to allow resin to fall through into the tank.

These are some of the daily Engineering challenges I have to go through with Form 2 & resin in general. I hope Form 3 would address some of my Engineering concerns and perhaps improvement in reliability.

If you are an artist with surface A models or figurines to print then that is great, because their design focus doesn’t seem to include engineer users (and I really wish this isn’t true), because Formlab “could” be a great potential to replace commercial SLA machines (not yet) as Formlabs tend to focus their machines for “ease of use and some level of autonomy”, but proven unreliable in Form2 and not good enough for use in Commercial (SME) / Engineering environment.


Quite a list.

My engineering projects were in the 2 -55 story range. So although I was a mechanical engr. for a few years I find the FL2 and FL 3 to be just what I need for my current projects.