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

The main reason to use a STEP file is part accuracy. Obviously you would edit the part in CAD before Preform.

I’ve run into this issue myself. Getting decent parting lines from a STL can also be a pain because they are never exactly correct. Fusion 360 can scale (expand) STL files to allow for part shrinkage. You can also directly modify the STL in Fusion but things are limited (the part is a dumb solid when converted).

Just a guess but I would bet that Formlabs doesn’t have arc segments built into their code. Currently everything is straight lines (controlled by the mesh density). True arc segments are out there already for the Arduino (etc.) based CNC software so it might not be that hard for Formlabs to implement.

Doesn’t the printer run G Code internally? If that is so, it should be able to drive the laser spot along curved lines. I seem to recall our old 3D Systems SLA 250 did steer the laser spot to grid points, but the grid was very fine; certainly way finer than the roundness of the laser spot, or the cure of the resin… Not much point measuring your lumber to the wavelength of your laser, if you then use your thumb as a guide for yor hand saw!!

It may indeed be some form of G code but it’s coded by Formlabs. They don’t currently need arcs so they may not be in the code.

I’m not sure whether STEP files would give much advantage in Preform–STL doesn’t necessarily have as much precision but I wouldn’t expect the difference to be notable in the actual print, other than that you just need to make sure your STL file has enough subdivisions to avoid surface facets and it’ll probably be indistinguishable from a STEP file.

I am almost positive it is not G Code. Look at the OpenFL code examples of the .flp files. Open FL is only currently for the Form 1+ but I see no reason why they would go backwards after writing their own code to drive the printer.

Besides STL fine tuning issue, file storage / backup is a nightmare as well in the professional environment as PDM isn’t really design for this situation.

It true Form2 has 1GB of internal memory if we we ever decide to do a re-print on some thing few years ago while studying what tolerance works (trying to measuring things accurately is annoying), but the file most likely to receive some design changes. And in this case we had to all the way back to use original CAD file to generate it.

Because most of our design parts require fits with another components (SLA or otherwise). Also all our designed component is to conform with real manufacturing process by default and NOT generally conform to a specific FDM, SLA, SLS machine. Meaning we had to generate 2 copies and it a redesign every time we want to print something.

In turn this is becoming a very time consuming process when we want to integrate 3D prototyping FULLY to be part of our Product Development process, rather than print something every once in a while.

Even GrabCAD or Carbon 3D printers recognise this situation (STEP enabled imports) as they specifically sold their machines to those in a Mass Production environment.

I’m aware many people use STL file, because it a simple Z,X,Y coordinate that you can even read it in Note, which allow for many open source 3D software to pop up over the years to manipulate STL one way or another.

This is why I’m proposing STEP import along side STL. Just like GrabCAD Print, Carbon 3D or even Makerbot introduced last year. Download MakerBot Print 3D Printing Software | MakerBot 3D Printers

I’m not sure If Preform is a g-code curve path or not, as both Netfabb & Formlab was unwilling to take control / responsibility of it software development as I have concurrent issues using 32 bit oppose to 64 bit PC. Even though 64 bit Mac is available and yet they claim Preform was written by NetFabb, which was even listed with Form 1 machine (not 2).

I tried to resolve this issue with them behind the scene, considering most of the design that i do have NDA issues. I even went to the Expo and explain my frustration without any real solutions (one said use Mac the other said use Autodesk online which is 32 bit), but both companies finger pointing between Netfabb & Formlab about the software development and unwilling to give me direct contact to the software engineer in Boston and that was 4 months ago without any update. All they say there is no way of knowing when or if 64 bit will ever be introduced.

I short Netfabb is a good software to use, as it has STEP & even native SW import, but with too many overkill features in which I don’t need as I tried to rely as much design work to one software i.e. Solidworks.

Where did you hear that Preform was written by Netfabb?

It’s true that Preform has built in code or a module to do mesh repairs, that was provided by Netfabb, but the Preform code itself??

I seem to remember the repair feature being NetFabb but I could be wrong.



I heard it direct from both Autodesk & Formlab sales rep, but Autodesk was quite tight lip whether to include Form2 as part of Netfabb software machine library. If this is the case, you only need Netfabb instead of Preform.

Also, Netfabb have a resolution definition settings when converting from (STEP) > mesh (STL), which is something you definitely need, but currently only SW, NX & Creo have it (the ones I know), because by default this settings is needed to properly convert and fine tune between visual quality VS file size when managing STL file.

But one of the benefit (if done properly) is that with STEP you don’t even have to sacrifice visual quality (visible facets) for the sake of file size, which seem to crash more often when I load 200 - 300mb file with 2-5 million polygons, which was the limit for 32 bit machine.

I even ask my Formlab re-seller to test & confirm this for me.

But if you ever used Professional 3d printer software like CatalysEX from Stratasys you can actually print it as it is including polygon errors e.g. missing or even duplicated facets (due to bad conversion from Solid > Mesh)

Or even SLA Viper from 3D System as you will notice polygon repair is very easy to use with minimalist repair function, as it should be, because overuse auto repair functions (which ever configuration as featured in Netfabb) will result in polygon artifacts, which might as well use MeshLab or other open source to spent half of my time repairing STL “manually”, which I don’t have the time for.

I’ve been doing 3d printing for 15 years (professionally) in various 3D printers and STL repair is something that I rarely have to do without sacrificing visual quality, until now. And everyone I know who used CatalystEX will say Preform isn’t quite there yet.

I do like the simplistic of Preform repair more so than Netfabb (which isn’t a true Native Parasolid as it require conversion), but it does have some really useful features in which Preform should take notes (My personal preference):

1 - Import as Native & STEP, as we create & manage 3d contents using Solid / Parametric programs and NOT Mesh programs like 3d max, Sketchup or Rhino.** Similar idea to CAM www.youtube.com/watch?v=6yA9Bn8EYhY
(all be it using STL, but at least the resolution is as good as any mould tool. Considering I use this 3d printer to simulate what the part actually look like if come from a mould tool design, I’m not even talking about draft (& lesser extent radii), but things like contours, split lines, component fits and even physical tests as would have done to production parts as well.

2 - 64 bit, to increase threshold on polygon correction
3 - Resolution settings when converting, rather than fixed internally by default (possible to do it in Parametric Programs in Solidworks, but lacking key features as they are not in the business of creating mesh features in a solid program)

Resolution control are also important to minimise visible facets on model, but it compromise file size if using mesh type files e.g. STL & Obj (contain colour info if I recall)

Parametric Solids (such as Parasolid or STEP or native) tend to have much smaller file size while having true contour. It true the slicing software still have to convert into STL (If they included this feature), but at at least it handle the conversion internally, which means dynamical control resolution settings (while using RAM resource, rather than rely on SSD read speed), because it means I don’t have to save the same model in various STL resolution settings just to test out if Preform is capable of handling it or not.

SW resolution export settings:

Creo resolution export settings:

This is what happen when it has a loading or saving error due to file size.

It possible for me to trick the software to force print it, but not always work.

I need to achieve high resolution so to minimise visible facets (or scan line).

Resolution setting 5.0d down to 2.0d are just no good (finer settings = smaller the number), which I wanted to use resolution 1.0d to give me unnoticeable scanline or facets after print, but this is borderline impossible with the current software even though the hardware is capable of printing it.

As I ended up removing all of the rear detail of this battery cover and joining it to the main housing in order to achieve 1.0d (a dumb model), but unable to show the removable cover with batteries inside. The only way is to reduce resolution settings down to 1.5d in order to have a removable cover, but even so the facet is still visible. Even if sanding it, I have to remove a lot of material (and time) to have a smooth contour.

This illustrate the current limit of this software, Formlab say 200mb (combine total for entire build platform of parts), but I manage to get to up 291mb

So, it means sometimes I can only print 1 to 2 model at a time even though I still have half the space available (on X & Y).

Which is why 64bit on PC is so important as I can fully utilise all my RAM when performing STL check & Fix, loading or even saving into .form file while using Solidworks to generate it. It also proven by Formlab that 64 bit do work with my model when they use it on their Mac.

Besides opening the STL correction threshold, I’ve tested out the repair features on 4 different slicing software. Below is from 3D System (include STEP import and STL check & fix feature), which where first import to Netfabb before going into 3D system to tell me what is going on.

It shows most of the Netfabb fixes & repair where overkill and sometimes overdone even if using some of it auto features, where polygon count could go higher (i.e. file size) with virtually no improvement, (unless you knew exactly how to configure the repair function.

I’ve also used solely the repair function in 3D System (as it utilises 64 bit) before importing the STL clean up models into Preform and it works without crashing, As this allow me to bypass the auto checking feature in Preform, because I sometimes get crashes if I import direct to Preform as the auto checking feature require RAM in which 32 bit can’t handle along with it file size 50 - 200mb.

Anyone who use Parametric Solids > Mesh will know that polygon error is fact of life during this conversion process, more potential error than if export from native mesh programs. Therefore Preform should also accommodate those who use Parametric Programs (SW, Creo, NX) alongside those who operate STL / mesh programs (SpaceCaim, Rhnio, 3D Max, Maya, FreeCAD, SketchUP & etc.)

I’m not here to bash Preform, because generally it a good piece of software and it easy to use, but my frustration could easily reflect those within the entire engineering communities: who uses Parametric Programs while knowing the machine capability isn’t being matched up by the software. This is why I strongly recommend Preform to include:

  • STEP import (include native ideally)
  • Own Resolution Settings
  • 64 bit


As a workaround, would it be possible to locally decimate the geometries of the STL file, with meshmixer for example ? I think Solidworks could do with a selective resolution tool when exporting STLs but we can keep dreaming it will never happen. This may allow you to print your parts in full.

1 - It possible to use Meshmixer or Meshlab (or any other STL editor) to manipulate the geometries, but it very intricate (annoying) when trying to open up 25- 50 micron equal distance hole for example for transition fit with another components. It is best done using Parametric Solids programs to do this kind of work.

Although, Meshmixer is quite simple to repair (not as easy as 3D System Sprint), but it is a 32 bit software, meaning the problem still exist if opening & repairing STL files bigger than 200mb. You can easily tell the differences between MeshMixer(32bit) & MeshLab (64bit) if you have a complex 200mb+ file.

2 - SW, Creo & NX already have selective STL resolution tool when exporting, but the problem is that as you crank up the resolution, more polygon errors being produced. In-fact, officially SW stated that they don’t support anything above 0.5 million facets per model (around 5.0d ~ 10.0d res), because it not their business to refine this mesh feature considering they are a parametric modelling company not a mesh based modelling software.

Formlab & I have already tried with their Creo (they have Creo licence as well), but it has the same problem when exporting / convert into mesh.

Image below is an extract from Netfabb Manual, to explain the nature of Parametric Solid > Mesh conversion.
As, I’ve mentioned before, STL will always be an approximation compared to Parametric Solids.

Since Preform is a mesh based program, thats why it should include:

- STEP import (include native ideally)
- Own Resolution Settings (when converting)
- 64 bit

Who else here uses professionally 3D printers on a daily basis? Because most of the polygon issues are taken up by the slicing / mesh software not Solidworks. For example, you don’t have to clean up the STL model in another software before importing to the Stratasys slicing software (or 3D System Sprint), in fact CatalystEX program are so robust that they don’t even include the repair button while still giving a decent FDM print.

I’ve worked with professional SLA machines 10 years before and I rarely need to cleaning up any STL (until now).

Meshmixer is a 64bit app., not 32bit, and I have opened large STL files before 300MB+, although it’s very slow working with them.

The difference between MeshLab and Meshmixer is in the way each one handles the model rendering. The most time consuming part though is when you apply mesh modifying operations. MeshLab crashes a lot more often than Meshmixer when dealing with it on large meshes.

You forgot to highlight “precise approximation”.

It’s really not that hard, if you need a highly accurate print use one of the professional printers that you’ve used in the past.

Sub $5k machine does not equal $30k plus machine. There is a pretty large price gap here that generates a much larger software development group for the “professional” machines. 99% of the time the Formlabs machine has print quality easily as good as the more expensive “pro” machines.

I’ve owned Formlabs machines from the Form 1+ release on. I’ve never had an issue where a mesh crashed Preform. I think I might have seen this issue once before on the forums so it’s not a big percentage issue to fix. There are plenty of other issues and features that should be put in before this.

There is no excuse for crummy software when their Preform hardware was very well designed. I absolutely agree the case of buying hardware to fund their software development, but not when they already sold thousands of these machines by now, you can actually tell their markup unit cost is higher than most desktop printers out there.

Also, they chose to develop in Mac over PC was entirely their choice (not Autodesk) and when asked about giving a Windows 64 a shot, they blankly refuse to answer or even studying it or even allow me as a customer for a direct contact.

Also, these slicing software have been around in the market for a very long time (and very reliable too). The first ever enthusiast 3D machines I’ve heard was 10 years ago, before that it was the LOM but it gone since the late 90s.


So no, the idea to develop a reliable software isn’t solely about machine cost to fund the operation, but rather spending time with the problem at hand and from what my distributor have said Formlab are too busy developing their SLS first before refining their own SLA software.

Also, let me tell you something, professional consumables materials tend to be much more expensive than in the enthusiast market (FDM = 8-10x or SLA = 2-4x) . e.g. FDM 1kg filament usually cost $350 or 10L SLA Watershed = $2.5k, they charge this much not because to fund their development, because they can due to lacking in competition, why do you think Stratasys & 3D Systems for the past 10 years have been using lawsuit against smaller company or buy them up?

At one point Formlab have to pay proceeds to these anti-competitive corporations. In some enthusiast machines they call FFF instead of FDM for this very reason. Or the fact that 10 years ago the explosion of enthusiast machines was no accident, because this was the time when the patent expire for 3d system & Stratasys as they had been in the market 30 years prior.

Also, professional slicing software that I use can actually date back from the early 2000 with few superficial updates.

Unlike OS coding, they don’t normally specify number lines of codes to their software, but below should give you a basic understanding.

Preform still have someway to go

I have really lost you now.

What software are you speaking of here? Preform? NetFabb? MeshMixer?

What is this supposed to show? I mean I get it, is the property sheet for various app installers, but what are we supposed to look for?

Again, most have not had this issue.

We tend to become pin point focused in our industries. I notice things in cars that 99.5% of people will not. Almost 30 years in the industry has allowed me to be very focused on exactly what I do. The same goes for moldmakers and design/style weenies. They will see the resolution you’re talking about but to the average person it’s simply not a big deal. If it’s a concern simply charge them for the expensive prints.

The big printer companies defend their patents to protect their investments and development costs. It’s pretty easy to just steal another company’s design and go to market without those protections. Formlabs has patents for exactly this reason.

Margins are dropping for 3D systems and Stratasys. It’s not 2013 anymore and the stock reflects their financial performance. 3D systems is down around 25% last quarter on margin. Ouch…

The RepRap project had huge resources poured into both the firmware and software. It’s open source so you can’t see or track the hours countless people volunteered to get that thing running. It’s a much different business model when compared to a normal “for profit” business.

Historically mould makers and automotive industries uses surface modelling oppose to solid modelling programs such as Catia for obvious reasons, as I’ve touched on this myself (lightly). But when it comes to functioning / engineering side they used solid modelling at it core.

Parametric Solid > Mesh will always be an issue, so I’m not sure where you get the 99.5% considering there are whole sector of engineers who use Solid programs at it core (maybe including Catia parametric surfaces), but definitely not Mesh STL.

Ask any engineers and they will tell you STL should only reserved for 3d printing industry and nothing else.

Even in FEA, CFD analysis like Nastran, Ansys (even the build in ones in SW & Proe) they use dynamic resolution control and not import as STL to generate polygon calculations.

By default it is visually unacceptable, if I hadn’t increase the resolution settings. There are situation where I had to use 3d prototyping services (for engineering) and I can tell you that if they printed at default resolutions (usually less than 0.5 million polygons) then I will definitely see facets / scan line as I’ve demonstrated for these battery covers and I will be in a position to reject those samples considering most of them charge each model around $100 - $800 (depending on build volume, model complexity & other visual requirements).

So no, I will not accept default resolution as the norm considering I’ve already paid a lot for consumables, because 3d printing as a whole is actually a very simple idea. In fact the core functions is no different to those SLA machines back in the early 80s.

The best way to describe Stratasys & 3D system = Kodak cameras, high cost in consumables, unwilling to develop new technology (using film instead of digital), almost everything is lockdown rather than open source. So I have no issue to see these 2 shrink, because I’ve already lost faith with them at least 15 years ago.

Most of the RepRap settings and functions actually borrowed from Stratasys FDM 2000. The major differences (beside it size) is that RepRap is open vented to allow for much cheaper construction, whereas Statasys machines are enclosed heated chamber., meaning by default Stratasys machines require less tweaking with more reliable prints. I’ve noticed enthusiast market headed into this direction as well e.g. Z18, 3dgence, Cubicon, Sindoh, Vshaper.

In fact look at the CatalytEX you will notice there is hardly any tweaks when committing to a print, whereas RepRap software had to over complicate itself in order to make it work e.g. feed speed, temp, bed speed, shell & etc.

Open source is a good place to start (for any kickstarter machines), but not that good when it comes to refining it when having new hardware. Which is why Formlabs turn away from it which is partly sponsored by Autodesk Netfabb, because AutoDesk told me they wanted to unified most of the enthusiast machines under one software (go to > Machine library). But Netfabb become a red-herring since the departure of Amber.

I’m quite happy to buy Netfabb software, as it has the feature that I need, but not under the basis of a legacy software (without a chance of update) and slap on an annual subscription fee rather than a fixed pricing in which Autodesk said they decide not to develop Amber any further (you can probably still find it somewhere). In another word, Autodesk is using their software development to try to cash-in from the enthusiast market (very disingenuous).

What annoying is that the newer so called professional printers like Desktop Metal now copy the functions as found in enthusiast FDM software such as RepRap.

Anyway, pouring money can turn them into a controlling giant without any new innovation. For the 15-20 years working with professional machines, I haven’t seen any real innovation or development.

That s why I bought the Form2 as a secondary, like I said Form2 including Preform is heading in the right direction, but it software require refinement to fullfill those people who use Solid modelling / engineers as we (I) notices it will hit a brick wall when doing anything with details in which the Form2 is capable off.

I even forgo the traditional right-way up SLA printing technique before considering this machine. Eventhough it introduces more issues when printing under this technique.

However, they seem to think it OK (for now) not to develop continuous SLA for Form3 considering other machines are already out there such as continuous DLP (e.g. Carbon3D, Nexa NX1). Formlab even tries to rebut this claim by releasing this document.

Suggesting DLP pixel size is larger than their own UV point diameter (superior at x,Y resolutions), but you can actually get a 4k panel for $40 of this size.

Finally, Formlab is probably the most profitable enthusiast 3D printers company out there with world wide sales network for almost 8 years now. So yeah Formlab is a business.

I agree that’s what we use our Formlabs printers for.