How We Test MakerBot Print Quality: MakerWare

Posted by on Monday, March 25, 2013 in Uncategorized

torture test group_700x467

MakerBot MakerWare, our 3D printing software package, is an integral part of the MakerBot 3D Ecosystem. Powerful software helps us take the most advantage of our powerful desktop 3D printing and scanning products, as well as developments in our ABS and PLA filament. In other words, MakerWare is a huge factor in print quality.

Our goal is to constantly improve print quality, so how do we make sure that we are constantly improving MakerWare? Above is a picture of one print from our multi-stage testing process. We thought you’d like to know more about how we measure our progress and keep moving forward.


We release updates to both the MakerBot MakerWare application and the MakerBot Firmware on a regular basis. These are separate, but they have to work well together. MakerWare operates on your computer, while the Firmware operates on your 3D printer. Not every new release of MakerWare includes a Firmware update, but most major releases do.



MakerWare contains an algorithm that turns a 3D model into a set of instructions for your MakerBot. We call this algorithm the MakerBot Slicing Engine. It cuts a 3D model into “slices” or layers that your MakerBot can print, one on top of the other.

The accuracy and smooth operation of the Slicing Engine are a huge priority for us. Our engineers work to make it capture smaller and smaller details of a model more precisely. To test new developments in the Slicing Engine, we devised a comprehensive demo print, seen above. We make a lot of them, in both ABS and PLA filament, and on the Replicator, the Replicator 2, and the Replicator 2X. Check it out.


1.  The walls of the spiral are just far enough apart that they can’t be filled with internal spurs (see #3 on this list). They have to be filled in with lots of tiny lines, and generating this infill can slow down the Slicing Engine. We’re speeding that up.

2.  These blocks test dimensional accuracy. A block of a certain size in the 3D model should be sliced and printed as a block of that same size.

3.  This is a version number for the design of this test print, but really shows the performance of internal spurs. Internal spurs are single lines of extrusion used to fill the gap between outlines when that gap narrows to a single extrusion width. Here we can see internal spurs are turned off, because the numerals are printed as outlines, rather than a single wall.

4.  In the 3D model, this tower gets progressively narrower and goes all the way up. This element tests what the smallest features are on a 3D model that the Slicing Engine will capture and turn into print instructions. Here, the tower stops at approximately half the height of the tower in the model and we can measure it with calipers.

5.  This empty spot that looks sort of like a lake in the middle of the print, shows us what happens when a single layer contains solid model surface bordered by infill. The effect is hard to see here, but basically our engineers are looking for a flat surface.

6.  These elements test bridging, or extruding filament in mid-air with no support material. They’re designed in different directions and lengths to demonstrate that the Slicing Engine has properly coded the direction of the filament extrusion. It’s kinda complicated.

7.  These walls demonstrate external spurs. With the appropriate settings, the Slicing Engine will see features this small and interpret them as a single extrusion width.

8.  The cylinders test how the Slicing Engine is capturing curves. We want to ensure that a circle in a 3D model is interpreted as a circle of the same dimension by the Slicing Engine. We know the size and shape in the model, so we can test the final result with calipers. We want it to be perfect, and we’re getting there.

9.  We make a lot of these test prints. This little gap makes it easy to get the piece off the build plate.

This is our torture test for the slicer. Do you have your own? Do you have a torture test for the mechanics of your MakerBot? Share it with us, and tag it “torture test” for everyone else to find easily. You can see our model here on Thingiverse.



The so called slicing “torture test” is just one part of our testing regimen. We also look at several dual extrusion prints to monitor layer transitions and clean color separations. Here are a couple examples.

dual extrusion_700x459

The blocks between the dual extrusion prints are part of an overall print quality test. They represent each of the standard quality settings in MakerWare (Low, Medium, High) and show us dimensional accuracy, layer height resolution, surface quality, and several other criteria. We keep these in an archive to compare the print quality from each new release with previous versions.

nut and bolt_700x467

Rafts and support material are other huge software challenges we are conquering, so when we test new MakerWare and Firmware updates, we test these elements, too. Take a look above at this nut and bolt set, designed by Thingiverse star aubenc. This model doesn’t actually require a raft, and you can see below that the top surface is much nicer when printed without a raft (left) than when printed with a raft (right). That’s something we’re fixing.

bolt surface_700x467

Here is just one of the 3D models we use to examine support material behavior, the Stanford Bunny, scanned by Marc Levoy for the Stanford 3D Scanning Repository.


We make this upside down with support material turned on, and then test for ease of removing the support material. On the left you can see the model printed with support material, and the middle what it looks like with the support material removed. There are some surface imperfections that you don’t see when the model is printed right-side-up without support material, as designed.


This is a window into how we test to make sure we are always improving. Do you have suggestions or feature requests? We are always interested to hear feedback from the amazing group of people that make up the MakerBot community. Email [email protected]


Tagged with , 13 comments

13 Comments so far

  • Emmett
    March 25, 2013 at 3:04 pm

    Pretty good, but I would recommend more non-rectilinear features. Spurs are easy when they are exactly a constant width, but if you difference a large sphere out of a wall so that it almost makes a hole, you’ll run into a variety of narrowing segments that show all kinds of problems. Bridges that start as a knife-edge are also tricky, but would work just fine with the right algorithm. Multiple bridges in different directions on the same layer is also important.

  • AJ Sherer
    March 25, 2013 at 5:22 pm

    How about showing some sense of scale in your photos? I have a feeling you guys printed these huge…

  • Andrew
    March 26, 2013 at 9:38 am

    @Emmett, thanks for the feedback. You should share your own Slicing Engine test!
    @AJ, great idea. The nut and bolt set are larger than normal, you’re right, but the other pieces aren’t large. For future photos, we’ll toss a coin or something in there. Thanks!

  • Brad
    March 27, 2013 at 4:45 pm

    Hey guys. Just wondering how you’re coming with getting supports compatible with dual extrusion?

  • Jamie Laing
    March 28, 2013 at 4:26 pm

    This is great, thanks so much for sharing this sort of thing.

    I’d like to second Brad’s question… it would be so great to have support material automatically come out from a secondary extruder, for example one with water soluable PVA on it. It would be great to take a complicated model and dunk it in water to remove the support material!

  • Jarod
    March 28, 2013 at 11:02 pm

    So why dont you have these test prints in your retail store. It would be nice to see what your machine can do. Or is your print quality a secret?

  • Mark
    April 1, 2013 at 10:33 pm

    When doing a test print. Why not print something that can be used. Or part of something larger and can be made and given away. Something that we all need. Something to learn with, a learning add of some type that can be give to local schools. A ruler. Or a number of smaller tools that could be given to the RED Cross to help those in times of need.

    As we start the real interactive age. Think of it this way. How many books could have been “test page printed over the last 15 years and given away to schools. Or a review on CPR. OR the steps to the heimlich. Or simple math review of some type. What a wasted paper it was, Lets not let that happen with 3d test printing.

    How about a 3d view of geometry simple laws. Pythagorean theorem any1. This time in 3d. Wow how we could all learn in new ways.

  • Aaron
    April 12, 2013 at 12:49 am

    I like how they put a tilt shift blur on the photo so you think there printing super fine details and perfect bridges. I downloaded the stl for this “makerware test object” on thingiverse, its friggin HUGE!!

  • Tunell
    May 3, 2013 at 2:22 pm

    An issue I’ve noticed that is not addressed in this model is internal ring/shaft dimensions. Like for Vertical bolt shafts. On many prints that are ring shaped, or have other hollow geometry that passes vertically through the entire model, the dimension is ALWAYS undersized.

    This seems to be because on internal hole dimensions do not take into account the width of the toolpath like the oudside dimensions do.

  • mike
    May 16, 2013 at 11:06 am

    Can we get a test cube, sphere, or some object that is like 4cm x 4cm, for calibrating 3D software apps.
    I load up a few of the test objects and they are like 100 foot tall in my 3D app {Lightwave 3D ver 11.5}
    I Load up the stl in Blender, and its huge.. same in Lightwave..
    I want to build, model objects, but i am having problems getting the scaling correct.

    And Thank you


  • [email protected]
    June 13, 2013 at 4:34 pm

    If you ever merge with Stratasys, hope that you will swallow their knowledge on CatalystEX and hopefully improve the slicing (and the whole set of parameters as well…).

    For example, their support of the support (Uprint’s, no joke) feels like a breeze compared to makerbot’s. Shure it is not the same kind of stuff but my makerbot 2x needs to be tweaked for almost every print, if I want to obtain a fair and precise result.
    I am not desapointed by the “X” in the product’s name…

    Achieving good (not fair) print quality is difficult with the makerbot 2x, especially with bottom layers printed on a raft. Trajectories of the tip really also need some improvements: sometimes the space between two filaments looks like a pothole ! And it should’nt.

    Funny to notice that Markerbot’s specs are pretty close of the Uprint’s in theory, it should be easy to reach its print quality someday, no ?

    As a switcher (from Uprint outrageous ABS pricing world to Makerbot’s cheap one), I would like to be able to have access to a good database on printing tips for user, something so detailed that it would be boring to read.

    Keep up the good job and hope to find some smart improvments in Makerware soon.


  • AnimMan
    August 15, 2013 at 2:45 pm

    Lightwave always refers to 1 unit = 1 meter. That’s why you are having scaling issues. When importing you will ether need to scale by .3048 or by 3.28 depending on whether the object is too large or too small.
    It is a simple meter to feet or feet to meters conversion.

  • Andre
    August 17, 2013 at 3:16 pm

    I’m curious if the Makerware software has features to resolve an issue I am having. Downloading STL files from CAD sites ( specifically) the prints come out with walls (shells) super thin, and often with holes or gaps between the layers, even though there’s no indication of problems with the objects within Makerware.

    So I have been playing around with the different settings — travel speed, heat, etc — and still haven’t found a solution, which leads me to here.

    So I’m looking for ways to optimize the printing process at printer’s highest resolution, and I’m not sure if my solution resides within Makerware or there’s just something strange from the STL files from 3dcontentcentral.


Leave your comment


xhtml: You can use these tags: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>


What can we help you with today?
I want to chat with Sales.
I have a question about an existing order.
I have a technical question about my device.
Existing Orders
For faster service, enter your order number
(found in your confirmation e-mail)