Thing-O-Saurus Claw by gavinmurphy

Thingiverse citizen gavinmurphy scanned the cast from a fossilized claw of an Allosaurus from the Peabody Museum with Autodesk’s 123D Catch.

It’s interesting to think about how many generations this plastic replica is removed from the original claw itself.  The first generation would be the original claw from the Allosaurus.  Depending upon the type of fossilization, the original claw would be replaced over time by mineral deposits in the shape of the original claw left in the surrounding material.  This would make the fossil itself either a second or third generation copy. ¹  The fossil was then cast into a mold once it was dug up, making that the third or fourth generation copy.  The mold was used to cast a replica, which would be a fourth or fifth generation copy.  That replica was photographed by gavinmurphy, making the fifth or sixth generation copy.  Those photos were converted into a 3D image by 123D catch, making the sixth or seventh generation copy.  The DWG or OBJ file created by 123D would have to be converted to an STL for printing, making the STL the seventh or eighth generation copy.  The STL was then printed as you’ve seen above, making the eighth or ninth generation copy.  The photo above would then be a ninth or tenth generation copy.

Whether nine or ten generations, this awesome claw is pretty much a must-print item if you’re looking to show off what a 3D printer can do.  (Oh, you want a dinosaur?  How about a life-sized dinosaur claw?!) ²

Untitled by gavinmurphy

This thing brought to you by

1. Second generation if the original claw became a fossil directly.  Third generation if the fossil was created as minerals took the shape of the claw from the surrounding material.  Also a third generation copy if the original claw disintegrated leaving a claw-shaped depression in the material around where it used to be. [↩]
2. Do you remember that bit from Jurassic Park where the kid calls the veloceraptor a six foot turkey?  I now finally have a good reason to memorize Sam Neill’s awesome response. [↩]

The Smiths by Tony Buser

MakerBot’s own Tony Buser has been scanning and sharing his family on Thingiverse with quite possibly the most creative use of the new “multiple derivative” feature yet!  (Each person was scanned separately and then arranged together into the picture above.)  Tony has thoughtfully memorialized his family and friends with these scans.

The possibilities opened by cheap ubiquitous 3D scanning technology are nearly endless.  A 3D printed family portrait, a movie of 3D scans morphing from one month to the next showing how a person grows, replacing the pen marks on the door jamb for marking height, or maybe even a way to easily assemble as seamless family portrait without everyone being in the same location at the same time.¹

1. Also, who will be the first to upload a Tony BuSta? [↩]

Amy Buser Reconstructed

People have been using the Microsoft Kinect with 3D printing for a while now using excellent programs like Kyle McDonald’s KinectToStl.  However, until recently, most programs can only capture one side of an object which creates a kind of relief sculpture.  To get around this limitation, you could take multiple scans and manually merge them. (hard)  Others like the blablabLAB calibrates and positions multiple Kinect sensors around a scene. (expensive) Last year Microsoft demonstrated something called Kinect Fusion that allows you to carry the Kinect around and dynamically capture all angles of a scene in real time.  Unfortunately, they did not release any software.  Profactor has just released a beta version of free software called ReconstructMe that works a lot like Kinect Fusion.

I’ve scanned a number of things so far, check out the reconstructme tag on Thingiverse!  ReconstructMe works a lot like the Polhemus scanner we used to scan Stephen Colbert where you walk around and wave the Kinect across a scene to capture all sides.  Although the resolution is lower, at least you don’t have to dust your hair in corn starch!  As a matter of fact I’ve found the best way to scan a person is to have them sit in an office chair, point the Kinect at their head, and then slowly spin themselves in a circle.  Once you have a raw scan, I suggest using the free version of NetFabb Studio Basic to rotate it, Cut away the parts you don’t want, and then Repair it to make it solid and suitable for 3D printing on your MakerBot.  The Ponoko blog has an excellent video explaining the process.  You can also place objects on a turntable, like a lazy susan and spin it by hand.  Just make sure that anything ReconstructMe sees within it’s scanning area all rotates in the same way.

There are some limitations to ReconstructMe.  It is Windows only.  In order to do real time reconstruction, you need a fairly powerful video card as it does the calculations on the GPU.  There is an offline recording option that allows you to record on a slower computer and then process it later using a more powerful computer.  However, you don’t get the realtime feedback alerting you when you move too fast or go out of alignment.  Due to the low resolution of the Kinect camera, it’s not that great for scanning small things.  For that, you might want to try something like Spinscan.  However, for scanning large objects like people – it’s awesome!  So go download ReconstructMe and be sure to tag anything you make on Thingiverse with the reconstructme tag.

Untitled by tbuser

This thing brought to you by

Untitled by tbuser

This thing brought to you by

Untitled by tbuser

This thing brought to you by

There’s been some excitement recently about a clever (and effective) way of turning objects into 3d models from photographs: Autodesk 123d Catch.  MakerBot Support’s own Brian Stamile has used it to get some very good results, and has even spawned a project idea (in a tweet):

Shopping for fun items to scan at Goodwill. Buy it, scan it, print it on a #MakerBot, donate it back. That’s the plan. Project Scancycle.

Here’s how it works: take photos of your item on a neutral background.  Rotate around the object, snapping a photo every 15 or so degrees, from a few different heights: above, below, from the side.  It’ll take about 40-75 photographs for 123d Catch to create a good 3d model.  It doesn’t work well with objects that are very shiny.  You’ll also want to open the resulting .obj in your favorite 3d modeling program (MeshLab, MeshMixer or Blender, perhaps) to clean it up and export to .stl.

123d Catch is currently a free beta, so try it out now…unfortunately it runs only on Windows at this time.  If you want to participate in Project Scancycle, just tag your Thingiverse item with “Scancycle” and/or tweet it with the hashtag #Scancycle.

Untitled by BrianStamile

This thing brought to you by

The Be Your Own Souvenir campaign continues. Having just wrapped up a spot at Ars Electronica Festival in Linz, Austria, the blablabLAB team is heading to London for Alpha-ville for what should be another great festival. If you’re gonna be in London between the 22nd of 25th of September, you should go check ‘em out, just like the love birds out there in Linz.

Here’s something a bit different: a MakerBot Thing-O-Matic features prominently in Adri Schokker’s installation “Reconstruction of a Catastrophe.”  As you can see in this video, the the Thing-O-Matic is used in conjunction with a novel video interface to create human figures for the installation.

The piece deals with the human connection to disastrous events; it uses the MakerBot-created items to actually place the viewer in the recreated scene.  The piece is not graphic, but the relationship it creates to the piece is an uncomfortable one for the viewer; it’s a bit darker than most of the projects we share here, so use your judgment before showing it to younger people.  That said, this is an innovative and thoughtful piece of art; we’re glad that a MakerBot Thing-O-Matic helped in its creation.

The awesome folks over at blablabLAB just took their show on the road to Romania, where they continued their “Be Your Own Souvenir” project, this time printing the scans on their brand-new Thing-O-Matics!

Check out the video about their project here.

They sent along a few pictures, and I’m struck by the fact that this looks like a very fun event.  More photos after the jump!

Read the rest of this entry »

Gnome and Gnome

There is no doubt Tony Buser has definitely done more for the 3D printing community than anyone else when it comes to advancing gnome duplication and teleportation technology.  However, I’m convinced that his SpinScan open source software and hadware has a larger potential besides assisting in the controversial practice of gnome cloning. ¹  Tony hasn’t finalized the materials list, but the final project would probably involve a decent web camera with good low light performance² , a cheap laser³ , a stepper driver, a stepper motor, an arduino, a few bearings, threaded rod, and some nuts and bolts.  The whole lot would set you back around $200 and significantly less if you can scavenge a few parts.

So, if you could scan and print anything, what would it be?⁴

Spinscan by tbuser

Untitled by tbuser

This thing brought to you by

1. I mean, the anti-gnome-stem-cell lobby is just insane! [↩]
2. Perhaps around $100 [↩]
3. He got a $4 laser from eBay [↩]
4. But, perhaps a better question is…  what are you waiting for?! [↩]

SIGGRAPH 2011 - Portable, super-high-resolution 3-D imaging from MIT

It used to be that creating highly detailed microscopic scans required huge expensive pieces of equipment, vibration isolation tables and hours of processing.  Researchers at MIT have developd a cheap small and portable 3D scanner about the size of a soda can that can detect features as small as 0.0001 mm tall and 0.0002 mm wide – and it can create the 3D images nearly instantaneously.

I cannot wait to plug one of these into my Thing-O-Matic!

Hattip to SlashDo

80 mm tall scan, printed with a Thing-O-Matic

I’ve been working with 3d scanning for over two years now, building my own 3d scanners and writing code to work with commercial scanners like the Kinect. When I started my artist residency at MakerBot, I knew I had to combine 3d scanning and 3d printing. One of the first ideas we had was to set up a 3d photo booth at MakerBot to scan visitors. So I wrote an app called KinectToStl which first launched at the MakerBot NY user group. I modified it a bit based on feedback from that event, and it’s since been fired up for the 3rd Ward Make-A-Thon and Maker Faire Bay Area 2011. Other artists and hackers are exploring the app, like 4nchor5 la6.

I’ve been maintaining a GitHub repository of all the code I write while at MakerBot, and the source is available for download there. If you don’t want to deal with compiling it, there’s also an OSX app available. The project has a few external dependencies that make it slightly difficult to compile right now, primarily because openFrameworks is nearing a release and hasn’t stabilized yet. The source would be most useful for someone interested in recreating this in Processing or another environment.

Once you’ve downloaded the app, plug in your Kinect and start the app, you’ll see three tabs and four sliders.

Screenshot from KinectToStl

The first tab is where the relevant options are:

• zCutoff controls how far away your cutoff plane is, in centimeters. Anything farther than this will disappear into the “background” (flat area).
• fovWidth and fovHeight control the field of view (or angle of view) multiplier in each axis. This acts like a “zoom” function, and lets you zoom into the center of the scan.
• stlSize controls the width of the STL file, in millimeters. I generally stick between 60 and 80 mm.
• useRandomExport is an experimental export option that only works when fovWidth and fovHeight are set to 1. It will attempt to do a more natural triangulation of the model.

Once you’ve framed the subject and have your settings right, click on “exportStl”. This will save an STL file to the /data folder next to the .app. They’re named like screenshots, “Kinect Export 2011-05-25 at 18.21.13.stl” These files are ready to be 3D printed. They’re scaled properly, and sitting flat on the bed. Because they have a high triangle count, they can take anywhere from 10-20 minutes to run through Skeinforge.

The second tab (click in the the second rectangle) is for lighting options if you’re having trouble seeing the live preview. The third tab is for more advanced options. I’ll mention a few:

• temporalBlur is what causes the 3d model to fade over time, and helps with the accuracy of the scan. It’s like the “exposure time” setting on a camera: shorter exposure is noisier, longer is more sensitive to motion but less noisy.
• smoothingAmount controls the level of blurring on the model. This can make a visually more appealing or better skeined model at the expense of loss of detail.
• backOffset controls the size of the backing plane. If you want more layers of backing, turn this up.

I’ve found that the best parameters for printing these models are with 3-4 shells and almost 0 infill. They’ll print quickly like this, and should be fairly structurally sound.

I also recommend playing with the “OSX Kinect Application” posted by CidVilas to Thingiverse, it has a lot more options and a prettier interface.

If you’re interested in scanning an object in a more complete way, I recommend taking multiple STL exports from KinectToStl and doing boolean joins on them after cutting off the backing plane with an intersection operator. If someone tries this out, I’d love to see the results!

All the scans we (myself and MakerBot) have been taking with this app are being posted to the 3dphotobooth Thingiverse account. So if you’ve been to any of the recent events mentioned above, check it out!