Posts Tagged ‘ meshlab ’

How To ‘Print The Bay’

Several people asked me if I’d share the specifics of how I completed my previous print, “Print The Bay“.  Sure!

ptb_front

As mentioned in the post, I used a great (and relatively new tool) by Thatcher Chamberlin called “Terrain2STL” to do the heavy lifting of the map geometry creation:  I had looked at the well documented techniques by Thingiverse user Shapespeare (aka Whitney Potter, as documented on his “Grand Canyon” thing here) and was about to start, but discovered several pieces of the software chain were PC only:  I’m currently on Mac.  So while I started trying to figure out an alternate pipeline, I found a random blog post on Terrain2STL, and everything changed :)  With Terrain2STL, it’s as easy as defining a region in Google Maps and hitting “download” to get a 3d printable stl of the terrain (although as mentioned below, some work is still needed).

After doing several test prints I got in contact with Thatcher, making suggestions as I became more familiar with the tool.  One of which was a way to add “extra thickness” to the ground, and the ability to “lower” the water:  In places like the San Francisco bay, much of the land is at sea level.  When I’d print the map, you couldn’t tell where the shoreline was.  Thatcher was able to implement these new “drop” features, and created a special “Bravo” page for Terrain2STL that exposes them.

Based on all that, here’s the overall process I used:

  • Terrain2STL has a maximum capture volume (the red box it creates):  Looking at the map in the software, at maximum capture size, it would take four regions to grab the area I was after (mid to south SF bay).  It has restrictions though:  The capture volume can’t cross over any major grid lines, so I was also gated by that spec.   Since there’s currently no accurate way to do multiple captures exactly next to one another all at the same time, I’d eyeball each red capture box next to one another, using landmarks on the map (usually street labels) to line one up to the next.  When exporting the data, I set “Base Height” to 4, and “Water Drop” to 2.  For any future ones I do I’d probably set this to 8 & 4, simply to make the printed water a bit thicker.  After download, I had four new stl’s on disk.
  • The stl’s aren’t perfect though:  Thatcher has been working on improving the quality of the data, but I still found a couple issues with them:
    • Holes in the data:  You need water-tight stl’s for printing, and these unfortunately weren’t  (Thatcher mentioned the datasets the tool pulls from has ‘holes’, the tool just emulates what it finds).  Trying to keep my toolchain free/open source, I turned to Meshlab:  It will easily fill in holes.  After bringing the stl into meshlab, access ‘Filters -> Remeshing, Simplification and Reconstruction -> Close Holes’.
    • Wrong aspect ratio:  When creating captured regions, based on what part of the world you’re in, the red capture region will be square (at the equator), or more rectangular the closer to the poles you get.  However, the download stl data is always square.  To fix this, I took a screenshot of one of my capture regions, cropping it to the exact size of the red rectangle.  I then pulled up the resolution info from it (on Mac you can RMB on the file and do a “get info”), divided width by height, and I got my scale value.  To set the scale, in Meshlab access ‘Filters -> Normals, Curvature & Orientation -> Transform: Scale’. Turn off ‘uniform scaling’, and set the X axis scale value to the computed aspect ratio.  Optionally you could scale it in Blender (below).
    • Verts stretching below the ground : It can be hard to notice when zoomed out and looking at the mesh in 3d, but on 3/4 of them there were several verts that were slightly lower than the base plane.  When brought into your slice software and dropped on the build plane, this causes the bulk of the mesh to be one or more layers above the platform, and it won’t stick well at all when printed.  Again, trying to keep the toolchain open source, I turned to Blender: Importing each STL into Blender, I switched to the front and side views, and scanned the bottom edges of the mesh:  Any verts that were too low I simply selected and moved back up above the base layer.  I then re-exported them back out as stl.
  • With the stl’s ready for print, I brought them into my slicer (in my case, Makerbot Desktop) and set them to maximum size.  Since I was going for a translucent effect on the ocean, I did a pre-slice preview to see how many layers thick the ocean was (in this case 2 layers), and set the firmware to pause at that height for a filament change.  I started the print using translucent ‘Neon Blue’ PLA, then swapped to a light gray PLA to finish.  Each region took 2.5-3.5 hours on my Makerbot Replicator 1 (Sailfish firmware).
  • To show the translucency of the ocean I had a custom piece of transparent light blue acrylic cut at TAP Plastics (I live close to one making it easy).  While I was there I picked up two tubes of epoxy that dries clear.  After cleaning the prints and acrylic I slathered a bunch of epoxy on the back of each map region, and pressed them into the acrylic, leaving it to set up for a few hours .
  • I wanted to add additional relief to the mountains.  To do this I shot matte-black spray-paint directly onto the mountain regions, and then quickly wiped it off with a rag soaked in lacquer thinner.  Since this is PLA, there is no reaction with the thinner (ABS would be another story…).

I waited for the sun to rise, and shot the above image.  Hope you find this helpful, and let me know if you have any questions.

Check out the article written on it at 3dPrint.com.

2014
12/18
CATEGORY
art
CG
projects
TAGS






Comments Off on How To ‘Print The Bay’

3D Printing, step one…

Back in March I received my Makerbot Replicator after a two month wait (new product, backordered).  I’ve been trying to get around to post about it, but the (first) Replicator was fraught with multiple problems.  I won’t get into the specifics, but I will say that the Makerbot support personal were very accommodating with helping me resolve the issues, and last week my replacement arrived, fully functional.  The original one I received was a “dual extrusion” model, meaning it could print two alternating colors.  But many of the problems I encountered stemmed from this tech, so I ultimately decided to get the single-extrusion (single color) model.

The Replicator doing it's thing: I covered it in plastic-wrap to help cut down on drafts, that can have an effect on print quality. It's print volume is 8.9" x 5.7" x 5.9", or just under 300 cubic inches.

—-

The first thing I get asked is “what are you going to do with it?”, which is still a question I ask myself.  I think back to two points in my past, and how they changed my future:

The first was in the mid 80’s when my family purchased an Apple ][c.  Up until that decade, computers weren’t popularized in the home, and I feel that was a major turning point for many people:  You survived just fine without one, but once you started using one you couldn’t imagine life without it.  Extrapolate forward to the internet and I think anyone would have a hard time disagreeing with it.

The second was when I finally wrapped my head around ‘Object Oriented Programming‘ while learning Python & Processing:  Up until that point, all my programming had been ‘functional‘.  I’d always heard about this ‘OOP’ stuff, but couldn’t grasp how it would help me.  But just like the old Apple, once I started using it not only I couldn’t imagine programming without it, but I was able to program much more complex systems.

I see 3D printing possibly falling into this category:  I currently don’t envision a wide-spread practical use for it in my daily life, but I have a strong inclination that once I understand the technology and integrate it into my routine, it will find a home among those concepts of the past that have shaped my future.  Or, it will be a very expensive door-stop.

The second thing I get asked is ‘how does it print?’.  There is a variety of 3d printing tech out here, and this style of printer does what’s called ‘extrusion’:  It squeezes a line of 1.8mm ABS filament (ABS is the same plastic Lego’s are made out of) down through a heated nozzle (220 deg C) out of a .4mm tip onto a ‘heated build platform’ (HBP, at 110 deg C).  Before the print started, a 3d model was ‘sliced up’ into individual layers, and the printer will print each layer, one on top of the next by extruding down lines of this filament.  Each layer sticks to the one below it.  When you’re done, you have a physically modeled 3d object in ABS plastic.

Three rolls of different colored 1.8mm ABS filament.

—-

To start learning the system, I’ve been printing designs that came with the printer, and interesting items I’ve found on Thingiverse.  If you want to keep track of the things make\like\design, you can checkout my profile there.  Or, you can just look at this current picture:

Successes include: Big squirrel, tiny squirrel, revolver pencil-holder, 'Minecraft Creeper', 3d printer, iPhone case, whistle (very loud), shark, twisted shot glass (with calibration cube inside).

I’ve also had a lot of failures, mainly based on the two color printing.  Below is an informative pic.  What it does show off however, is the hexagonal ‘infill’ pattern that is automatically generated by the software.  This both provides support for the item, and helps when printing the ‘tops’ of things:

Failures include a robot, a checkered ball, multiple cupcakes, shark, 3d printer.

—-

The next step is to make my own items to print, and there is a lot of software is needed to pull it all together (going in reverse order of usage):

To generate the ‘G-code‘ files the 3d-printer will access to create the print, I use ReplicatorG:  This tool only officially opens stl files, so I have to make sure whatever model-generating solution I use can export to the stl format, or I need mesh conversion software.  ReplicatorG can be used to do simple transformations to the mesh to fit the print volume:  Translate, rotate, scale, and adjusts the print settings:  what percentage of  ‘infill’ the object has (how solid it is), filament feedrate settings, etc.

To deal with mesh file format conversion from the source data to stl for ReplicatorG, I use MeshLab:  It seems very capable in converting one file format to another, and can help ‘clean up’ holes in the mesh.

To modify mesh I didn’t create, I’m starting to use MeshMixer:  Most commonly it’s used to help generate a flat base on the mesh for printing.  There is a good tutorial video on this here.

Finally, you need software that will generate the thing you want to print.  You can design 3d items two main ways:  One is to model it in 3d, the other is to procedurally program the model.  I want to do both actually:

The 3d modeling approach:  I’ve been using Maya (at work) since it was released, and am very comfortable modeling in it.  However, it’s a $5k piece of software, so some other home solution is needed.  The Makerbot page “3D Design Software 101” does a good job of providing an overview of which (affordable) software one can choose, and where to go from there.  I have experience with SketchUp, so that’s the software I’m current investing my modeling mindshare in.

On the procedurally modeled approach, my current plan is to use Processing extended with the libraries HE_mesh and/or toxiclibs:  They are both add-on libraries for Processing to make dealing with mesh a lot easier, and can export to the stl file format.  I’d love to do all this with a Python implementation instead, simply because I like the language more, but it’s hard to beat Processing for its ease of use.  Another option would be OpenSCAD, but I don’t know if I’m up for learning another language right at this moment.

—-

So to date, what have I learned?

  • Large items with a flat base don’t need to have a ‘raft’ printed first.  But if the item is small, or the base is some crazy shape, a raft really helps.
  • Always set the HBP (heated build platform) to 110 deg (Celsius):  ReplicatorG likes to set it to 100 deg by default.  If you don’t, you could get print edge curling.
  • Knock off any hanging filament before the print starts:  It can screw up the initial extrusion and cause it not to stick to the HBP.
  • Print directly from the SD card, not over usb:  I’ve had usb ‘screw up’ a couple times, canceling the print.

—-

What is next?  Modeling in 3d, printing my own designs.  My ultimate goal is to cast these in aluminum via sand-casting (see my previous post on melting metal).  Honestly however, the hardest thing is just figuring out what to make… when you can make anything.

2012
06/17
CATEGORY
art
CG
TAGS






2 comments