Recently, I’ve been working with several MU Geosciences students to refine the method we’ve developed for 3D printing topographic maps from DEM files. While printing topography isn’t necessarily new in and of itself, we wanted to create a method that we could publish for wide use, and wanted to ensure that the method is cartographically sound as well. To fully refine the method, we’ve been experimenting with various aspects of 3D printing maps. The first, after our proof-of-concept model that included only topography at 1:24,000 scale, was to print a complete map of Tioga County.
Our goal was to create, as I said, a complete and cartographically sound map, with controlled horizontal and vertical scales, a implement a vertical extent that used 3D print slices as contours, while also including the standard map elements (title, legend, scale representation, credits) that differentiate the product as a complete map instead of simply a model.
To do this (using the method we published before) MU student Jonathan Bagg created a model of Tioga County PA at 1:250,000 using contours provided by our data sharing agreement with the Tioga County GIS Department. The contours were interpolated to a DEM raster 11″ (279.4mm) x 11″ in size. The scale of 1:250,000 was used for this project because it was the largest commonly used map scale that would allow the county to comfortably fit into the 11×11″ dimensions. We used a pixel size that allowed the MakerBox Replicator Z18’s maximum print resolution of 0.1mm of filament, creating an output raster of 2794px x 2794px (don’t you love the simplicity of the metric system?).
Further, the raster’s interpolation was simplified to the minimum and maximum values of the contours data, allowing us to more precisely control the raster’s represented values to those even 20-foot values — significant because we wanted to control exactly how the slices produced by the printer would align to the vertical relief of the model. In other words: we wanted to start at the even number of the contours so that each printed slice — layer — would directly correspond to those even numbers in elevation.
An 8-bit output was used for the raster, acceptable because we did not seek a map with a vertical exaggeration where more than 256 values of differentiation were necessary (0.1mm slice resolution x 256 shades of gray = 2.56cm vertical relief in map distance; at 20′ per slice = representation of 5,120′ field vertical relief with a vertical exaggeration of nearly 10x; plus, Tioga County’s total relief is only around 1500′).
In 3DS Max, we created an 11″ (279.4mm) x 11″ plane with 1000 edges and used the displacement modifier to model the topography to a vertical exaggeration of 4.1x, which both allowed the slices to accurately represent the relief while still creating enough variation at this scale to represent the topography as a whole.
To this plane, we added extruded text for a title, scale statement, and credit. We also added a horizontal scale bar, using extruded simple rectangles and text, and a north arrow by extruding a simple geometric shape and the letter N. A vertical scale was created by extruding text labeling elevation at 500′ increments, extruded to the appropriate height for the elevation labeled.
MakerBot Desktop estimated the model, pictured below, would take 44 hours to print and use 514g of filament.
Results & Discussion
Hot off (still on) the build plate, here’s the Tioga County map (click any pic below for a larger version). It took right around 56 hours total to print. There’s definitely some refinement needed on the text, especially at the bottom of the map where the letters were apparently too thin for the printer to properly parse. It also has some print flaws, little bubbles and whatnot on the surface that we’re trouble-shooting.
As a whole, the product was not terrible. Converting my design knowledge from 2D to 3D is taking some practice — I debated the size of the labels endlessly because in 2D, they looked ridiculously large. Unfortunately, I make many of the labels at the bottom too small for this medium, leading to articulation problems from the printer.
The topography, seen below in the oblique view, seemed to mostly work well. It does represent the topography of the county fairly precisely without being too exaggerated, at least at a glance (more on that in a second).
The title text (below) looked okay enough. The text is clear and extruded enough to be legible but not obnoxious.
The text at the bottom was just plain disappointing. It’s ugly and completely unfinished because the lines were too small to properly articulate with the filament, even though they were thicker than the 0.1mm resolution of the printer.
The left side’s text was also disappointing. The scale bar, north arrow and north “N” came out quite well, but the text surrounding it came out really poorly. It’s disappointing, if for no other reason than the filament expenditure (500+g, so around $25) and the time (56 hours!). Simplifying the content of the text will allow us to make the letters larger and “fatter” in the next iteration.
The vertical scale (below), which I thought was clever in its simplicity worked, mostly. The text, like most of the rest of the text, needs to be larger and “fatter” to articulate correctly with the printer — lesson learned. The other issue, though, is that shortly after this photo was taken, the 2500′ marker had already broken off. While the simplicity of the text for the scale is effective in terms of clarity and design, the fragility is a problem that we will have to address.
Another issue, this one unrelated to the text, is found on the topography. Looking closely, you can see that there’s something of a “grid” visible on the features.
I do not think this is an issue of the filament’s resolution or printing ability of the MakerBot, rather an issue with the GIF->3DS Max conversion process. The GIF was sized 2794x2794px to create an 11″ (279.4mm) map at 0.1mm resolution. This was displaced on a surface in 3DS Max with the maximum of 1000 edges. However, obviously, 2794px is not evenly divisible by 1000, leading to this conversion artifact. In the next iteration, we will be changing the raster output to 2793px, use a corresponding map size of 279.3mm, and use 931 edges in 3DS Max (2793/931=3).
Finally, a problem that I do think comes from our printer is the quality of the print. There are definitely some noticeable flaws in the printing, especially here in the bottom right corner, where some “bubbling” and holes are evident.
We are investigating some ways to troubleshoot this issue and are confident that we’ll find a way to avoid this on future prints.
We will be incorporating some of these refinements into our next printing project, which is to print a model of the Pine Creek Gorge in its entirety, taking four tiles to do so. The results of that print will be presented as part of a display to the Tioga County government as a thank you for funding our printer.
The other step that I will be doing, as soon as I have some time, is publishing a refined printing guide that will include the various calculations and steps necessary to ensure a truly representative set of contours produced by the printer’s slices. That will probably take a week or two, but soon!
Finally, with further experimentation, I intend to produce a guide to using text on 3D printed maps — which fonts and styles work, how large they need to be to articulate, how far extruded they should be, etc. A guide like this should prove relatively valuable as this application of 3D printers becomes more commonplace throughout the cartographic community.