Monday, June 8, 2009

3D hypsographic shaded print

It is interesting how things coincide…

In end of May I was in Wellington teaching “Cartography with ArcGIS” course and thought “I should write a post about that 3D print I did...” And several days later there is a post about 3D prints on Free Geography Tools blog.

Back to the 3D print I did. The print was done quite a while ago, for a NZ ESRI User Group Conferecne  in October-November of 2006. About one month before the Conference I got an e-mail with brochure about 3D printers from Contex. One of the vendors at the conference has these printers but they don’t offer print services. Eagle Technology is the main sponsor of the conference so was in contact with vendors. Luckily, there is another company in Auckland that has 3D printer and it is interested in collaboration. That company is 3D Print. When we met they showed me promotional GIS/mapping model they have. It was a standard size (30x30cm) tile with a city in flat area with satellite image draped on top. It wasn’t impressive since tallest feature on print was about 2cm tall. I have suggested that we put something together – I can provide data/model and they would print it. So we started working on this project.

I have access to topographic data for whole New Zealand including the 20m contours lines.  The question was what area to choose. One of immediate candidates was Wellington area for several reasons: conference is held in Wellington and terrain has interesting features. It didn’t take me long to create a DEM and hillshade from available data. Then I started experimenting with the colours. I have experimented a bit with elevation colour ramps (like described in this post). Here are some test combinations applied on top of Swiss hillshading. There are subtle differences visible in highest mountain region in southeast.

Here is one of the first tries I liked the best.
This is looking promising and I though it would be nice if I could get bathymetry data for Cook Straight. Luckily, National institute of Water & Atmospheric Research (NIWA) was very interested and I got the shapefile data for Cook Straight. And it was really exciting to work with the data and see the results. Cook Straight has very interesting features as you can see. From supplied data I created the DEM as displayed below.
Next step is to combine two datasets. The result is shown below.
Now it is back to the colour schemas. NIWA has a colour ramp they are using but I didn’t like it that much.
And here is why. On its own I wouldn’t mind using it but it in my opinion it does not work well with elevation tint.
Sea bed with logo 
Here are some test screenshots.
After some testing I have decided to go with blue colour ramps and here are most likely candidates.
small land and sea _ blue
small land and sea _ blue 3

With colours almost sorted there was a next challenge – what data format will work the best for 3D print? According to printer’s manual it will work with VRML, 3DS and DXF/DWG data. I had a bit of problems with exporting data as VRML (I was using ArcGIS 9.2 pre-release then). Once exported VRML didn’t really give us the result we were expecting. There was lot of trouble reading the file and getting the texture (picture correctly positioned on top of model). In next try, I made a 3DS file and send it to 3D Print. Few days later I have also sent a DXF file and the image file of the final colour separately so they could experiment and choose the best option.

Eventually, the figured out the best data format and the way to print this (I am not too sure what format was used in the end) and they made first test 3D print from supplied data and this was the result.
Looking at this we have agreed that this is nowhere near to what we have expected. To make it better 3D Print made it bigger, to cover 4 tiles and to have a base and side walls. Each tile is 30x30cm and height depends on the features of terrain. The result was much better. The model was in our Auckland office and then taken to Wellington by one of our staff members about two weeks before the conference. Unfortunately, two of four tiles got broken in transport because the box with print was checked in at the airport. And that was under two week before the Conference!

Then the good guys at 3D Print decided to do more tweaking and make another run/copy of the model. The main reason for tweaking were the colours. All colours came out lighter then everybody has expected or liked. To my surprise, new model looked so much better than the first one.

Here are the photos of the second model taken with my Sony DSC P7 camera (old and not very good).
As you can see even new printed tiles broke but this is not visible when put together.
For the print we have included logos of all parties involved. Can you spot the spike under letter A in EAGLE?
There are some nice features of the terrain visible even if I use just two tiles.
Here are some close up shots to illustrate terrain and undersea features.

On the photo above you can see ‘terraces’ result of undersea configuration – large areas with very small slope. The same effect is noted on Free Geography Tools blog.

For this print we decided not to use solid base in order to save time and material for printing. Here is an underside of one tile.

Lessons learned
This was an interesting experience for several reasons. Firstly, working with new printing technology. Using 3D prints can really bring out the characteristics of the terrain (in this case). These are especially attractive for students and for people who can’t easily create mental picture of an area looking at the topographic map. 3D print as a company was very keen to work with us and demonstrate use of this technology in GIS and mapping.

Choosing symbology was quite difficult considering I had to combine bathymetry and terrain in single product. This would not be that difficult if it was a medium with known characteristics, like paper. In the first try colours used looked fine on screen and test print on paper but when used in 3D print they came out pale. To highlight the coastline I used orange colour that was too strong any line symbol was too wide. Lakes and swamp outlines were also too strong and too wide.

Printing in 3D is not cheap. One tile 30x30cm costs around $300 NZ. All together this printing costs were around $3,000 NZ. Once printed you should take care of the prints since there are quite fragile.
If I get the opportunity to do this again I would be delighted and hopefully the process would be less troublesome.

No comments:

Post a Comment