Daniel Huffman, a cartographer, writes about map design, projections, shaded relief techniques, and the history of mapmaking.
Commenters find his story compelling and inspiring, and HN links it for map-making techniques and his thoughts on elegance versus accuracy.
Sure.
For the Moon the (images) data seems to be bucketed with regard to pixels-per-degree (PPD), the more coarse data set being 64ppd data (finer resolution for the Moon goes at least as high as 512ppd that I have found). Sadly, my workflow (image editors, and Blender as you mentioned) are already straining with the 64ppd data-set so I did not try to go higher resolution. It means though that my Moon is wildly small — the lunar bases that seem to be a paltry 7x7 or 9x9 voxels are in fact the size of a small town.
I began with NASA's "CGI Moon Kit" [1] which, as I say, is 64ppd. There is both a "color data" image and an "elevation data" image (where the greyscale value of the image represents elevation). Polar data was garbage (and would look wildly distorted in my tile-based voxel algorithm so I confess that I tossed it — cropped everything outside 60-degrees latitude basically - I just limit fuel in the game to keep pesky kids from trying to get too deep into the polar regions).
Even at 64ppd and poles cropped, I still ended up with two images that were 23040 × 11264 pixels in size. This turned out to be a challenge for most of the image editing apps I have. Blender handled it (but barely).
With both color + elevation images I followed an online tutorial [2] meant for cartographers wanting to use Blender for relief shading. NASA's moon map had no shadows (probably a good thing). I was not going to try to real-time shade/light the voxel terrain so I settled on "burning…
I think there is something important that we lose when we increase the accuracy of a representation at the expense of the elegance of the representation. Illustrations to photographs is a great example here, but another one is in maps.
I don't know if anyone here ever used this, but there used to be a system called LineDrive [1] that would give you "napkin" directions instead of overlaying the directions on a map. It used to be on Microsoft's MapPoint but apparently decomissioned [2].
This was most useful in the time when you would print out directions instead of getting them dynamically, but I feel like there's still a place for it -- I don't need to necessarily know the details of every twist and turn in the road so much as a schematic overview, even when getting turn-by-turn directions. It would be great to see this revived in some way, because I think the representative maps are much more clear for most use cases.
[1] http://vis.berkeley.edu/papers/rtmaps/
[2] https://somethingaboutmaps.wordpress.com/2011/03/08/remembering-linedrive/
This reminds me of a Bing (MSN?) maps plug-in a while ago that was built by, I think, an Indian guy who went over to MS, which would plot a summarized route map from point A to point B where the main roads in your route were labeled, esp where you had to turn or get on a diff road. This was before phones with GPS came about.
The roads looked somehwat like chicken scratch (straight, not many curves). Like if you were mapping directions in sand/dirt with a stick in explaining to someone how to get some place.
Ah, yes, it was Linedrive: https://somethingaboutmaps.wordpress.com/2011/03/08/remembering-linedrive/
Reminds me of LineDrive [1], a feature of MapBlast that would give you directions in something that looked like how a human would write out directions, where roads were simplified. The overall result is reducing a map to only the parts that are important to someone that needs directions along a specific route.
A bit of a moot point in today's world where everyone has a GPS enabled device capable of giving them turn-by-turn directions interactively.
[1] https://somethingaboutmaps.wordpress.com/2011/03/08/remembering-linedrive/
I don't get it why ML is needed. Just download a DEM (for example, an aerial LIDAR map or SRTM), import it in QGIS along with your base map raster, choose the appropriate shading parameters and combine the two with appropriate blending.
edit: and if you're feeling fancy, you can render the relief shading in blender: https://somethingaboutmaps.wordpress.com/2017/11/16/creating-shaded-relief-in-blender/
The author Daniel Huffman's story is quite compelling and inspiring:
https://somethingaboutmaps.wordpress.com/about/
The long version is even better:
https://somethingaboutmaps.wordpress.com/2011/03/02/on-salvation/