I've seen criticism of Beck's famous Tube map as geographically misleading, but it would be hard to be even approximately geographically accurate without eating up a LOT of space.
As to his map being the first of its sort of schematic map, I think that one can test that hypothesis by looking at maps of urban-rail systems of various cities over the years. Someone might have already done that, but I'd have to do a lot of searching to find out.
In any case, I found someone's work on automatic drawing of rail-network schematic maps, Martin Nöllenburg's publications; his master's thesis was "Automated drawing of metro maps". However, he had not published his source code, though I think that it might be possible to develop a passable imitation of it.
It just so happens that Damien Merrick, Tim Dwyer and I have a paper being presented at ISVC in dec on metromap layout. It is a much simpler algorithm than Nöllenburg's, but which produces compelling layouts. There is an implementation in 2geom.org in src/2geom/toys/fitter
The source seems to have bitrotted slightly, I'll look at it on the weekend. It is much much faster (real time vs hours), though given the time to make a new metro line, performance may not matter ;) It may end up in inkscape soon.
I've seen criticism of Beck's famous Tube map as geographically misleading, but it would be hard to be even approximately geographically accurate without eating up a LOT of space.
ReplyDeleteAs to his map being the first of its sort of schematic map, I think that one can test that hypothesis by looking at maps of urban-rail systems of various cities over the years. Someone might have already done that, but I'd have to do a lot of searching to find out.
In any case, I found someone's work on automatic drawing of rail-network schematic maps, Martin Nöllenburg's publications; his master's thesis was "Automated drawing of metro maps". However, he had not published his source code, though I think that it might be possible to develop a passable imitation of it.
It just so happens that Damien Merrick, Tim Dwyer and I have a paper being presented at ISVC in dec on metromap layout. It is a much simpler algorithm than Nöllenburg's, but which produces compelling layouts. There is an implementation in 2geom.org in src/2geom/toys/fitter
ReplyDeleteThe source seems to have bitrotted slightly, I'll look at it on the weekend. It is much much faster (real time vs hours), though given the time to make a new metro line, performance may not matter ;) It may end up in inkscape soon.
Here is the output from the london rail network.