[postgis-users] line simplication

[David Blasby]

Mapserver's simplification is effective and easy.  When it loads in spatial
data, it transforms each of your points to an integer (x,y) location on the
resulting Image.  If it finds two or more points in a row that are in the same
pixel, it ignores all but the first.  The resulting picture looks the same as
rendering each of the points in the line/polygon.  This method is really fast
and easy.

A proper line simplification (like D-P) works on the input data in world
coordinates, and is much more costly to compute. You'll probably find that, at
LAN speeds, you can
    QUERY + TRANSMIT + MAPSERVER SIMPLIFY + RENDER
faster than you could
    QUERY + HARD SIMPLIFY + TRANSMIT + MAPSERVER SIMPLIFY + RENDER


Alternatively, you could have two columns (or even two tables) in your database
- one with full resolution data, and one created using the simplification
method.  In mapserver, you use a scale-dependent layer system to show the full
resolution data when you're zooomed in and the lower resolution when you're
zoomed out.

The main drawback is that you're doing your simplification based on
line-segments, not on the entire feature.  For example, if you're rendering a
highway feature that is made up of 50,000 individual segments, you're still
going to be rendering 50,000 segments no matter how much simplification you
do.  Ideally, you'd want to create a highway line and simplify it!

dave