Somewhat less 8-bit hachures

    Newer Older

    The previous version was biased toward diagonals because it didn't take into account that diagonal movement from each point is a longer distance than moving straight up, down, left, or right, and that therefore it shouldn't move that way unless the drop is sqrt(2) as long in that direction. This also comes out a little darker because it is only invalidating a circle around each point instead of a square.

    Better viewed large.

    probable koz, Gem Images, Ross, and 1 other people added this photo to their favorites.

    1. michal migurski 24 months ago | reply

      The confused fuzz on flat land is particularly interesting. Are your particles traveling uphill or down?

    2. t knouff 24 months ago | reply

      The main part of the San Andreas fault stands out so distinctly in these images. These are great. Thanks for sharing them.

    3. Eric Fischer 24 months ago | reply

      They're traveling downhill, but there probably ought to be a threshold that keeps them from drawing if they only go a step or two. I should let them go uphill too and see what that looks like.

    4. Eric Fischer 24 months ago | reply

      It looks pretty much the same if you let them go uphill instead. I think there's just a lot of noise in the elevations.

    5. michal migurski 24 months ago | reply

      Is this SRTM 30m? It has some terrible data in flat areas, just awful. I had to skip it entirely in DEM tools in favor of a crossfade between SRTM 90m and NED 10m.

    6. Eric Fischer 24 months ago | reply

      This is the SRTM1 (1 arc-second/30 meter) data from I think what's going on in flat areas is that it can't tell the difference between buildings and land so you get whatever random thing the radar bounced off of (or nothing when it failed to bounce off of shallow water). Are there even worse problems in other areas?

    7. Eric Fischer 24 months ago | reply

      By the way, the algorithm is:

      Make a list of the coordinates of all points and sort it by elevation
      For each point, starting from the highest elevation:
      ; If it is still valid, plot that point and invalidate it
      ; While there are any valid neighbors:
      ; ; Choose the neighbor with the steepest drop from this one
      ; ; Plot that point and invalidate it
      ; ; Set that point as the current point to choose the next neighbor from
      ; For each point along the path just plotted:
      ; ; Invalidate a radius around the point equal to the square root of its distance along the path

      Indenting with semicolons since that was trendy last week.

    8. michal migurski 24 months ago | reply

      Try NED 10m, the files are large but when downsampled the quality is great.

    keyboard shortcuts: previous photo next photo L view in light box F favorite < scroll film strip left > scroll film strip right ? show all shortcuts