[GRASS-user] New modules in add-ons: v.in.geoplot, r.xtent + r.burn.frict

[Benjamin Ducke]

Hi Hamish

by "boundaries" I actually meant physical entities that
act as movement blockers in a model, such as broad rivers,
chasms, etc., not the GRASS geometry type "boundary".

r.grow adds too many cells for my purpose
v.to.rast gives me stepped lines, which is why I wrote
the little mapcalc filter.

A better solution would probably be to add proper
anti-aliasing to the API, G_bresenham_line_AA() or
whatever and then add an AA option to v.to.rast.
But I have no time to spend on that at the
moment, especially since I wouldn't know if there are
any applications beyond what I needed it for.

If there is general interest in anti-aliased rasterization,
it would be worth a thought. We would need to 
discuss how to do it properly in a GIS context, though.

Ben

----- Original Message -----
From: "Hamish" <hamish_b at yahoo.com>
To: "grass-user" <grass-user at lists.osgeo.org>, "Benjamin Ducke" <benjamin.ducke at oxfordarch.co.uk>
Sent: Tuesday, September 29, 2009 2:49:17 AM GMT +01:00 Amsterdam / Berlin / Bern / Rome / Stockholm / Vienna
Subject: Re: [GRASS-user] New modules in add-ons: v.in.geoplot, r.xtent + r.burn.frict

Ben wrote:
> r.carve actually "carves" rivers into a DEM, subtracting
> their depths from the DEM.
> 
> r.burn.frict does essentially the same job as v.to.rast
> would, except it performs some anti-aliasing on line edges,
> so there is no case like this:
> 
>    0100
>    0010
>    0001
> 
> where a cost surface algorithm that scans for 8 directions
> could "slip" diagonally through the cells that represent a
> boundary line.
> Instead, the line would be rasterized like this:
> 
>    0100
>    0110
>    0011
> 
> This is (only?) useful for rasterizing lines onto friction
> surfaces. It's really just a primitive little script that I
> wrote because I needed it for territorial modeling with
> r.xtent which involves cost surfaces.


if a module does its job well, then great. congrats. but FYI a
couple other ways to accomplish the same thing if it helps.
(not sure it will, once done r.mapcalc is pretty hard to beat)
- r.grow[.distance]
- v.type boundary,line or v.cat add to boundaries -> v.to.rast
  (see man page)
- not G_bresenham_line() but ...?


Hamish







------
Files attached to this email may be in ISO 26300 format (OASIS Open Document Format). If you have difficulty opening them, please visit http://iso26300.info for more information.