[OSM-dev] Filled oceans in tiles at home - request for development work

Dirk-Lüder Kreie osm-list at deelkar.net
Mon Jan 22 17:07:40 GMT 2007

Hash: SHA1

[this should've gone to the list]

Frederik Ramm schrieb:

> I cannot quite explain the gap between the last two numbers, as I've
> used the distributed tile generation during the last few days to re-
> generate about 30.000 tiles which were either outdated or missing,
> and I had hoped to close that gap. I guess that many of the 41852
> tiles that have OSM data might contain data that is not rendered, so
> that a request for rendering one of these tiles results in an empty
> tile.

We need a better way to distinguish empty tiles from ocean tiles,
because they are equally uniform, and thus both very small filesize.
A hack around that would be to give the sea-tiles an artificial
"structure" (waves in the png would be over the top, but would create a
significantly different filesize) instead of just one plain color.

> I'll have to investigate this because otherwise I'll continue to
> request rendering of tiles that I think must be missing, only to find
> that rendering them has no effect!

Currently the tiles at home project doesn't make sure a rendering job (one
level 12 tile with all tiles in it's area down to zoom 17) really gets
rendered and uploaded after it was sent out to a client.

>  From the work done in the previous days I can say that with the
> current participants, about 10k level-12 tiles can be rendered on
> tiles at home in one day.

I'd say even more on average, the queue emptied 6:45 CET (5:45 UTC)
today morning, that's about 10k render jobs in about 14 hours.

however this number fluctuates significantly if the rendered areas
contain a well mapped out urban area vs. only one feature (road,
coastline) in the area.

Dirk-Lüder "Deelkar" Kreie

Version: GnuPG v1.4.6 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org


More information about the dev mailing list