[Qgis-user] Qgis-user Digest, Vol 77, Issue 41

[watercress at gmx.de]

Am 12.07.2012 12:25, schrieb qgis-user-request at lists.osgeo.org:
> Message: 7
> Date: Thu, 12 Jul 2012 12:11:53 +0200
> From: Tim Sutton <lists at linfiniti.com>
> To: Bernhard Str?bl <bernhard.stroebl at jena.de>
> Cc: qgis-user at lists.osgeo.org
> Subject: Re: [Qgis-user] Which android tablet
> Message-ID:
> 	<CALCNqkakHUNK3DWc2=MpjpTWDGOZ=B84GWeVpsqh_83nTq5p_A at mail.gmail.com>
> Content-Type: text/plain; charset=ISO-8859-1
>
> Hi
>
> On Thu, Jul 12, 2012 at 12:01 PM, Bernhard Str?bl
> <bernhard.stroebl at jena.de> wrote:
>>
>> Am 12.07.2012 11:48, schrieb watercress at gmx.de:
>>
>>> Hi,
>>>
>>> has anyone tested the performance of qgis on android.
>>>
>>> I want to collect and store the position of 50 trouts tagged with radio
>>> transmitters directly in the field.
>>>
>>>
>>> What I did so far:
>>> I have to aim at the targets from different locations at the river bank.
>>> What I do now is simply draw lines on printed orthophotos. The fish is
>>> located where the lines intersect. It is always quite a mess to
>>> determine your own position. I could save the origin with a gps-device
>>> but this takes too much time, would include "postprocessing" and adding
>>> another step will increase the source of error.
>>>
>>> ... and it is not possible to walk into the stream ...
>>>
>>>
>>> I am thinking of using a tablet, locally store the orthophotos (5 to 8
>>> pieces, format: .jp2, each around 130 MB) and save the points (e.g.
>>> postgis). All metadata could be entered at once. No paper work. I know
>>> where I am and can easily store, edit,.. fish data in the field.
>>>
>>> But I am bit worried about the performance. I takes about two til four
>>> minutes to load those raster-layers at my desktop (Ubuntu 12.04, 16GB
>>> Ram, QuadCore,...). The performance appears very poor.
>>
>> This is weird. I am not a specialist with raster formats but one of my
>> aerial photos (geotif) 281 MB loads and displays within seconds, this is
>> with pyramids (maybe you need to build these?)
>> I even put some 50 of these photos into a GDAL virtual raster (VRT), it
>> takes some 20 seconds to load but displays almost immediately.
>> My machine is similar to yours hardwarewise (less RAM, though) running
>> OpenSUSE 64 bit.
>>
> I think the jp2 drivers are quite slow. A nice solution is to load
> your rasters into an mbtiles sqlite database - its not the best for
> size, but its very good for performance.
>
> Note it needs a recent (unreleased?) version of GDAL.. Otherwise
> convert your data to tiffs as Bernhard has done and optimise them for
> performace by e.g. creating pyramids.
>
> Regards

Hi Tim and Bernhard,

thanks a lot. the displaying performance highly increased after
converting the raster set to geotif.

Regards,

Jens