[udig-devel] tif render error.

[Jody Garnett]

Vince Darley wrote:
>> It looks to me like you are mixing two things: GeoTIFF (where the 
>> details are encoded in the tiff file) and world plus image; where the 
>> details are defined in files with the same base filename as the tiff.
> Hmm.  How can I tell whether the tiff files have the geographic 
> details encoded?  My colleague who was looking at this told me that 
> nothing at all worked until he got hold of the tfw/tab files from OS.
That could be the case; raster support does not get really good until 
the 1.2 series (which has not seen a public release). Personally I use 
gdalinfo (a command line tool) to check what metadata is known for 
geotiff files.
>> Can you make yourself a ".prj" file using the Well Known Text 
>> available on the advanced tab of the CoordinateReferneceSystem dialog 
>> in uDig?
> Can you explain more precisely what I have to do, and then I expect I 
> can do that.
For a file "foo.tiff" and "foo.tfw" you need a seperate file "foo.prj" 
that contains the well-known-text definition of the projection you are 
using.

Here is the page in the FAQ that has an example:
- 
http://udig.refractions.net/confluence/display/UDIG/How+do+I+open+an+image+in+uDig

>>> Unfortunately when I import one of these images and add it to my 
>>> map, it isn't placed anywhere near the other data, because it seems 
>>> to be using coordinates (British National Grid?) in the range 
>>> 575000, 305000.
>> I see that number (575000) in both your tfw file and your definition 
>> table? Are you defining the range of three axis? Or only two ...
> I'm not really sure.  These are all just standard OS data, and I'm 
> just trying to use it.
I answered my own question:
5.000000000000    (size of pixel in x direction)
0.000000000000    (rotation term for row)
0.000000000000    (rotation term for column)
-5.000000000000    (size of pixel in y direction)
492169.690845528910    (x coordinate of centre of upper left pixel in 
map units)
5426523.318065105000    (y coordinate of centre of upper left pixel in 
map units)
>>> On a related note, there are about a thousand of these tiles.  How 
>>> can I add them to the map in one go (and have them shown in the uDig 
>>> user interface as one entity) rather than as a thousand separate 
>>> layers in the map.
>> On trunk I have support for "Image Moasic" basically a shapefile with 
>> each record indicating what "tile" goes there .. that is the "right" 
>> way to do things; and I am afraid it is not available in 1.1.x.
> Is it worth me testing out trunk?  Or is that just going to hit a 
> bunch of other issues?  (is there a trunk sdk available for testing 
> with?)
Vince trunk is only for the brave right now; how much time do you have? 
I really enjoy trunk and am working there day in and day out; in fact 
after uDig 1.1.0 goes out I don't expect to work on it again execept for 
bug fixes.  That said there is a lot of work left to do on trunk ... 
until a couple of "show stoppers" are fixed it has not been worth us 
making an SDK.

There are direction here for building from source code:
- http://udig.refractions.net/confluence/display/ADMIN/Home

The main reason to experiment with trunk now is if you need access to 
some of the exclusive raster formats like MRSID; or that image moasic 
above. You could also backport the image moasic work (my notes are here 
http://udig.refractions.net/confluence/display/HACK/ImageMoasic).

Jody