On Thu, Apr 3, 2008 at 1:32 PM, Martin Desruisseaux
<[EMAIL PROTECTED]> wrote:
> Simone Giannecchini a écrit :
>
> > but still there can be some
> > combinations for the RasterSymbolizer elements for which we can get
> > images that have like 7 bands or images with just one band but with
> > negative values.
> >
>
> I see two distinct cases here:
>
> * Images with just one band (with or without negative values) are
>  usually backed by IndexColorModel. As long as the values stand
>  in no more than 16 bits, it can work even if there is negative
>  values (using complement-by-2 integer arithmetic). So in this
>  case it is not clear to me why we don't apply the color palette
>  associated with the image?
>

I am not talking about raster with IndexColorModel, I should have
probably pointed that out.
Typical exmple would be elevation data, which is most part of the time
store in rasters with single band of type short which cannot be
visualized direclty.


> * Images with more than 3 bands can be splitted in two case:
>
>  a) They have a ColorModel (this is possible even if we have 7 bands,
>     provided that the guys who created the image was disciplined enough
>     for building a - maybe custom - ColorModel), in which case we should
>     be able to inspect the ColorModel in order to detect what seems closer
>     to gray, RGB or whatever bands.
>
>  b) They have no ColorModel. This one is an open issue. But I wonder: do
>     we control the processing chain that produced the image, and if so can
>     we revisit it and try to figure out why they don't have a ColorModel,
>     instead than applying arbitrary choices?
>

Same above, if the image has a "nice color model" there are not so
many problems, but for example I have some landsat geotiff coverages
with 7 bands  and it is not straightforward to decide which is the
color model for them. Users should use raster symbolizer to pick out
the RGB or Gray bands via the RasterSymbolizer but if they do not do
that (this seems more like an error to me) we need to decide what to
do (much like the visible band  approach).
It is not clear to me what you mean about the second part of the last
paragraph, but here I am talking about doing some default manipulation
of raster data which does not have intrinsic visualization info
attached in the case where the users give us no clue how they want
them to be rendered or the users gives us erroneous information. In
such a case instead of throwing an error we should try to render
something meaningful, much like openev, qgis or arcmap do.


Simone.


>
>        Martin
>
>



-- 
-------------------------------------------------------
Eng. Simone Giannecchini
President /CEO GeoSolutions S.A.S.
Via Carignoni 51
55041 Camaiore (LU)
Italy

phone: +39 0584983027
fax: +39 0584983027
mob: +39 333 8128928


http://www.geo-solutions.it

-------------------------------------------------------

-------------------------------------------------------------------------
Check out the new SourceForge.net Marketplace.
It's the best place to buy or sell services for
just about anything Open Source.
http://ad.doubleclick.net/clk;164216239;13503038;w?http://sf.net/marketplace
_______________________________________________
Geotools-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/geotools-devel

Reply via email to