[mapguide-internals] Tile Palette RFC changes

[UV]

Hi Jason,

I made a ColorPalette-1.0.0.xsd and a new Mapdefinition-1.1.0.xsd to 
test this..
I had to add the approrpiate extensions into Maestro to be able to edit 
the colorpalette ;)

But after some discussions regarding use cases and usefulness we dropped 
that idea...
In the end the palette defined by the mapdefinition is very fragile and 
the mapping from and back to the colorpalette is not clearly defined.
We discussed to use an attempt to use the Labels of the Layers to 
identify the colors.... or even logical names....
this is not as yet implemented because I have not seen the speces but 
should be easy to do later.

Further, I was thinking this could be used as a kind of preset setting 
for the mapdefinition (make a color scheme,,, and be able to save it.. 
then make another one) but you might as well just copy the 
mapdefinition... Manually duplicating a complex mapstructure might be 
tedious but its probably not worth the effort to implement externally 
visible palettes.

The base colors are applied to GIF images also.

Jason Birch wrote:
> Hi UV,
>
> I noticed that some changes had been introduced into "MapGuide RFC 60 - improvement of color palette quantization for PNG8 tiles"
>
> http://trac.osgeo.org/mapguide/wiki/MapGuideRfc60
>
> It looks like the code changes are now entirely transparent to the end user.  IMHO, keeping this stuff from touching the repository is a good thing.
>
> I do have a couple questions:
>
> - Will this work for any request for an 8-bit image, not just Tile requests?  I would be super-happy to see this done for standard GetMapImage requests; the colour shift between the legend and the rendered image is a large concern not just for tiles.
>
> - Is the palette extracted/stored to the runtime map regardless of the format requested, or is this code path only taken for PNG8 requests?  I'm probably showing my ignorance of the rendering code here :)
>
> If you feel that there is value in a server configuration item to control whether this optimisation is applied at all, and what algorithm is used, then I'm not opposed to that. Given tile cache generation times and resource usage, anything to keep this under control is good.  I could see one setting being used on the tile generation server, and another on the dynamic map rendering server. 
>
> Jason
>
> _______________________________________________
> mapguide-internals mailing list
> mapguide-internals at lists.osgeo.org
> http://lists.osgeo.org/mailman/listinfo/mapguide-internals
>
>