[mapguide-internals] RFC60 - PNG8 color palette improvementspleasecomment

[Jason Birch]

Ah...
 
I misunderstood.  I thought the precomputed values would be the only ones available in the palette.  With this in mind, much of my comments are invalid.  Additional thoughts follow.

> So the remaining use case is a server serving many fairly
> simple maps in different image formats....is that a common
> use case?

I can only speak to my own use case, but I'm going to be serving multiple maps with base layers.  Actually I'm hoping that Fusion 2 will allow me to create dedicated tile-based map definitions, and then include those within other map groups, but I haven't looked into that yet.  Anyway, one of the base layers that I'd want would be orthophotos, which I don't think really lend themselves well to 8bits, so I'd want to have a mix of 24bit and 8bit base layers.  Not sure how critical this is though; I will likely end up serving the orthos as a dynamic layer.

> The color palette is only used as a hint, it does not omit
> the color quantization happening in the tile service, it just
> adds the base colors

Colour me epiphanized :)

More thoughts/questions:

I'm not convinced that there is value in retaining the current method when creating 8bit tiles; it's flawed in all non-trivial use cases.  If the propsed method is effective, I would prefer not to see an additional configuration directive.  Rather, I would like to see it used as the default method for rendering 8bit tiles, with the potential (now or in the future) of overriding the runtime-calculated palette with a manually crafted one on a per-scale-range basis.

Does the propsed process include the background colour from the map definition as well as point/line/polygon colours for each individual layer?

Does this process inspect symbology (esp. image-based symbols) and label definitions in determining what colours to add to the palette?

Jason