question about colors

[Malcolm Williamson]

On Sun, 6 Apr 1997, Tian Qing wrote:

> Hi,
> 
> I have a question about the image color within GRASS.
> 
> I tried to compose  a color spot image with 20 level colors,
> it finished,  but I couldn't show it.
                    ^^^^^^^^^^^^^^^^^^
I'm not quite sure exactly what you mean by that. 20 color levels means 
20 shades of red x 20 shades of green x 20 shades of blue, or a total of 
8000 colors. As the display program has to look up a color entry for each 
pixel, this tends to make for a slow display. Just as a note, color 
ranges in GRASS will display much quicker than long color tables (though 
I realize that this isn't much help with imagery).

> 
> I tried to compose it  with 256 level colors,
> it couldn't finish in 3 hours.

Let's see, that's 256^3, or 16,777,216 entries in your colortable. There 
could be problems.

> 
> Is there a limit for the color level to compose color
> image within GRASS? at most 10 level?

Not that I know of - I've certainly worked with 16 color levels in 
r.combine before.

> 
> My computer is Silicon Graphics workstation. It supports
> 16777216 colors( 24 bit).

Fine, but you can't deal with a 16 million line lookup table for dynamic 
display. The GRASS Xdriver will only display 256 colors (and I'm not sure 
that it's any different for the IRIS driver), so it's a moot point 
for most of us. Although it is theoretically possible to print 16 million 
colors from GRASS using ps.map, generating the maps would take forever. I 
generally use 12 to 16 colorlevels when I want a good looking map on a 
24-bit dye sublimation printer.

Remember, in most commercial remote sensing packages you are displaying 
three different bands at the same time to derive your 24-bit image; you 
are not writing a 24-bit lookup table. By the way, have you played with 
the SG3d module? It _does_ make full use of your display capabilities 
(and it's really cool, as well!).

Have fun,
	-Malcolm Williamson

> 


> Thanks a lot.
> 
> Tian Qing
> 
> 
> 
> 
>