[GRASS-user] r.colors generates different colors than QGIS plugin when same rules used?

[Tim Holland]

Hello, 

Sorry for what is becoming a very long question (and perhaps too many
images).  

I put up two more images that are probably better at demonstrating what I'm
talking about.  They are zoomed in on the same section of the Landsat
composite, but with one image made in GRASS d.rast
(http://i1020.photobucket.com/albums/af326/timothyholland/HLB_brovey_GRASS_Zoom.jpg)
and the other in QGIS
(http://i1020.photobucket.com/albums/af326/timothyholland/HLB_brovey_QGIS_Zoom.jpg). 
Is the issue just that QGIS has greater color depth than GRASS d.rast or the
PNG driver?  That's what these look like.  Is there a way to get greater
color depth in GRASS?  

Thanks again, 
Tim







Tim Holland wrote:
> 
> Hello List, 
> 
> I am trying to generate PNG files of a few different maps using r.colors,
> d.rast, and the PNG driver.  
> 
> The images I get from this process end up being colored slightly
> differently than when the equivalent images are generated in QGIS.  It is
> far more convenient to use the PNG driver right in GRASS (because I can
> set a script running and print out many images at once), but I prefer the
> color results I get with QGIS.  
> 
> In GRASS, I am using
> 
> r.colors map=ForestChange rules=colorTable.file
> d.mon start=PNG
> d.rast map=ForestChange
> d.mon stop=PNG 
> 
> with the rules file (colorTable.file) as follows using the rgb values for
> each level:
> -15 255 0	0
> -0.5 255 240 180
> 0 white
> 0.5 180 230 130
> 15 0 70 0
> 
> (the idea with this particular rules file is to generate deforestation /
> reforestation maps where zero change is shown as white, with beige-red
> representing increasing levels of deforestation, and light green - dark
> green representing increasing levels of reforestation).  
> 
> Then, I take the same raster file (ForestChange), load it in QGIS, and set
> its colormap according to what I think are the same rules:
> http://i1020.photobucket.com/albums/af326/timothyholland/QGIScolorTable.jpg
> 
> The two resulting files don't look the same, however.  The GRASS PNG
> output has a far more yellow-ish tone.  I think the QGIS one is quite a
> bit nicer.  GRASS:
> http://i1020.photobucket.com/albums/af326/timothyholland/GRASS_PNG_map.jpg
> QGIS:
> http://i1020.photobucket.com/albums/af326/timothyholland/QGISmap.jpg.  I
> have tried the various flags in the r.colors module (-g / Logarithmic
> scaling; -e / Histogram equalization; and the default), but all seem to
> return a similar result to the image which is given here).  
> 
> As a test, I also took a landsat composite image that I had generated
> using r.composite.  In this case, the color table was set by r.composite
> without any input from me, so I was really expecting the same output. 
> However, again, the QGIS output has much more natural-seeming colors. 
> GRASS:
> http://i1020.photobucket.com/albums/af326/timothyholland/HLB_brovey_GRASS.jpg
> QGIS:
> http://i1020.photobucket.com/albums/af326/timothyholland/HLB_brovey_QGIS.jpg. 
> 
> Does anyone know if there is a way to better approximate the QGIS color
> results using GRASS directly?  According to the QGIS GUI, color
> interpolation (for the forest maps, at least) is linear, which should be
> the most straightforward interpolation, shouldn't it?  
> 
> Thanks for any help / advice anyone can provide. 
> 
> Best, 
> Tim
> 
> 

-- 
View this message in context: http://n2.nabble.com/r-colors-generates-different-colors-than-QGIS-plugin-when-same-rules-used-tp3558680p3558754.html
Sent from the Grass - Users mailing list archive at Nabble.com.