[Qgis-developer] Raster colours

[Etienne Tourigny]

Jim, thanks for taking the time!

On Wed, Aug 1, 2012 at 2:55 PM, jjg <j.j.green at gmx.fr> wrote:
>
> Etienne Tourigny-3 wrote
>>
>> Iif you use a browser interface (like I implemented) it can get very
>> crowded.
>>
>> For example, the Color Brewer gradients have 7 variants each - no
>> sense in having 7 entries for the same palette.
>>
>
> No, there are 12 gradients with 3 segments, 12 gradients with 4,
> ... No actually there are 38 containing green and 72 not containing
> green... No actually there are ...
>
> Get my point? There are dozens of ways to categorize a set of gradients,
> and there is no obvious way to say which is better. How would you categorise
> the ds9 collection?

I would not categorize the ds9 collection - the palettes are all different.

The Color Brewer case is quite simple in my (perhaps
graphically-challenged) mind - group all the Blues_xx together, the
BuPu_xx together.

I am simply talking about "variants" of the same palette (with varying
number of colors) rather than classification, which can be arbitrary
as you point out.

GnBu_03, GnBu_04, ... , GnBu_09 => GnBu{_03,_04,...,_09}

>
>
> Etienne Tourigny-3 wrote
>>
>> When you present the palettes in a big page (like on your website),
>> it's ok to show them all, but in an application I find it's easier to
>> group them.
>>
>
> I see this as a problem which needs a creative GUI solution rather
> than an artificial and labour intensive categorisation.  How does
> PhotoShop solve this problem?  (I'm asking genuinely, I've never
> used PS).

Again, I am talking about groupping together palettes based on similar
names, it's not really labor intensive but can lead to some cases of
grouping error.

>
>
> Etienne Tourigny-3 wrote
>>
>>>     I can see it might be useful to have a description for each
>>>     subdirectory.  Say
>>>     - short name (same as directory name)
>>>     - long name (essentialy the text in the directory link on parent
>>> directory)
>>>     - description (essentially the first sentence of the page text)
>>>     so
>>>     "seq"
>>>     "sequential"
>>>     "Sequential colour schemes designed by Cynthia Brewer"
>>>     This info in a file called DESC.xml in the directory cpt-city/cb/seq/
>>
>> Yes that would be great, especially the description and long name. You
>> might put all information (description and copyright) in one file
>> though.
>>
>
> That would not be possible since there must be one DESC for each
> directory, but one might not have a COPYING in each directory.

ok

>
>
> Etienne Tourigny-3 wrote
>>
>>> 3) is a different matter altogether, some of these selections are updated
>>>     every day (most popular downloads etc), so I suggest we do the above
>>>     first an learn the lesson before trying this ...
>>
>> ok.  It would be cool to have xml files for those eventually, or a way
>> to parse the web pages that contain the lists.
>>
>
> Probably be easiest to have a fixed format XML file rather than trying
> to parse the HTML (which may well change)
>
>
> Etienne Tourigny-3 wrote
>>
>> Also - Tim also wrote to me that it would be interesting to distribute
>> within QGis a selection of gradients that allow distribution. Do you
>> know how I could search the archive for such gradients - except for
>> the obvious grep?
>>
>> On your site you write "those under GPL, Apache-like, Creative commons
>> or MIT licences allow distribution (under some conditions)" - which
>> restrictions are those, are they specified per-licence or per-author ?
>>
>
> Either you can browse the website and look at the copying links at
> the bottom of the page, or look at the COPYING.xml in the package
> (the latter generates the former).  That file applies to that directory
> and all of its subdirectories.
>
> You will need to judge for yourself whether the licence text allows
> redistribution,  for GPL, Apache, public domain this is clearly yes;
> for "no distribution allowed" this is clearly no, but there are some in
> the middle.  The distributor take the legal risks and so it is only right
> that the distributor makes this judgement.

ok thanks
Etienne

>
>
>
>
>
>
>
>
> --
> View this message in context: http://osgeo-org.1560.n6.nabble.com/Raster-colours-tp4991140p4992506.html
> Sent from the Quantum GIS - Developer mailing list archive at Nabble.com.
> _______________________________________________
> Qgis-developer mailing list
> Qgis-developer at lists.osgeo.org
> http://lists.osgeo.org/mailman/listinfo/qgis-developer