[Zoo-discuss] Request For Changess: customize styling of output geospatial ComplexDatas

[Fenoy Gerald]

Dear Remi,
many thanks for your proposal. This would be a very nice enhancement for the automatic publication of processing results.

As you proposed to use a branch for your development, it supposes that you have a ZOO-Project developer account. 

So the first step to proceed with your developper account is that you should read and confirm that you agree to adhere to the committer guidelines [1].

Once you will have confirmed that, I will send an email to the ZOO-Project PSC to request for a vote on your addition to the ZOO-Project committers list [2].

Note that we have noted your great contributions in the past, in example the ticket #131 [3].

We would be very happy to count you in our developper team.

I hope to hear back from you and that you will agree with your integration into the ZOO-Project committers.

Best regards,

[1] http://zoo-project.org/docs/contribute/dev.html
[2] http://zoo-project.org/docs/contribute/contributors.html
[3] http://zoo-project.org/trac/ticket/131

> Le 7 avr. 2017 à 11:25, Rémi Cresson <remi.cresson at teledetection.fr> a écrit :
> 
> Dear all,
> 
> I want to propose the following RFC for the zoo-kernel:
> 
> Currently in zoo, geospatial images outputs of a WPS process can be published as WMS through MapServer, which is very convenient for visualization. An extremely useful optional functionality, consisting in tagging the complex ProcessOutput whith "msClassify=true" in the .zcfg process metadata, enables a classes-oriented styling of the images (i.e. random color table). This is achieved in producing a mapfile whith the appropriate style. 
> While this absolutely rocks, I feel that there is still some room for improvement of this functionality. In my opinion, I would be really interesting to propose more than 2 styling ways (i.e. more than just "msClassify=true" or "msClassifiy=false"). Let's not reinvent the wheel for image vizualisation and styling: take QGIS for instance, which is quite operational on this side. QGIS proposes the essential methods to allows the vizualisation of any kind of images. Personally, I use 99% of my time the following ones:
> 
> 1. linear histogram stretching between min and max values of the image (e.g. radiometric indice image visualization)
> 2. linear histogram stretching between mean-2*std and mean+2*std values of the image (e.g. natural colors image visualization)
> 3. random pseudocolor for monoband images (e.g. classification image visualization)
> 4. custom pseudocolor for monoband images (e.g. thematic image visualization)
> 
> It would be a great improvement to allow the using of those styles (and more?) in the MapServer service, conditioned by the WPS process .zcfg metadatas of the desired output image. I propose therefore the following changes in metadatas for output images styling in the .zcfg:
> 
> -remove msClassify[0..1]: boolean
> 
> -add msStyle[0..1]: enum of {linear, random, custom} Description: when "msStyle=linear", the image is styled using a histogram stretch (all bands independently). When "msStyle=random", the image is styled with a random color table (equivalent to the obsolete "msClassify=true"). When "msStyle=custom", the image is styled with a custom mapfile.
> -add msStyleLinear[0..1]: enum of {minmax, meanstd, ...} Description: those enums set the low and high values for the linear histogram stretch.
> 
> -add msStyleCustom[0..1]: custom mapfile filename, e.g. "myNiceMapfile.map"
> 
> Again, those are just suggestions, and are proposed as a basis for discussions with the PSC and the zoo community. 
> I could implement this RFC in a dedicated feature branch.
> 
> I am waiting for your feedback, suggestions, opinions.
> Regards, 
> Rémi
> 
> _______________________________________________
> Zoo-discuss mailing list
> Zoo-discuss at lists.osgeo.org
> https://lists.osgeo.org/mailman/listinfo/zoo-discuss



Gérald Fenoy
http://wiki.osgeo.org/wiki/User:Djay