[Geoserver-users] [Geoserver-devel] problem styling tiff coverages

Guys, you all know how it feels when you ask a (for you important) question on the list, and nobody answers. Please don't do that do me...

When no one answers, you really don't know whether 1) nobody read your message; 2) nobody knows the answer, 3) you misformulated your question; 4) everybody thinks your problem is nonsense; 5) nobody likes you; 6)...

So please give me some hints.

To freshen up the minds: I have problems using styled (sld) 1 band geotiff images in geoserver-trunk: the sld's seem to be ignored for geotiff files. The original message is below.
I have send the original message to the dev list; cc-ing now to the users list too to reach a bigger public.

Cheers,
Vincent Schut.

--- original message: ---

Hi all,

I am encountering problems when trying to apply an sld with
rastersymbolizer to my coverages. It's a bit hard to really find out
what is wrong, but it seems styling does not like geotiffs (or vice
versa). I'll try to give a summary of what I tried and hope it rings a
bell somewhere...

- Applying a style (like dem.sld, raster.sld, or a customized version of
one of those) *does* work on the supplied gtopo30 sample raster;
- Applying a style (idem) on my custom coverage does *not* work; the
raster gets rendered in grayscale, whatever I try. This raster is a
single band, tiled, Int32 geotiff, no colortable;
- Applying a style (idem) on a geotiff version (non-tiled, uint16, 1
band, no colortable) of the gtopo30_sample (created with gdal_translate)
does *not* work, neither does a baseline tiff with tfw file version;
- Applying a style to a fresh copy of the gtopo30 sample dataset
(created new dir with these files, created new coverageDataset, create
new Coverage) *does* work;

This is what I could think of to narrow down the problem, and leads me
to the following conclusion: styling does not work for tiff files. I
might be wrong and it sounds kind of illogical (why would it not work
for tiffs, but work for gtopo30 rasters, while I suppose that the
rendering chain is basically the same except for the file reader...),
but I have no idea how to investigate this any further. So I hope some
of the imaging/wcs guru's can shed some light on this (and fix it, I
hope :)).

Some additional info: using sun jdk 1.5.0-09, jai and jai-imageio from
cvs (synced yesterday), geotools (2.3.x branch) from svn (yesterday),
geoserver svn trunk from yesterday, deployed in tomcat-5.5.

Cheers from a bit desparate Vincent (I /do/ need raster styling...).

-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys - and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
_______________________________________________
Geoserver-devel mailing list
Geoserver-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/geoserver-devel

Vincent Schut ha scritto:

Guys, you all know how it feels when you ask a (for you important) question on the list, and nobody answers. Please don't do that do me...

When no one answers, you really don't know whether 1) nobody read your message; 2) nobody knows the answer, 3) you misformulated your question; 4) everybody thinks your problem is nonsense; 5) nobody likes you; 6)...

Nobody knowing an answer read your message unfortunately afaik.
I'm learning the inside workings of WCS now, so I may be able to
give you an answer in a few days.
Anyways, I can confirm what you said, WCS seems to ignore the SLD
with Geotiff images. I guess it assumes Geotiffs are always images,
whilst in your case each pixel has a physical value, right?

Cheers
Andrea

Andrea Aime wrote:

Vincent Schut ha scritto:

Guys, you all know how it feels when you ask a (for you important) question on the list, and nobody answers. Please don't do that do me...

When no one answers, you really don't know whether 1) nobody read your message; 2) nobody knows the answer, 3) you misformulated your question; 4) everybody thinks your problem is nonsense; 5) nobody likes you; 6)...

Nobody knowing an answer read your message unfortunately afaik.

Hmm that's what I feared. Sometimes it's hard to write your message the right way so you get the attention from the right person(s)... Glad you took the time to answer, thanks. Btw I had also send it to Simone directly hoping that he would be the right person, but got no answer... I hope I have not offended him by doing that...

I'm learning the inside workings of WCS now, so I may be able to
give you an answer in a few days.

That would be great. If this is not solved quickly, I'm afraid I'll have to fall back on mapserver for most of my raster rendering. Obviously I'd rather do everythin with geoserver now. Note this is not a threat or so, but just to tell you guys I think it is a serious flaw which is a showstopper for me regarding the use of geoserver with raster imagery. Most of my rasters are not rgb images, but classified rasters. Alas my java skills are less than basic, so I don't feel like being capable of hacking geoserver myself.

Anyways, I can confirm what you said, WCS seems to ignore the SLD
with Geotiff images. I guess it assumes Geotiffs are always images,
whilst in your case each pixel has a physical value, right?

True, usually stuff like landuse classes etc. I'd say that SLD should be ignored with images that advertise their bands as rgb or rgba, and that at least all single band imagery should be handled by its SLD. IMHO (no offense) It's kind of dumb to let this choice depend on file format, instead of file/bands type.

Cheers
Andrea

Vincent Schut ha scritto:

Andrea Aime wrote:

Vincent Schut ha scritto:

Guys, you all know how it feels when you ask a (for you important) question on the list, and nobody answers. Please don't do that do me...

...

Anyways, I can confirm what you said, WCS seems to ignore the SLD
with Geotiff images. I guess it assumes Geotiffs are always images,
whilst in your case each pixel has a physical value, right?

True, usually stuff like landuse classes etc. I'd say that SLD should be ignored with images that advertise their bands as rgb or rgba, and that at least all single band imagery should be handled by its SLD. IMHO (no offense) It's kind of dumb to let this choice depend on file format, instead of file/bands type.

Heh, there's a reason why Geoserver WCS is still beta :slight_smile:
We have it because two nice italian guys spend the (big) effort needed
to set it up, but we cannot ask them to work for free all the time :-p
The reason WCS was set up is because they needed to display big satellite and airborne images, and it does a good job on this use case.

If the SLD issue it's a small thing, I think it'll be fixed soon, but other than, we can make no promises. That's open source, things get
done when someone volounteeers to do them, or someone else sponsors
development.

Cheers
Andrea

Hi Vince,

sorry for not answering before … we are working on upgrading the GeoTools RasterSymbolizer. The new version (when ready) will be able to handle even GeoTIFFs.

Cheers,
Alessio.

On 1/15/07, Vincent Schut <schut@anonymised.com> wrote:

Andrea Aime wrote:

Vincent Schut ha scritto:

Guys, you all know how it feels when you ask a (for you important)
question on the list, and nobody answers. Please don’t do that do me…

When no one answers, you really don’t know whether 1) nobody read
your message; 2) nobody knows the answer, 3) you misformulated your
question; 4) everybody thinks your problem is nonsense; 5) nobody
likes you; 6)…

Nobody knowing an answer read your message unfortunately afaik.
Hmm that’s what I feared. Sometimes it’s hard to write your message the
right way so you get the attention from the right person(s)… Glad you
took the time to answer, thanks. Btw I had also send it to Simone
directly hoping that he would be the right person, but got no answer…
I hope I have not offended him by doing that…
I’m learning the inside workings of WCS now, so I may be able to
give you an answer in a few days.
That would be great. If this is not solved quickly, I’m afraid I’ll have
to fall back on mapserver for most of my raster rendering. Obviously I’d
rather do everythin with geoserver now. Note this is not a threat or so,
but just to tell you guys I think it is a serious flaw which is a
showstopper for me regarding the use of geoserver with raster imagery.
Most of my rasters are not rgb images, but classified rasters. Alas my
java skills are less than basic, so I don’t feel like being capable of
hacking geoserver myself.
Anyways, I can confirm what you said, WCS seems to ignore the SLD
with Geotiff images. I guess it assumes Geotiffs are always images,
whilst in your case each pixel has a physical value, right?
True, usually stuff like landuse classes etc. I’d say that SLD should be
ignored with images that advertise their bands as rgb or rgba, and that
at least all single band imagery should be handled by its SLD. IMHO (no
offense) It’s kind of dumb to let this choice depend on file format,
instead of file/bands type.

Cheers
Andrea


Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net’s Techsay panel and you’ll get the chance to share your
opinions on IT & business topics through brief surveys - and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV


Geoserver-devel mailing list
Geoserver-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/geoserver-devel

Eng. Alessio Fabiani
Vice President/CTO GeoSolutions

http://www.geo-solutions.it


Andrea Aime wrote:

Vincent Schut ha scritto:

Andrea Aime wrote:

Vincent Schut ha scritto:

Guys, you all know how it feels when you ask a (for you important) question on the list, and nobody answers. Please don't do that do me...

...

Anyways, I can confirm what you said, WCS seems to ignore the SLD
with Geotiff images. I guess it assumes Geotiffs are always images,
whilst in your case each pixel has a physical value, right?

True, usually stuff like landuse classes etc. I'd say that SLD should be ignored with images that advertise their bands as rgb or rgba, and that at least all single band imagery should be handled by its SLD. IMHO (no offense) It's kind of dumb to let this choice depend on file format, instead of file/bands type.

Heh, there's a reason why Geoserver WCS is still beta :slight_smile:

No beta's, no fun :slight_smile:
It's just that I had never heard about this 'bug', though following the geoserver lists pretty closely. So I assumed this part would be implemented already.

We have it because two nice italian guys spend the (big) effort needed
to set it up, but we cannot ask them to work for free all the time :-p

Sure, and did I already say their work is highly appreciated? If not, here's again :slight_smile:

The reason WCS was set up is because they needed to display big satellite and airborne images, and it does a good job on this use case.

If the SLD issue it's a small thing, I think it'll be fixed soon, but other than, we can make no promises. That's open source, things get
done when someone volounteeers to do them, or someone else sponsors
development.

Which I fully understand. If we could sponsor, we would. It's a pity we can't currently, I'd like to express my thanks in more ways than just emails.
Hey, good luck further implementing wcs! And don't worry, I'm still mostly a very happy geoserver user :slight_smile:

Cheers
Andrea

Vincent Schut wrote:

Guys, you all know how it feels when you ask a (for you important) question on the list, and nobody answers. Please don't do that do me...

When no one answers, you really don't know whether 1) nobody read your message; 2) nobody knows the answer, 3) you misformulated your question; 4) everybody thinks your problem is nonsense; 5) nobody likes you; 6)...

You've just about perfectly summed up Warnock's Dilemma...

http://en.wikipedia.org/wiki/Warnock’s_Dilemma

Unfortunately I have no idea on the raster side. Andrea and Brent are working to get up to speed on raster's so we can put 1.5 out as a stable release, so perhaps they can help work through this.

best regards,

Chris

So please give me some hints.

To freshen up the minds: I have problems using styled (sld) 1 band geotiff images in geoserver-trunk: the sld's seem to be ignored for geotiff files. The original message is below.
I have send the original message to the dev list; cc-ing now to the users list too to reach a bigger public.

Cheers,
Vincent Schut.

--- original message: ---

Hi all,

I am encountering problems when trying to apply an sld with
rastersymbolizer to my coverages. It's a bit hard to really find out
what is wrong, but it seems styling does not like geotiffs (or vice
versa). I'll try to give a summary of what I tried and hope it rings a
bell somewhere...

- Applying a style (like dem.sld, raster.sld, or a customized version of
one of those) *does* work on the supplied gtopo30 sample raster;
- Applying a style (idem) on my custom coverage does *not* work; the
raster gets rendered in grayscale, whatever I try. This raster is a
single band, tiled, Int32 geotiff, no colortable;
- Applying a style (idem) on a geotiff version (non-tiled, uint16, 1
band, no colortable) of the gtopo30_sample (created with gdal_translate)
does *not* work, neither does a baseline tiff with tfw file version;
- Applying a style to a fresh copy of the gtopo30 sample dataset
(created new dir with these files, created new coverageDataset, create
new Coverage) *does* work;

This is what I could think of to narrow down the problem, and leads me
to the following conclusion: styling does not work for tiff files. I
might be wrong and it sounds kind of illogical (why would it not work
for tiffs, but work for gtopo30 rasters, while I suppose that the
rendering chain is basically the same except for the file reader...),
but I have no idea how to investigate this any further. So I hope some
of the imaging/wcs guru's can shed some light on this (and fix it, I
hope :)).

Some additional info: using sun jdk 1.5.0-09, jai and jai-imageio from
cvs (synced yesterday), geotools (2.3.x branch) from svn (yesterday),
geoserver svn trunk from yesterday, deployed in tomcat-5.5.

Cheers from a bit desparate Vincent (I /do/ need raster styling...).

-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys - and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
_______________________________________________
Geoserver-devel mailing list
Geoserver-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/geoserver-devel

-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys - and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
_______________________________________________
Geoserver-devel mailing list
Geoserver-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/geoserver-devel

!DSPAM:1003,45ab834743111665516417!

--
Chris Holmes
The Open Planning Project
http://topp.openplans.org

Chris Holmes ha scritto:

Vincent Schut wrote:

Guys, you all know how it feels when you ask a (for you important) question on the list, and nobody answers. Please don't do that do me...

When no one answers, you really don't know whether 1) nobody read your message; 2) nobody knows the answer, 3) you misformulated your question; 4) everybody thinks your problem is nonsense; 5) nobody likes you; 6)...

You've just about perfectly summed up Warnock's Dilemma...

http://en.wikipedia.org/wiki/Warnock’s_Dilemma

Doh, nice read :slight_smile:
Cheers
Andrea