Hi,
I have a strange effect: font rendering is different on a geoserver instance when accessing the same tile directly, and via the built-in tile cache.
See the two attached images as a demonstration. The URLs for these images are:
direct (non-cached), looking good:
http://openaviationmap.tyrell.hu/geoserver/OAM/wms?service=WMS&request=GetMap&version=1.1.1&layers=oam&styles=&format=image/png&transparent=false&height=256&width=256&srs=EPSG:3857&bbox=2152466.7165105627,5987771.047747566,2191602.4749925733,6026906.806229578
cached, looking bad:
http://openaviationmap.tyrell.hu/geoserver/gwc/service/wms?service=WMS&request=GetMap&version=1.1.1&layers=OAM:oam&styles=&format=image/png&transparent=false&height=256&width=256&srs=EPSG:3857&bbox=2152466.7165105627,5987771.047747566,2191602.4749925733,6026906.806229578
This is on a geoserver running on Ubuntu 12.04 LTS with Oracle JDK 7, and with ms corefonts installed (the effect is the sameon Ubuntu 12.10). the font definition for the text in question is:
<sld:Font>
<sld:CssParameter name="font-family">Arial</sld:CssParameter>
<sld:CssParameter name="font-size">8</sld:CssParameter>
<sld:CssParameter name="font-style">normal</sld:CssParameter>
<sld:CssParameter name="font-weight">normal</sld:CssParameter>
</sld:Font>
I wonder what the source of this issue is, and how can one resolve it?
best regards,
Akos


Interesting. You didn't change the SLD afterwards ?
If it really is a problem in the caching layer, it appears to be while encoding the PNG-file, since the JPEG looks okay (ignoring artifacts resulting from the JPEG compression).
http://openaviationmap.tyrell.hu/geoserver/gwc/service/wms?service=WMS&request=GetMap&version=1.1.1&layers=OAM:oam&styles=&format=image/jpeg&transparent=false&height=256&width=256&srs=EPSG:3857&bbox=2152466.7165105627,5987771.047747566,2191602.4749925733,6026906.806229578
-Arne
On 1/11/13 09:22 , Ákos Maróy wrote:
Hi,
I have a strange effect: font rendering is different on a geoserver instance when accessing the same tile directly, and via the built-in tile cache.
See the two attached images as a demonstration. The URLs for these images are:
direct (non-cached), looking good:
http://openaviationmap.tyrell.hu/geoserver/OAM/wms?service=WMS&request=GetMap&version=1.1.1&layers=oam&styles=&format=image/png&transparent=false&height=256&width=256&srs=EPSG:3857&bbox=2152466.7165105627,5987771.047747566,2191602.4749925733,6026906.806229578
cached, looking bad:
http://openaviationmap.tyrell.hu/geoserver/gwc/service/wms?service=WMS&request=GetMap&version=1.1.1&layers=OAM:oam&styles=&format=image/png&transparent=false&height=256&width=256&srs=EPSG:3857&bbox=2152466.7165105627,5987771.047747566,2191602.4749925733,6026906.806229578
This is on a geoserver running on Ubuntu 12.04 LTS with Oracle JDK 7, and with ms corefonts installed (the effect is the sameon Ubuntu 12.10). the font definition for the text in question is:
<sld:Font>
<sld:CssParameter name="font-family">Arial</sld:CssParameter>
<sld:CssParameter name="font-size">8</sld:CssParameter>
<sld:CssParameter name="font-style">normal</sld:CssParameter>
<sld:CssParameter name="font-weight">normal</sld:CssParameter>
</sld:Font>
I wonder what the source of this issue is, and how can one resolve it?
best regards,
Akos
On Fri, Jan 11, 2013 at 9:53 AM, Arne Kepp <arne@anonymised.com> wrote:
Interesting. You didn’t change the SLD afterwards ?
If it really is a problem in the caching layer, it appears to be while encoding the PNG-file, since the JPEG looks okay (ignoring artifacts resulting from the JPEG compression).
Known issue under JDK 7 (which is not supported at the moment):
http://jira.codehaus.org/browse/GEOS-5486
Afaik GWC always sets transparent=true in its internal requests, regardless of how the tile was asked.
Cheers
Andrea
–
Ing. Andrea Aime
@geowolf
Technical Lead
GeoSolutions S.A.S.
Via Poggio alle Viti 1187
55054 Massarosa (LU)
Italy
phone: +39 0584 962313
fax: +39 0584 1660272
mob: +39 339 8844549
http://www.geo-solutions.it
http://twitter.com/geosolutions_it
That’s a really useful warning. Does a halo help ? (Obviously not always an option.) -Arne
···
On 1/11/13 10:06 , Andrea Aime wrote:
On Fri, Jan 11, 2013 at 9:53 AM, Arne Kepp <arne@anonymised.com> wrote:
Interesting. You didn’t change the SLD afterwards ?
If it really is a problem in the caching layer, it appears to be while encoding the PNG-file, since the JPEG looks okay (ignoring artifacts resulting from the JPEG compression).
Known issue under JDK 7 (which is not supported at the moment):
http://jira.codehaus.org/browse/GEOS-5486
Afaik GWC always sets transparent=true in its internal requests, regardless of how the tile was asked.
Cheers
Andrea
indeed, it works with JDK 1.6 thanks
···
On 11/01/13 10:06, Andrea Aime wrote:
On Fri, Jan 11, 2013 at 9:53 AM, Arne Kepp <arne@anonymised.com> wrote:
Interesting. You didn’t change the SLD afterwards ?
If it really is a problem in the caching layer, it appears to be while encoding the PNG-file, since the JPEG looks okay (ignoring artifacts resulting from the JPEG compression).
Known issue under JDK 7 (which is not supported at the moment):
http://jira.codehaus.org/browse/GEOS-5486
Afaik GWC always sets transparent=true in its internal requests, regardless of how the tile was asked.
it seems not, the example I’m showing does have a halo
···
On 11/01/13 10:24, Arne Kepp wrote:
That’s a really useful warning. Does a halo help ? (Obviously not always an option.)
On Fri, Jan 11, 2013 at 10:24 AM, Arne Kepp <arne@anonymised.com> wrote:
On 1/11/13 10:06 , Andrea Aime wrote:
On Fri, Jan 11, 2013 at 9:53 AM, Arne Kepp <arne@anonymised.com> wrote:
Interesting. You didn’t change the SLD afterwards ?
If it really is a problem in the caching layer, it appears to be while encoding the PNG-file, since the JPEG looks okay (ignoring artifacts resulting from the JPEG compression).
Known issue under JDK 7 (which is not supported at the moment):
http://jira.codehaus.org/browse/GEOS-5486
Afaik GWC always sets transparent=true in its internal requests, regardless of how the tile was asked.
Cheers
Andrea
That’s a really useful warning. Does a halo help ? (Obviously not always an option.)
A solid halo should help, yes
Cheers
Andrea
–
Ing. Andrea Aime
@geowolf
Technical Lead
GeoSolutions S.A.S.
Via Poggio alle Viti 1187
55054 Massarosa (LU)
Italy
phone: +39 0584 962313
fax: +39 0584 1660272
mob: +39 339 8844549
http://www.geo-solutions.it
http://twitter.com/geosolutions_it