[Geoserver-users] GeoTIFF layer anti-aliasing

Hello,

I have a GeoTIFF (with overviews) background layer.

When I am far from the raster layer resolution, this layer looks ugly.

- When I zoom in, at 300% the Tiff displayed with the "Eye of Gnome"
image viewer looks blurred but better than GeoServer. I am aware that
I can't make details appear ala NCIS!
- When I zoom out, at 33% the Tiff looks better in EoG as well.

I tried to enable bilinear/bicubic interpolation in the WMS
configuration panel but it doesn't help.

What can I do?
Gerald

Gérald Quintana ha scritto:

Hello,

I have a GeoTIFF (with overviews) background layer.

When I am far from the raster layer resolution, this layer looks ugly.

- When I zoom in, at 300% the Tiff displayed with the "Eye of Gnome"
image viewer looks blurred but better than GeoServer. I am aware that
I can't make details appear ala NCIS!
- When I zoom out, at 33% the Tiff looks better in EoG as well.

I tried to enable bilinear/bicubic interpolation in the WMS
configuration panel but it doesn't help.

What can I do?

Hmmm... what I'm wondering if "Eye of the Gnome" is using overviews
or computing an optimal scaled out image on the fly ignoring
overviews.
If the latter is true, you may want to recompute the overviews
in your tiff file using a different interpolation _for the overviews_,
see: http://www.gdal.org/gdaladdo.html

Cheers
Andrea

2008/4/22 Andrea Aime <aaime@anonymised.com>:

Hmmm... what I'm wondering if "Eye of the Gnome" is using overviews
or computing an optimal scaled out image on the fly ignoring
overviews.

Testing with other image viewers, I realize that EoG is very good at
downsampling/interpolating images.
The problem is not related to GeoTiff, using PNG+WLD instead is ugly as well.

If the latter is true, you may want to recompute the overviews
in your tiff file using a different interpolation _for the overviews_,
see: http://www.gdal.org/gdaladdo.html

I tried to add "-r average/average_magphase" to change interpolation,
but it doesn't help.

I guess, there is no solution. Thanks for your answer Andrea,
Gerald