- **status**: open --> closed-fixed
- **Comment**:
There are still small problems with georeferencing, but it seems to be due to
the usage of float instead of doubles, and it is much less than 1/2 pixel. See
#507. I close this ticket
---
** [bugs:#503] 1/2 pixel shift in image
Seems to be fixed for all legacy drivers. All tests have been made with a
normal RGB Geotiff image from IGN :
- was already OK with Commons imaging
- has been fixed by r6485 for GeoTIFF driver in
com/vividsolutions/jump/workbench/imagery/geotiff/RasterPainter
- has been fixed by r6478 for all
no problemo.
maybe you shouldn't test readers as such but ways of georeferencing? meaning
worldfile, gdal & geotiff parsing. probably easiest if you simply temporarily
comment/uncomment those in GeoReferencedRaster.. ede
On 9/14/2020 23:17, michael michaud via Jump-pilot-devel wrote:
> Thanks
Thanks Ede, it helped ;-)
I could make a fix. I still must check if it works for every loader though.
---
** [bugs:#503] 1/2 pixel shift in image georeferencement**
**Status:** open
**Milestone:** OJ_1.16
**Created:** Sun Sep 13, 2020 09:52 AM UTC by michael michaud
**Last Updated:** Mon Sep
Seems to me that Sextante Raster and Commons Imaging referencing are correct
and that other ones are shifted. I don't know why images loaded with Commons
Imaging are not referenced as others. Ede, any hint to help me understand which
loader uses which ImageFactory/ReferencedImage ?
---
**
- **Milestone**: undecided --> OJ_1.16
---
** [bugs:#503] 1/2 pixel shift in image georeferencement**
**Status:** open
**Milestone:** OJ_1.16
**Created:** Sun Sep 13, 2020 09:52 AM UTC by michael michaud
**Last Updated:** Sun Sep 13, 2020 11:32 AM UTC
**Owner:** nobody
There is still 1/2
There is the following comment from Stefan Steiniger in the
AddRasterImageLayerWizard class :
~~~
TODO: [sstein] Feb.2009 - I discovered a 0.5px offset towards south-east
for the envelope, in comparison with images loaded with
Jon's/VividSolutions implementation, if the envelope is obtained from
>There is still 1/2 pixel difference between image georeferenced with
Sextant Raster reader and images georeferenced with other readers. I have
not investigated yet and don't know which one is correct.
AFAIR that was a quite old issue in Sextante. Eventually (if I am not
wrong) Stefan corrected
---
** [bugs:#503] 1/2 pixel shift in image georeferencement**
**Status:** open
**Milestone:** undecided
**Created:** Sun Sep 13, 2020 09:52 AM UTC by michael michaud
**Last Updated:** Sun Sep 13, 2020 09:52 AM UTC
**Owner:** nobody
There is still 1/2 pixel difference between image