Would this Geopackage be suitable for bringing imagery into QField on Android? Or is there a more suitable way to incorporate imagery to use on a tablet without utilizing large amounts of data?
Thanks for your effort. I'd appreciate to see in some next video (:)no rush(:) using single-band raster loading and storing using geopackages. We have a LiDAR based DEM... over 100Gb in Erdas img file... :/
Ouch! To be honest, even if you can put that in a GeoPackage and compress it a lot, you probably shouldn't. It's the individual raster pixel values that are important, and that will be destroyed by compression... Probably better of with another solution. Maybe resample the data set into new layers for different resolutions... That way you can tie the layers to scale in your project and only show/use the full resolution in the largest scales.
Very nice. I've been playing with google earth engine scripts as a nice really fast way to download sentinel imagery. Will have to figure out if I can still convert that to geopackage.
Is the Geopackage the quality solution?,... I mean, this database format have tools or characteristics about the QA & QC matters of the geospatial & attribute Data?,.. great video,... greetings.
Are you restricted to just 3 rasters for your virtual build? I tried R,G,B, NIR from Landsat 4-5 and the color was very different between the compressed and the virtual. Should limit this to 3 bands?
I dont really get how resolution levels work, i have been searching some docs but not found an explanation. How does 2 vs 256 levels effect resolution of tiles. Is it that it cut the raster in 2 vs 256 parts?
Assuming you mean pyramid levels. Level 2 will generate a new raster at 1/2 the resolution. Level 4 at 1/4 resolution. 512 will generate a pyramid raster at 1/512 resolution. All tiles are default 256x256 pixels large, and that is applied to all pyramid levels, as well as the base resolution image.
Cannot write raster. Error code: data source creation Cannot create new dataset C:\Users\luisa\Desktop\teste.gpkg: Only 1 (Grey/ColorTable), 2 (Grey+Alpha), 3 (RGB) or 4 (RGBA) band dataset supported for Byte data type. How can i solve this problem, the source files are in geopachage... Strange...
@@luisalmeida9904 You can probably use the raster "Translate" command. In there you can set "-expand rgba" as additional command-line parameters. The result will not have pyramids, but that can be fixed separately with the "Build overviews" command.
hey klas that was really hepl full and since geopackage was added to qgis it saved my life im using it more than i tought i would do and when teaching qgis i focus on explaining how important is this i exported 5 years of data of my work into geopackages with compression feature which saved me alot of time finding the .sho .prj .xml files of each vector and raster data i used i just dump everything i need in geopackage !! i love ur video they are very friendly i have a question what compression u recommend for discrete data? stay safe klas cheers 🍻🚴
I usually use "default" and set QUALITY=100 in custom settings. This will keep the highest quality of the data. If you can accept slight artifacts try lowering the value some. The default is 75.
Awesome, I was hoping you'd put this up as a tutorial Klas :-)
Thank you so much.
Would this Geopackage be suitable for bringing imagery into QField on Android? Or is there a more suitable way to incorporate imagery to use on a tablet without utilizing large amounts of data?
Great tutorial, thank you very much!
Thank you very much for this lecture
Thanks for your effort. I'd appreciate to see in some next video (:)no rush(:) using single-band raster loading and storing using geopackages. We have a LiDAR based DEM... over 100Gb in Erdas img file... :/
Ouch! To be honest, even if you can put that in a GeoPackage and compress it a lot, you probably shouldn't. It's the individual raster pixel values that are important, and that will be destroyed by compression... Probably better of with another solution. Maybe resample the data set into new layers for different resolutions... That way you can tie the layers to scale in your project and only show/use the full resolution in the largest scales.
@@KlasKarlsson Well, actually I did not find any big problems reading and displaying that HUGE img file in QGIS.
@@KlasKarlsson Thanks for your response.
Did you change the Resampling Algorithm to "Average" by accident by scrolling your mouse wheel or could one choose the defaul "Nearest neighbour"?
Thanks for the tutorial!
Very nice. I've been playing with google earth engine scripts as a nice really fast way to download sentinel imagery. Will have to figure out if I can still convert that to geopackage.
Is the Geopackage the quality solution?,... I mean, this database format have tools or characteristics about the QA & QC matters of the geospatial & attribute Data?,.. great video,... greetings.
Are you restricted to just 3 rasters for your virtual build? I tried R,G,B, NIR from Landsat 4-5 and the color was very different between the compressed and the virtual. Should limit this to 3 bands?
I dont really get how resolution levels work, i have been searching some docs but not found an explanation. How does 2 vs 256 levels effect resolution of tiles. Is it that it cut the raster in 2 vs 256 parts?
Assuming you mean pyramid levels. Level 2 will generate a new raster at 1/2 the resolution. Level 4 at 1/4 resolution. 512 will generate a pyramid raster at 1/512 resolution. All tiles are default 256x256 pixels large, and that is applied to all pyramid levels, as well as the base resolution image.
@@KlasKarlsson right ok thank you! I have had a difficult time conceptualize it 🙂 That make sense!
Cannot write raster. Error code: data source creation Cannot create new dataset C:\Users\luisa\Desktop\teste.gpkg:
Only 1 (Grey/ColorTable), 2 (Grey+Alpha), 3 (RGB) or 4 (RGBA) band dataset supported for Byte data type.
How can i solve this problem, the source files are in geopachage... Strange...
For this method your source data need to be three band with 8-bit values.
@@KlasKarlsson thanks. Is there a easy way to convert my rasters on qgis?
@@luisalmeida9904 You can probably use the raster "Translate" command. In there you can set "-expand rgba" as additional command-line parameters. The result will not have pyramids, but that can be fixed separately with the "Build overviews" command.
@@KlasKarlssonapears a mensagem Saint that raster does not have tablet...
Used 20+ jp2 files to create a virtual raster but when trying to export it to Geopackage with Klas' settings I get the same error message.
hey klas that was really hepl full and since geopackage was added to qgis it saved my life im using it more than i tought i would do and when teaching qgis i focus on explaining how important is this i exported 5 years of data of my work into geopackages with compression feature which saved me alot of time finding the .sho .prj .xml files of each vector and raster data i used i just dump everything i need in geopackage !! i love ur video they are very friendly i have a question what compression u recommend for discrete data? stay safe klas cheers 🍻🚴
I usually use "default" and set QUALITY=100 in custom settings. This will keep the highest quality of the data. If you can accept slight artifacts try lowering the value some. The default is 75.
Tx