vincentsarago
commented
2 years ago @zacharyDez 👋
your COGs don't have CRS or geo_transform set, but use GCPS (ground control points) for geo-reference.
Sadly you can't use cogeo-mosaic cli directly for this.
- write a small CLI
# cog_gcp_info.py from rio_tiler.io import GCPCOGReader
import click from rasterio.rio import options import json
@click.command() @options.file_in_arg def main(input): """Read tile.""" with GCPCOGReader(input) as cog: bounds = cog.geographic_bounds click.echo( json.dumps( { "geometry": { "type": "Polygon", "coordinates": [ [ [bounds[0], bounds[3]], [bounds[0], bounds[1]], [bounds[2], bounds[1]], [bounds[2], bounds[3]], [bounds[0], bounds[3]], ] ], }, "properties": { "path": input, "bounds": cog.geographic_bounds, "minzoom": cog.minzoom, "maxzoom": cog.maxzoom, "datatype": cog.dataset.meta["dtype"], }, "type": "Feature", } ) )
if name == 'main': main()
2. Run CLI on your files
```bash
# using parallel is optional
$ cat 1996.txt | parallel -j 4 python -m cog_gcp_info {} > 1996.json- run cogeo-mosaic cli
create-from-features
$ cat 1996.json| cogeo-mosaic create-from-features --minzoom 8 --maxzoom 12 -o 1996_mosaic.json --prop
erty path
zacharyDez
Hi @vincentsarago,
I've been trying to test out the creation of a mosaic from a repo of radarsat imagery of the canadian federal government. I saw some issues related to s3 paths and replacing with the equivalent HTTP URLs. I can curl the URLs defined within the .txt file.
I initially created the list of images with:
I lazily just deleted all the radarsat imagery that did not have the same configurations to avoid the datatype errors. I then ran
cogeo-mosaic create 1996.txt -o mosaic.json. I'm getting the following warnings but I am not sure how to interpret them:The resulting mosaic has no tiles:
The output .txt is:
Any suggestions on next steps to solve the issue? Any other tips are more than welcome.
Thanks