Open mortac8 opened 4 years ago
Hi, @mortac8 . Thank you for posting this issue. I'll create an example that demonstrates this along with some documentation. It may take a couple days. I'll update this issue when that is complete. Thanks!
Hi, @mortac8 . Sorry, I haven't found time to create an example, but for what it's worth, this library should automatically load the minimal amount of data (including from internal or external overviews). It uses georaster, which uses geotiff.js, which is the code actually doing the fetching of tiles. I'll try to write something up more formally later and add that to the documentation.
Thanks for the info. I am on the fence on using your library vs one of these. My goal is to work with a large GeoTIFF file (~3GB) that has 16-bits per channel (RGB) which will be 'enhanced' through our algorithms (calculations need to be done on the 16-bit bands) and finally displayed onscreen in the standard 8-bit space.
https://github.com/christian-werner/leaflet-geotiff https://github.com/danwild/leaflet-geotiff-2
https://jsfiddle.net/mortac8/unjmv7o8/1/
Is that the general idea? Any way to fix issues where some of the tiles seem not to line up like on the roads?
Yup. The misalignment of the roads appears to be a by-product of both reprojecting and sampling the raster at the same time. There's a few possible solutions such as tweaking the resolution param and I can look at improving the core algorithm, but the easiest would be to reproject the raster into the resolution of the map. Assuming you are using Leaflet with the standard 3857 projection, could you reproject the enhanced 8-bit GeoTIFF into 3857?
I can also try to investigate this more thoroughly.
My goal is to load the minimal amount of data from a COG with internal overviews. Is this possible or do I need to be using external overviews?
I want to either load the entire image (using the internal overviews ex: 500x500) or load the actual data when far zoomed in. Any breadcrumbs to help me with this or best practices advice would be appreciated. Thanks!