Merci Even,
I can confirm that the workaround you suggested works correctly.
>From the performance I was getting beforehand, I strongly believe that the
mixup happening on the overview levels was causing my hat/lr dataset to
never be used at all.
Thank you for the quick response,
Thomas
On Mon, Ju
Salut Thomas,
There were actually 2 issues:
- contrary to what the documentation stated, OVERVIEW_0_ metadata
items were simply ignored, hence you only got overviews starting at the
1/4 reduction
- and the documentation also lied when implying that it would
automatically add overviews o
