Re: Fast X-array rasterization
Philipp Kats
Hi Brendan, Thanks! Indeed, that what I am trying to achieve. and your hint to ints is great! In fact, I could boil it down to store binary masks - feature id would be stored in the Z dimension itself, and all pixels on one layer will likely have the same shared feature-based value after all. Moreover, this way I could calculate this 3d shape once - and then lookup values and perform computations on this cube for any feature's property I want. So this will definitely boost my computations by a lot. The second one also sounds very legit but sounds a little trickier to implement - we'll see. Still, I wonder if it would make sense to (somehow) use gdal_rasterize bands here...
On Mon, Sep 10, 2018 at 11:06 PM Brendan Ward <brendan.charles.ward@...> wrote: I'm not entirely sure I follow what you are aiming for here yet. Are you trying to aggregate across features per pixel to calculate various statistics? Like, what is the average feature size of all features that intersect this pixel?
|
|