Sebastian Hanß

> From a computational point, we need to calculate all classes anyways and then either just return one specific class or all. Not necessarily, I think. I'm just hacking a...

We are working on a class-level moving window implementation (#211). When we have that, it should be trivial, right?

> How is that related to a moving window? CWED is just another metric as far as my understanding goes... @bonushenricus wrote: > Yes, I could actually do what you...

I agree. We already have terra support, don't we? And internally we use plain matrices, so there is probably not too much to do. Also, we can just support raster,...

I worked a bit on the terra branch with the idea to move from `raster::focal` to `terra::focal` because `terra::focal` supports vectors as output for the focal function. I think this...

I have no idea about this particular metric, but here is my shot :-) > 1. > According to the FRAGSTATS manual, we need "[...] landscape area, including all internal...

> That would at least help with the padding yes. But we still have the problem that we can not distinguish between NA that are within the landscape (e.g. water...

> Regarding NAs in general, that's where FRAGSTATS is a bit unclear, but I think they do conceptually make a difference between the landscape boundary and NAs that are within...

> Not sure if we are actually talking about the same smile I am thinking of the example where e.g. a circular polygon was used to mask the original landscape...

 like so?