Stefan van der Walt

@stbnps My apologies--somehow (!) this issue slipped under the radar. If you are still interested, I'd be happy to pick up the thread.

Unfortunately, the code as posted also does not have a license.

@jni Is this functionality in addition to what exists in `scipty.ndimage.distance_transform_*`?

The `apply` seems superfluous. How about just `filter_forward` and `filter_inverse`? Also, we need to keep at least `inverse` as an alias (with deprecation), since that was public before.

I will send Paul Debevec an e-mail to sort out the licensing issue.

Before I sent the email, I had a look at the paper itself. That code is really nothing other than a straightforward implementation of the ideas in the paper, so...

The example images are not in the public domain.

Just to double check, but are you sure HDR cannot be done simply using numpy operations? I.e., do we absolutely need for-loops?

@paalge CC is Creative Commons, not public domain. Any government datasets (e.g., the NASA images mentioned above) are public domain, though. You are right: it takes significant effort and dedication...

Thanks, both of you, for moving this forward.