Eevee

- [ ] There are some TODOs and obvious broken things - [ ] More docs probably - [ ] Extract metadata from the original image on read

A dev has told me e's doing this: ``` def unanimate(im): if len(im) == 1: return im ret = Image() ret.append(im[0]) return ret ``` Yikes.

I don't know what else I had planned here, but they could use a once-over. Also a bunch of tests. A nice bar to reach would be parity with Go's...

This would be neat. Basically implement `convert`'s various custom-pipeline things as a Python API. So something like: ``` def filter(pixel): if pixel.position.x > pixel.position.y: pixel.color = black else: pixel.color =...

Currently no way to inspect or change a single pixel. Better fix that. Keep in mind that IM's API is intended for batch work: it's built around requesting a rectangular...

Taking some influence from the Magick++ API, though only so much. Current concerns include that I seem to be breaking HDR; colors outside `[0, 1]` get trimmed. Will this interfere...

Should they be iterable, or treated like enums (#5, #6), or just be a big cluster of incidental special parse cases?

http://www.python.org/dev/peps/pep-0368/ It defines a (surprisingly in-depth) API for a standard-library imaging module, with a lot of similarities to PIL. It was never implemented, but never rejected either; it just doesn't...

If nothing else it would be good for forcing me to use and flesh out the core of the library.

Would be neat to be able to import/export a whole frame as an array of w × h × channel. Also see what PIL does.