Jonathan Behrens

> Also a short remark on PixelType, can we avoid mixing the color interpretation with in-memory layout of pixel and channel? And then note that this greatly increases the complexity...

> > The problem with using ViewMut directly is that it is parameterized on Pixel, which would prevent DynamicImage from implementing AsPixels. > > Also already partially solved on a...

> I'm not oposed to having blur and rotate etc. on DynamicImage but I do not see the need for a trait to achieve this ergonomically. If DynamicImage was the...

> Implementing the exact same thing on DynamicImage as on all ViewMut commits it to a particular layout and representation or at least requires it to reorganize its data to...

I'd caution that this change could be quite disruptive for downstream software that consumes Blender-produced glTF files without being able to know what scale factor was used. The core problem...

An ImageBuffer derefs into a slice. I think you should just be able to pass that to qcms::Transform::apply(...).

Unfortunately, no. Some of the individual decoding crates provide a bit more information, but we don't have any common abstraction for that (yet) at the level of this crate. You'd...

We'll need to add support for transparently loading CICP profiles before we start writing them back to files. Otherwise we'd end up pretty much just writing "unknown" or "sRGB" regardless...

@lukehsiao blitting operations tend not to be that hard to implement with simple loops, but there are a bunch of variations that people can want. In the example described above,...

What's wrong with looping through individual pixels? [That's literally what imageops::overlay does.](https://docs.rs/image/0.21.0/src/image/imageops/mod.rs.html#127)