Re: [Gimp-developer] New Image/Color Management option

[Øyvind Kolås <pippin gimp org>]

On Sun, Jun 5, 2016 at 5:14 PM, Gez <listas ohweb com ar> wrote:
> > There is also flipping to/from formats with alpha. Pre-multiplied and
> > non-premultiplied alpha. As well as flips to/from higher and lower
> > bit-depths as well as to/from CIE Lab. By necessity of the operations
> > involved, working on linear data is another one of these constraints
> > that should be fulfilled. Whenever possible the developer/designer of
> > image processing algorithms should be burdened with reducing the
> > number of parameters, as well as sticking with good defaults - making
> > efficient work-flows possible. If you continue calling it "babl
> > flips"
> > I will start calling your non-flipping-babl a lobotomized babl - you
> > could also collapse "RaGaBaA float" into "RGBA float" along with
> > "R'G'B'A float" to make babl even less capable of providing internal
> > color / pixel-representation management for GEGL and thus GIMP and
> > other applications using GEGL. Before scaling or blurring an image a
> > user would then have to run a pre-multiply filter and after-wards
> > un-premultiply filter.
>
> The fully automatic choice of alpha association is not always the best
> way to go.
> Sure, there are many documented cases where you know exactly what's the
> most adequate association, but then there are also cases (which are not
> corner cases at all) that completely fall apart with an automatic
> selection.
> For instance, take an EXR file with luminescent transparent pixels.
> That's a perfectly valid case that is used extensively in VFX to
> composite fire, light effects and any kind of emissive phenomena that
> produce light but don't occlude light from the background.
> It's important to note that what I mention is not a hack used by
> artists at all. It's a valid alpha compositing situation described both
> in the original porter-duff paper and the EXR specs.
> If you feed that data to a program that decides a strategy of alpha
> association, the most probable outcome is that the information in
> pixels with zero alpha is discarded.
> There is an infamous thread at Adobe Forums that had the most renowned
> imagers ranting about how Photoshop was screwing image data because
> programmers decided for users and made assumptions.
>
> That's why it's important to know what artists will do with your
> software before making such assumptions.
> I'm not saying that every single option in a graphics program should
> come with a toggle, but you can't choose for users if you don't know
> what they need.

babl being able to deal with pre-multipled/associated and
non-premultiplied / non-associated alpha/transparency and know if a
buffer contains one or the other is a neccesary for being able to make
the choice of doing things automatically or not in GEGL/GIMP ; remove
this capabilitity from babl - like removing the capability to request
RGB data in a linear variant of the used RGB color space, removes the
ability to deal with thing automatically and requires full manual
control by the operator.

/pippin - digital media artist and toolsmith