The Intel TBB library indeed provides a flow graph since v4.0—that's interesting, thanks!
On 20 December 2014 at 00:58, Luc-Eric Rousseau <[email protected]> wrote: > > I always talk about stuff from my programmer-colored glasses. I say > there isn't a lot of value in the fxtree code, or older > compositing/paint code in general, because today anyone can download > an image library, openimageio, and write you own fxtree-like > compositor within a few days. I think the Intel libraries might have > you covered with all the threading and graph evaluation. Want to > write a paint app. you can look at the gimp source code, use the Cairo > library for vector graphics, etc. I tool the image lib from illusion, > then wrote my own operator evaluation code, if I can do it it's not > complicated. But I think there are open source libs for that too. > > Now doing correct floating point compositing, multi-channel workflows, > tile based/memory management handling, that's a whole other ballgame. > Then you go in 3d space, it's yet another ballgame. Nuke is another > ballgame. > The basics are always easy, and eventually the bar moves up and those > basics become commoditized, which is the word I guess I should have > used rather than "worthless". That's why there are so many text > editors these days, while it was a programming feat to make a text > editor in the early days but today it's a well known problem. > > On Fri, Dec 19, 2014 at 2:32 AM, Matt Lind <[email protected]> wrote: > > In your opinion, what would've needed to happen with the FXTree to make > it a > > 'real contender'. > > > > Matt > > > > > -- Christopher Crouzet *http://christophercrouzet.com* <http://christophercrouzet.com>
