Good, because over the weekend I felt like posting right here. :-) Actually, I have not dig too deep in GEGL, but for the Graphs idea. But what I want to start a thread here about is the following: some of you might remember that I've talked on the Gimp lists a couple of times of what I've called a "Custom Layer Combination Mode". The idea is that one could be able to represent the way pixels in two (or just one) layers are combined together to be shown on screen, by a set of simple mathematical expressions. WIth the simpler of them all, one could just as easily re-create GIMP's existing modes: normal, addition, darken only (with an "if" operator like C's "? :" ), grain merge. etc... More elaborate expressions could come up with real itneresting results, not to talk about starightforward results, like fine-tunning grain merge. But, why did I brought this subject here? Because I actually wrote some code to do the above. While I did not do a compiler to go from human readable C like expressions to one think fast to compute, I wrote a combination Engine that would interpret a Stack with the operators, and thus be able to perform any number of __generic__ computations upon combining 2 bitmaps. In my code, I used a 16 bit fixed point computing scheme to do the calculations. GEGL will need other models for operating with other types of image than 8bpp. My proposition is that an engine similar to mine be available in GEGL, so that a node could pick such generic custom expressions - in fact, a mini programming language - and use it to combine two or more images into a third. So what? Anyone want to take a look in my code? (last I fiddled with it, we were on GIMP 1.3.14 yet). If there is anyone interested, I could mail the text,m or maybe, we could open a GEGL bug report and follow this thread there. Regards, JS -><- On Monday 08 March 2004 12:52, Henrik Brix Andersen wrote: > Seems to be working :) > > On Mon, 2004-03-08 at 15:51, Daniel Rogers wrote: > > this is only a test. > > > > you know, to see if this list works. . . > > > > sent Monday March 8, 6:50 am PDT > > > > -- > > Dan
