“GIMP 2.10’s core will be 100% ported to GEGL, and all of the legacy pixel fiddling API for plug-ins is going to be deprecated.” I’ll honestly admit I have no idea what they’re on about (I can’t know everything about everything), but it appears to be quite a big deal.
Apparently they accidentally did this?
No wonder 2.8 hasn’t been released yet.
It is. Take a look at gegl.org
The whole of the graphics core has been moved into their own graphics core library (GEGL).
The big deal will come if other alternative UI can be written (by third parties?) with a more sensible user interface that is easy to used by beginners, including children. In theory this development should make this easy to do.
That’s nice, can things like web applications, CLI programs and other things use this library? I see a lot of potential in this.
If you want a CLI program for image manipulation, Imagemagick is already really good.
So going out on a limb here what it could mean is it being used possibly in a desktop like GNOME/KDE or maybe even in something like LibreOffice?
Are you familiar with the new single window interface that is in gimp 2.7?
http://www.gimp.org/release-notes/gimp-2.7.html
I’m not sure what could be better than that. I’m also not sure how changing the graphics core in gimp allows for alternative UI’s. GEGL itself has been available and capable for a while.
If I’m not mistaken the GEGL engine also supports OpenCL, which is great news as from what I’ve been reading OpenCL support in the Gallium3D drivers via the Gallium Compute state tracker is looking like it could be ready by the merge window for the fall releases which would allow for at least the Evergreen, Northern Islands and NV50 series GPUs to have support initially.
http://www.x.org/wiki/XorgEVoC/GalliumCompute
I dunno when Intel will have support though.
Many of the next gen ARM GPUs are also supposed to have OpenCL support http://www.arm.com/products/multimedia/mali-graphics-hardware/mali-…
*Almost completely OT*
I hope like hell some company is smart enough to make a tablet/netbook with a detachable keyboard like the Acer Iconia with a 1920×1280 res screen, the 2.5Ghz quad variant of the Cortex A-15 and the Mali T658 GPU. Something like that would have incredible performance and be amazing for all current HD content(have the video player automatically rotate for pixel perfect 1280x720p and 1920x1080p) as well as be very good for gaming and even some actual work as the CPU wouldn’t be too bad, but the GPU is supposed to be slightly more powerful then that of the PS3’s Cell, so OpenCL based multimedia editing apps would perform very well on it.
“Once the core is completely ported, it will be a minor effort to simply ^aEURoeswitch on^aEUR high bit depths and whatever color models we^aEURTMd like to see. Oh, and already now, instead of removing indexed mode (as originally planned), we accidentally promoted indexed images to first class citizens that can be painted on, and even color corrected, just like any other image. The code doing so doesn^aEURTMt even notice because GEGL and Babl transparently handle the pixel conversion magic.”
Making high bit depths available will strengthen the GIMP feature set (for those that need it and for those who don’t, it removes the excuse).
The hooking into use of the GPU, where possible, is very cool indeed – especially now that 10+ megapixel are commonplace.
Being able to drop a block of complex legacy code will help the project to focus elsewhere. I still find myself hoping they will rework the UI.
The biggest strength behind GEGL, other than allowing the program to tap into the GPU more and more if this OpenCL development works well, is to bring non-destructive editing to Gimp (see this for example: http://en.wikipedia.org/wiki/Core_Image for an API supporting non-destructive image editing).
The drive for the implementation of GEGL in Gimp is to support 16 bit color.
This is one of the main “pro” gripes, that photoshop can support 16 bit (and more i think) and gimp only 8 bit.
Many a flaming artist rants on why you need and others one why you don’t need it for certain tasks like webdesign…
Edited 2012-04-18 11:57 UTC
Should perhaps be noted that this is 16-bit ‘per-channel’ and not 16-bit total color space. I can’t say I’ve ever been hampered by this myself but then again I’m probably as far away from a professional photographer/retoucher/etc as you can be and of course having 65536 (I assume?) possible levels beats having only 256 in terms of flexibility and precision.
As for the Photoshop/Gimp gui thing, apart from single window mode I’ve always thought they were extremely similar although obviously not matching 1:1, I tried single window mode in the current Gimp dev version and I found that I prefer the floating window mode, YMMV but either way it’s good that both preferences are supported from now on.
Out of curiousity, how do graphics designers work with 16 bit per channel, when afaik displays and gfx cards (generally) can only show 8 per channel?
At the risk of answering my own question, the only advantage I see is that effects would be more accurate, such as blending and what not. But with a single layer, 16 bit doesn’t make much sense to me.
The ability to tone-map pictures is the biggest advantage for photographers.
For more complex compositions, it’s just as you said, it increases the precision so there is less data loss or aliasing artifacts. Imagine you draw a grandient and then want to apply levels or curves, it would still look good instead of aliased.
It’s not about display. In the end, when a picture is finished, it is generally bumped back down to 8 bit / ch color. The advantage is when working with the image you don’t loose color information.
For instance if you manipulate the color curve in a way the crushes part of the color data, and then come back later and manipulate the color curve again in a way that stretches that color range back out, it will look gross because there is no data left to draw information from, so the computer has to interpolate and guess.
More simply put, if you do the exact same operations and all other variables the same, aside from bit depth the end result of the picture worked on in 8 bit color will look grainy, less detailed and the gradients will look stair-stepped.
Also GEGL supports more than 16bit, it can do 32 bit float color space too. The advantage of GEGL to something like Photoshop in this case is that in photoshop fliters have to be explicitly programmed to support multiple color depths. With GEGL, the filters work is done by the GEGL backend, so filters support all color-depths and color-spaces supported by GEGL.
Blender took a huge boost with the UI refactoring in 2.6
Now gimp will definitely go there to with the new UI (no more floating windows, hooray!) and now high bit depths.
Floating windows are useful for designer having multiple physical monitors. For that reason, that option will stay.