When talking about imaging on Linux, you won’t be able to ignore The Gimp for long. Love it or hate it. Actually, Gimp vs Photoshop is a bit like Apple vs PC, Linux vs Windows, Canon vs Nikon (I am Pentax, btw) or ManU vs ManCity: lots of whining, rants, trolling and flaming. So let’s have a look at The Gimp with a photographer’s eye.
The Gimp (web, wiki) has been one of the first widely distributed, successful open source program; it has been part of any Linux distribution for as long as one can remember and it still is. Development has had its ups and down and is now close to a 2.4 release, which we’ll use as a basis for this entry. I don’t intend to do a detailed report, just to highlight a few dot points from a photographer’s point of view.
One of the most controversial facet of The Gimp is its interface: the programs opens several window when used, one for your picture, one for the tools, one for the layers… I rather like that since it gives more freedom to put your picture on one screen and your tools on the other one and since we’re talking about Linux, you can devote one virtual screen to The Gimp. Mind you, I never extensively worked with Photoshop since:
- I am on Linux 🙂
- I’d rather buy a new lens than a box of software…
The 2.4 version brings quite a few new / nifty features, like a great way to select / crop rectangular regions (flexible and easy), a nice channel mixer to produce black and white images, color management (more on that later) , an enhanced Text engine and a nice preview of image rotation.
There has also been a reorganisation in the menus that make them more logical even if not perfect…
Not quite there yet features
The Gimp supports layers but no effect layers. So you cannot go back to your “brightness and contrast” settings and touch it up later. You have to use the “undo”, which is no way as flexible.
The Gimp also supports a lot of scripts that allow you to do about anything like drop shadows, drawing flames or giving a “glow” look to your image. Few are really useful for photography, except (maybe) red-eyes removal, unsharp mask (this one definitely), blurs (for masks and selections). There is also a Noise Reduction dialog (but not as powerful as NoiseNinja or Noiseware) and that’s about the ones I use.
Where the deal falls through
The problem with The Gimp is that it only processes images with 8bits / channel. So you have 256 nuances of Red Green Blue or (in case of grayscale) 256 nuances from black to white. This is roughly as much nuance our human eye can distinguish so it is fine in theory, but in practice, when you start correcting your image with curves, levels (you name it), you end up with 230, 200 or so levels and you have visible degradation in your gradients (skies…). It is even worse with black and white.
So that makes The Gimp a “no-no” for any serious photographic work under Linux. I only use it for a “quick and dirty” touch up of a image just to see how it would look when croped and touched up, but any “serious processing” (for printing or archiving) is not done in The Gimp.
There is an ambitious project called GEGL (web, wiki) which is an image manipulation library. The idea is that you apply operations (say brightness/contrast) to your base image and get the final output either on screen or saved in a file. The beauty of it is that you can modify, add or suppress operations (think “effect layer”) and that your base image is not changed (think “Nikon capture”). And this library works internally in 32bits floating point (think “millions of nuances in Red Green Blue”). That library is going to be at the heart of the next iteration of The Gimp (2.6, but I would bump it to 3.0 🙂 ).
There is also a big effort going on in a rethink of the user interface (web) to be implemented in the next iteration, so there is much to look forward to.
The Gimp has great perspectives with GEGL (overcoming its 8bits limitation) and a UI redesign. However, remember that Gimp is an open source project with people working in it “when they have time”. So this can mean that the next iteration is in a quite distant future.
I think The Gimp has also a problem defining itself: is it a tool for photographers, for web designers, for icon designers? I would say that quite a bit of the confusion regarding the UI is due to that. There is no way to have an easy fitting-to-all interface if the software behind is complex and offers lots of possibilities (and powerful photographic software is quite complex, although not as much as, say, 3D modelling).
So good things to look forward to, with a grain (or two) of salt…