Converting digital images to B&W is a bit like the Swiss fondue recipe: everyone has a different version and each person is convinced to have the best. I previously blogged about the subject with a “blind comparison” between different methods – with the comparison’s result.
Now during the recent Libre Graphics Meeting, there was a short talk (I don’t remember the speaker) about “What’s new in GIMP 2.6?” and one of the hotness is the very explicitly named c2g feature, conveniently buried in the GEGL operations (sub-)menu.
And guess what? c2g is a black and white converter… maybe the ultimate one.
The other good thing about conferences is that you get to actually meet people, so I had a chance to listen to some more explanations from Øyvind Kolås (aka Pippin), c2g’s author. I say listen since I was the least technical guy around the table so there was not much talking on my side; which is also why I hope this blog post is accurate – please correct me if someone knows better.
The idea behind c2g is to convert each pixel to B&W using an optimal white balance: the black and white points are calculated for each pixel according to its surrounding, giving the closer pixels more weight (and following a gaussian curve). And basically this mimics the way our human eye perceives the image – how we perceive a color (or grey nuance) depends not only on its value, but on its environment.
You probably came across this optical illusion: both greys in the middle are of the same density (value) but one is perceived clearer than the other because it is surrounded by a darker tone than the other one.Now if I process the image with c2g, here is what I get:See the idea? The absolute value of the greys in the center are “corrected” to “fit” our perception. By the way, this is with every c2g value left as default, except samples set to 10 – resulting in a longer processing time, but less noise.
The best way I would describe what c2g does is masking done automagically right for maximum contrast. For the purpose of illustration (rather than for artistic purpose) take this photo:
And this is what you get with c2g – default parameters except a radius set to 280 for a more dramatic effect on a small image:
Just for comparison’s sake here is the Luminosity (from HSL):
Øyvind has a comparison between several more B&W conversion methods (including c2g, obviously).
What do you think? If c2g is not the ultimate B&W converter, it is darn close to it, if you ask me. By the way, here is the comparison’s image processed with c2g: