Photography Stack Exchange is a question and answer site for professional, enthusiast and amateur photographers. It only takes a minute to sign up.
Sign up to join this community
In a previous question about soft lenses, it was suggested that much of the current practices regarding soft focus involve selectively soft focusing certain parts of the image and mattdm suggests its not as simple as a little blur.
What digital methods exists to give a selective approximate soft focus to photographs?
(I'm open to answers in Lightroom, Photoshop, and the GIMP.)
There are two separate parts to this; first, how to replicate the effect of a soft focus lens in software, and second, how to do what Jay notes in your other question is common practice in modern portrait photography: selective softening.
There are several techniques for this. First, here's a sample image (a photo of my daughter):
And here's that image with a simple gaussian blur.
Following the suggestion in the comment below, I've replaced the original examples with versions where I turned the sliders up a little higher than I had originally, so the effect is more apparent. It could be even more dramatic, but this is as un-timid as I could make myself go.
I found a pretty decent tutorial for one approach here: Soft-Focus Emulation in Photoshop. In summary, the process is:
I found this technique to be a bit tedious in Gimp, due to the lack of adjustment layers; you end up recreating and destroying the middle layer over and over again as you work to get the right effect. Anyway, a sample result:
This particular approach is often called the Orton Effect, after a film technique used by photographer Michael Orton.
Another common digital approach:
This is much simpler, and produces nice looking results with many images, although I think it's less technically similar to what a soft focus lens does. Result:
And finally, a third (and simple) technique based on Stan Rogers's answer. The steps here are based on Gimp; the actual details should be similar in Photoshop.
Result:
For comparison, I think it's handy to open each of these in its own tab and flip back and forth: orig blur 1 2 3. I think the last approach works best, but the others would have their uses in some images.
For the second overall part to the question, I think usually a more simple approach is used, where one simply makes a blur layer and then erases the blur to varying degrees in areas where more sharpness is desired. (With no curves adjustment layer.) But that's not an area where I've done much work, since it doesn't fit with my own style and preferences.
I've tried it a little bit, and using one of the above approaches in combination with selective erasing can yield nicer final images than a simple blur layer. It's particularly important if you use the first technique, which can cause a lot of unwanted blur. If you have an image where you really want to post-process for this effect and have it look good, careful selective work is probably the thing to do.
This isn't really my thing. I'm more comfortable with the play-with-a-soft-lens-see-what-it-gets me approach. (For example, I enjoy attempting portraits with a zone plate, even if I haven't really gotten great results yet.)
It would probably help to start with a definition of "soft focus". "Soft focus" is not just another way of saying "out of focus", at least not in the photographic vocabulary.
The classic soft focus lens is the Rodenstock Imagon. It was a lens that had a relatively large amount of spherical aberration, which means that the center of the lens doesn't focus at the same plane as its outside edges. Used without its signature "sink strainer" diaphragm, it was soft to the point that it was unusable. The diaphragms had a relatively large hole in the center, which, by itself, produced a relatively sharp picture -- not tack-sharp by today's standards by any means, but much better than the lens used without the diaphragm.
Around that center hole were a series of smaller holes that allowed light to pass through the periphery of the lens. Light from the periphery would be out-of-focus relative to the light from the center, so there was an in-focus image with an out-of-focus image overlaying it. (The size of the smaller openings was controllable, so you could adjust the degree of softness.) The Canon 135/2.8 SF lens uses a different method to achieve a similar effect, but it's the spherical aberration (the fact that the lens can be simultaneously in and out of focus) that does the trick. That means that simply blurring the image won't work -- at the very least, it needs to be a combination of a sharp image and one that's been blurred somewhat. More than one blurred image with different amounts/types of blur can work better.
Beyond that, you need to keep in mind that shadows (no light) from the out-of-focus part of the image can't contribute anything to the image. In the end, a classic soft focus image will be characterized by an overall loss of detail in bright areas, and a sort of etherial "halo" that bleeds from the highlight areas into the shadows. You can approximate that with a blending mode that only lighten the underlying sharp image, and that has a blend range that ignores the darkest tones in the blurred image (and preferably one that can graduate this cut-off).
Most of the higher-end image editors I've used can do this (with a greater or lesser degree of difficulty). Frankly, it's a heck of a lot easier to use a plugin. I know that Topaz Labs has a couple of plugins that will do it, and I'm pretty sure that NIK Color Efex can do it as well. No doubt Alien Skin and others can supply a plugin for Photoshop that will do the trick, and there are a number of plugins for the GIMP that do classic soft focus effects (I haven't evaluated any of them, so I can't recommend one to you).
One strategy I have used is to make a layer over the original, do a Gaussian blur, and set it to 50-70% opacity.
I usually do this after cutting out the eyes and lips (when photographing people--the usual application for soft focus), so they retain a sharp focus.
I have no idea if this is considered best practice.
I implemented an attempt to simulate it in the way Matt suggested (method 3): blur with a certain window size, blend it with a alpha depending on the brightness (of the blurred layer, not the original) - though not linearly, but "gamma adjusted" by an "effect amount" setting, and applying it only when the result is brighter (Max(I1,I2)).
In the current form it "underdoes it", but note also that the brightness is not true due to clipping, so the white and bright greens are actually brighter and should account for more blur, as you see on the comparison below: (Left: normal lens. Middle: simulation Right: softfocus lens. From wikipedia http://en.wikipedia.org/wiki/Soft_focus )
Photoshop (as of about CS3) actually has a filter called "Lens Blur" which simulates bokeh. You can simply mask your image how you want it to be applied and run the filter.
It's actually kind of nice in that you can set the number of blades and many other properties. Adobe.com has a tutorial
