Focus Magic?

[davidedric]

Prompted by a recent post from Richard Lundberg, in this thread: Post Processing - How to Know When to Stop?

I had always understood that if an image was out of focus, that was it. And that sharpening techniques such as USM are not designed for, and don't help with it.

I read and saw the results from Richard's recent post, and was surprised and impressed. Not that I would expect to need Focus Magic regularly, I do try to get it right in camera, but it looks as if it might just save an important image.

For anyone out there who uses it, any comments?

As a supplementary, if you do use it, where does it fit in your workflow?

Regards,

Dave

[Shadowman]

Dave,

It depends on how I use it. As a plug-in it usually follows ACR, if I use it as a stand alone, I have to convert to a jpeg before using. There shouldn't be a need to convert, but sometimes I like to experiment with the other filters (defocus, incr resolution, motion blur) that are available. Motion blur and Fix out of focus are available in the plug-in, so I typically use that method.

[DanK]

I would find it very useful to have more discussion of Focus Magic, ideally with links to full-size images, or at least much larger images, that would allow serious evaluation of the program's effects. I tried it once a year or two ago, on an image that was actually out of focus rather than just not optimally sharp, but I did not get very useful results. That may well have been novice-user ineptness on my part. However, it is still sitting on my computer.

[Shadowman]

I would find it very useful to have more discussion of Focus Magic, ideally with links to full-size images, or at least much larger images, that would allow serious evaluation of the program's effects. I tried it once a year or two ago, on an image that was actually out of focus rather than just not optimally sharp, but I did not get very useful results. That may well have been novice-user ineptness on my part. However, it is still sitting on my computer.

I think the best place to start is with the program itself and its claims. There are tutorials on how the program works and comparisons between current existing products. In the end, it depends on your idea of what acceptable sharpness actually is. I find that the program works well with some images but not all and it depends on the initial level of sharpness.

http://www.focusmagic.com/

[NorthernFocus]

I would find it very useful to have more discussion of Focus Magic, ideally with links to full-size images, or at least much larger images, that would allow serious evaluation of the program's effects. I tried it once a year or two ago, on an image that was actually out of focus rather than just not optimally sharp, but I did not get very useful results. That may well have been novice-user ineptness on my part. However, it is still sitting on my computer.

I also have a copy of it and don't use it at all. I've not yet seen a case where OOF images can truly be corrected by Focus Magic or any other means. Sure you may be able to correct a full frame image well enough to end up with something that looks sharp when reduced to web resolution. But if an image isn't properly focused during capture then you are pretty much stuck with those limitations.

[DanK]

I think the best place to start is with the program itself and its claims.

John,

I have done that. However, they are not likely to publicize the program's weaknesses and limitations. I would rather hear the experience of users who are not interested in selling it, and I'd like to see what they have accomplished. My own meager experience was similar to that of Alaska Dan, but that could well be that I just didn't know how to use the product properly.

I also agree with Alaska Dan that looking at an image at web dimensions is not enough. It will show the worst flaws, but you can have an image that looks great on the web but looks awful in even a modest-sized print. I had a shot of a morning glory a few years ago that I handled very poorly in camera. I forget now what I did wrong, but it was serious. I was able to produce an image for the web that got me a lot of positive comments, but when I printed it at 8 x 10 (roughly A4), I had to throw it out.

(Boston) Dan

[IzzieK]

I also have Focus Magic and after using once or twice without satisfactory results, it's been sitting on my hard drive, barely used.

[Shadowman]

John,

I have done that. However, they are not likely to publicize the program's weaknesses and limitations. I would rather hear the experience of users who are not interested in selling it, and I'd like to see what they have accomplished. My own meager experience was similar to that of Alaska Dan, but that could well be that I just didn't know how to use the product properly.

I also agree with Alaska Dan that looking at an image at web dimensions is not enough. It will show the worst flaws, but you can have an image that looks great on the web but looks awful in even a modest-sized print. I had a shot of a morning glory a few years ago that I handled very poorly in camera. I forget now what I did wrong, but it was serious. I was able to produce an image for the web that got me a lot of positive comments, but when I printed it at 8 x 10 (roughly A4), I had to throw it out.

(Boston) Dan

Dan,

My point is look at the claims and judge the results for yourself. You can download a trial if you are really interested and although one user might find the performance amazing, it really boils down to what you perceive as improvement. I've tried a lot of plug-ins, some work some don't. Regarding Focus Magic, if you have an image that is reasonably sharp, that has a lot of patterns or text in the image and the degree of sharpness of those patterns or text are important to you then Focus Magic works wonderfully.

[NorthernFocus]

....I was able to produce an image for the web that got me a lot of positive comments, but when I printed it at 8 x 10 (roughly A4), I had to throw it out.

(Boston) Dan

Focus is really THE only real killer of digital images. I have an image of a sea otter taken in poor light, terrible white balance, and heavily cropped. Sea otter shots are all about the whiskers which of course require good detail. The post cropped image is only about 2.5MP from a 12MP D300 frame. It is pretty noisy (D300 is not a good low light performer). I had little hope from a technical standpoint but the content was good AND the focus was spot on so I spent several hours trying to recover the image with NR software, color enhancement, etc. It looked OK and printed well at 8x16 so I then used OnOne software's resizing program to create a 12x24 in. 300ppi version. I printed it on metal which is very unforgiving. The image is beautiful and the detail holds up under close inspection. I've sold several copies of it.

On the other hand, I have another full frame 16MP image of an otter and pup taken with a D7000 that is JUST not quite perfectly focused. The eyes look OK but the whiskers are JUST a bit fuzzy. But it is well exposed, low noise, good color, etc. and the content is excellent so I attempted to recover it. I used every sharpening technique I could find, various sharpening software programs, etc., and the largest acceptable print I can produce with it is 8x16 at 300ppi. I did enlarge it to 12x24 and printed it but it is basically a throw away at that size. It looks fine on canvas up to 16x32.

Moral of the story, OOF = SOL.

[Richard Lundberg]

Focus Magic OOF mode is a straightforward deconvolution that assumes a standard blur transfer function. It's just math, folks. They could improve it by further refining the blur function for prime lens and zoom modes. You can move the selection area and change the blur width while looking at the preview before and after. If that doesn' t help, you have really screwed up.
I use it, I love it. YMMV.

[DanK]

Focus Magic OOF mode is a straightforward deconvolution that assumes a standard blur transfer function. It's just math, folks. They could improve it by further refining the blur function for prime lens and zoom modes. You can move the selection area and change the blur width while looking at the preview before and after. If that doesn' t help, you have really screwed up.
I use it, I love it. YMMV.

Richard,

It's helpful to know that you have found it useful. That will spur me to try it again. But really, it's all math--this is digital photography, after all--and that fact doesn't mean that it will work well. From what little I have read, the problem is that the transfer function is ill-posed, so it is not clear what deconvolution function will work. Am I missing something?

I found tutorial 6 on the Focus Magic website useful. It deals with faces that are quite severely out of focus. It shows substantial but incomplete improvement, at the costs of some artifacts on the skin surfaces. My guess is that the results would have been far better if the image had been less out of focus.

In any case, I am going to go back to it again. You are not the only person who is enthusiastic. There are a fair number of people who have reported very good results with problems far less severe than those in tutorial 6. One person uses it routinely to offset a fairly severe anti-aliasing filter in his camera.

Dan

[NorthernFocus]

...It's just math, folks....

So does this mean we're all wasting a lot of money on expensive equipment? We should just shoot crap gear, buy a little software, and let the computer spit out world class photos?

[Donald]

So does this mean we're all wasting a lot of money on expensive equipment? We should just shoot crap gear, buy a little software, and let the computer spit out world class photos?

D...., that sounds a good idea. Might have meant I got more keepers from my session out in the fog this morning!

[Richard Lundberg]

It's not that it's ill-posed, whatever that means. It's that they use a generalized blur transfer function to deconvolve. For an extra hundred bucks per download they could probably tailor the blur function to classes or makes. You lens experts can mull that. I like it because it's a useful practical application of Fourier transforms that I have been involved with for all my engineering life, especially in anti submarine warfare sonar processing where we used FFT extensively.

[NorthernFocus]

It's not that it's ill-posed, whatever that means. It's that they use a generalized blur transfer function to deconvolve. For an extra hundred bucks per download they could probably tailor the blur function to classes or makes. You lens experts can mull that. I like it because it's a useful practical application of Fourier transforms that I have been involved with for all my engineering life, especially in anti submarine warfare sonar processing where we used FFT extensively.

How does it deal with the variability in the blur radius associated with distance from the focal plane? Can it handle that?

[Richard Lundberg]

I believe it does the deconvolution based on the blur width where the selection frame is placed. This may be why some people have trouble with it.

[NorthernFocus]

Well you convinced me to reload it and give it another try. Particularly since I've got a couple of shots like the one I mentioned that are JUST far enough OOF to be unacceptable.

[DanK]

It's not that it's ill-posed, whatever that means. It's that they use a generalized blur transfer function to deconvolve.

I'm not a mathematician, so I probably shouldn't have used the technical term, but an ill-posed problem is one that does not have a unique solution. In tutorial 6, they concentrate specifically on varying the blur width, but it doesn't entirely work. This is what you would expect if the deconvolution function is only an approximation of the unknown inverse of the blur transform function for a specific image.

Anyway, the real issue is the range of situations for which this works well. I'll be trying it again, thanks to Richard's comments, as will Alaska Dan. Perhaps others too can chime in with examples of when it works well and when not.

Dan

[Colin Southern]

The "focus magic" I like is where I point the camera and 1/2 press the release.

It focuses - like magic

Seriously folks, to me, a shot is either in focus or it isn't; I'm sure software like Focus Magic and Adobe's anti-shake do the best they can, but I've yet to see an example where they've turned a badly focused image into anything acceptable. If the image that Dave was impressed by (being Richards rework of the bird in the "PP - How to know when to stop" thread) - then sorry folks, if that's impressive then those people need new glasses; the original was hopelessly out of focus and the reworded image is also hopelessly out of focus, just looking a bit different.

Sorry Richard - I'm not "getting at you" - I just think folks need a reality check here. What I'm seeing is nothing more than a gimmick.

PS: I ran it through Photoshop's anti-shake filter just for comparison:



My suggestion is that if you get an OOF shot, trash it, and go take a better one. That's what I do. Trying to fix it in PP is like trying to make a silk purse out of a sow's ear.

[Richard Lundberg]

If what you uploaded is the antishake filter it didn't do anything worthwhile.
As DanK said "This is what you would expect if the deconvolution function is only an approximation of the unknown inverse of the blur transform function for a specific image." Exactly. Focus Magic is not sharpening. It is deconvolution of the blur function and theoretically should use the exact inverse transfer function, but it would be necessary to code a table for every popular lens. This probably would not make that much difference for ordinary lenses. Sharpening is basically edge detection and differentiation. Sharpening for output should still be done in moderation. Not to belabor the subject, but it is a valid, mathematically sound, method of focus recovery. I'm not sure why Colin didn't see any benefit, but it saves a shot for someone it's well worth it. It's another bit for your drill, or another vise grip to use as a hammer if you're like me.