Out of focus photo - Is this fixable?

KMW

Hi,
I haven't posted in a long time. People here have been very helpful in the past. I hope someone can help me repair this photo.

This was from a recent family session. It was not one of my easiest sessions. It was a sunny, cold, windy day and the little boy was extremely hyperactive. This is one of the few photos where they're all looking at the camera and the boy was not trying to escape. I was able to capture some great individual and parent/child photos but not many group photos.

The image was originally shot in RAW. I converted to JPEG to upload to dgrin. I've tried sharpening in Photoshop Elements before and it still does not look right. This image is untouched.

Please help!

Thanks,
Kath

http://www.willoughbyphotography.com/photos/661433047_V8H6M-M.jpg

http://www.willoughbyphotography.com/photos/661433047_V8H6M-X3.jpg

Wil Davis

Nice pics!

Here's very basic USM (Unsharp Masking) with PhotoShop (80%, Radius=2 pix, Thres=1)




…and this one has a colour tweak to warm it up a bit, and USM slightly less (70%):



Nice family! I think they look pretty happy and and not at all cold!

- Wil

BinaryFx

Kath, I think that the reduction to the web posting size has removed (averaged out) most of the focus issues! Keep this in mind for the print size, if you reduce the original down to a smaller pixel size for printing, then some of your work may be reduced (print size will have more pixels than for a website, so the averaging process may not help as much).

You may need to crop a smaller section at full size and post that for comments. In the full size photo, is this problem focus blur (like gaussian blur) or directional/movement blur in one specific direction (either from the camera or the subject)?


Regards,

Stephen Marsh
http://members.ozemail.com.au/~binaryfx/
http://prepression.blogspot.com/

Wil Davis

Stephen, I think that close examination of the X3 file which Kath posted will show that it's the former rather than the latter; I didn't have the .RAW file to work with, but certainly using the X3 file it was quite obvious. By way of comparison, here is a composite of Kath's posted original, and my colour corrected and sharpened version:



- Wil

BinaryFx

Will, I think that I have very different expectations when the title of the thread is "out of focus"!

Yes, the posted images are not "tack sharp", but were not as bad as what I was expecting from the topic title.

Apart from the focus, I would certainly luminosity blend the green channel into the composite to help with the faces being washed out.


Stephen Marsh
http://members.ozemail.com.au/~binaryfx/
http://prepression.blogspot.com/

Wil Davis

BinaryFx wrote:

Will, I think that I have very different expectations when the title of the thread is "out of focus"!

Yes, the posted images are not "tack sharp", but were not as bad as what I was expecting from the topic title.


Stephen Marsh
http://members.ozemail.com.au/~binaryfx/
http://prepression.blogspot.com/

You make a valid point about printing, and to be honest I hadn't thought that far ahead; I was just trying to help her out considering what she was working with. When I first saw the thread, my immediate reaction was if a picture is out of focus to begin with, then it is usually a total loss, but like you I was surprised to find that it was (IMNSHO) quite recoverable (well, good enough for the web at least).



- Wil

KMW

Thank you for your replies. The sharpened versions that Wil posted look great but I am concerned about print quality. If I do present this photo and my client orders an 8x10 or 11x14 will it look good in print?

Stephen, How do I blend in the green luminosity channel? Would it help if I shared the RAW file?


Thanks,
Kath

BinaryFx

KMW wrote:

Thank you for your replies. The sharpened versions that Wil posted look great but I am concerned about print quality. If I do present this photo and my client orders an 8x10 or 11x14 will it look good in print?

Stephen, How do I blend in the green luminosity channel? Would it help if I shared the RAW file?


Thanks,
Kath

Kath, simply put in text instructions:

In Photoshop -

1) Go to the channels palette and click on the green channel
2) Select all and copy the green data, then click the full colour composite channel again
3) Return to the layers palette and paste the channel data as a new layer
4) Set the layer to luminosity mode to retain the lower layers colour, making use of the contrast/luminosity of the upper layer

You may choose to only mask this edit into the facial area and not to affect the whole image...or you may not care that the whole image changes if you like the effect on the faces and other key areas.

I have recorded a quick 30 second flash video screencast of the process (5.5mb). The video is a bit different to the text above, I simply used channel mixer to do the same in a non destructive format that does not use the memory overheads of channels/layers!

Direct Link: Green Luminosity Blend
My Video Tutorial Folder: Video_Tutorials_Stephen_Marsh

As for print quality/size...can you give me the pixel dimensions of the final cropped high resolution image? How will you be printing this, inkjet or a true photographic output on say a Fuji Frontier or similar? On a less than perfect shot, in many cases, the smaller the output the better...


Stephen Marsh
http://members.ozemail.com.au/~binaryfx/
http://prepression.blogspot.com/

KMW

Thank you so much for the video tutorial. Unfortunately, I do not have a channel mixer. I am using Photoshop Elements 5. Is there another way to achieve this without the mixer? I do have the luminosity option.

I was not planning on cropping the photo, so the file will keep its original dimensions (3872x2592 240ppi). I use a professional print lab and I am not sure if they use Fuji Frontier. What do you mean by "the smaller the output the better?"

Thanks, Kath

BinaryFx wrote:

Kath, simply put in text instructions:

In Photoshop -

1) Go to the channels palette and click on the green channel
2) Select all and copy the green data, then click the full colour composite channel again
3) Return to the layers palette and paste the channel data as a new layer
4) Set the layer to luminosity mode to retain the lower layers colour, making use of the contrast/luminosity of the upper layer

You may choose to only mask this edit into the facial area and not to affect the whole image...or you may not care that the whole image changes if you like the effect on the faces and other key areas.

I have recorded a quick 30 second flash video screencast of the process (5.5mb). The video is a bit different to the text above, I simply used channel mixer to do the same in a non destructive format that does not use the memory overheads of channels/layers!

Direct Link: Green Luminosity Blend
My Video Tutorial Folder: Video_Tutorials_Stephen_Marsh

As for print quality/size...can you give me the pixel dimensions of the final cropped high resolution image? How will you be printing this, inkjet or a true photographic output on say a Fuji Frontier or similar? On a less than perfect shot, in many cases, the smaller the output the better...


Stephen Marsh
http://members.ozemail.com.au/~binaryfx/
http://prepression.blogspot.com/

BinaryFx

KMW wrote:

Thank you so much for the video tutorial. Unfortunately, I do not have a channel mixer. I am using Photoshop Elements 5. Is there another way to achieve this without the mixer? I do have the luminosity option.

Sorry Kath, I am one of those Photoshop snobs that does not consider that others would use anything else!

One can add the channel mixer option to PS Elements with a free add-on which can be found via Google.

Without the free add on, you would have to follow the previous text instructions, rather than the video. This presumes that you can copy the green channel data and blend it as a layer in luminosity mode (not using PSE, I am not sure if you can do this without the free add-ons to Elements).

If you have good raw capture conversion software, you should be able to render the raw data into a much better version, improving contrast and adding some "depth" to the skintone so that it is not so washed out etc. Prints often suffer from perceived lesser contrast more than the monitor, so it often pays to build more contrast into the print. BTW, adding more contrast will also help with the illusion of better sharpness/focus/detail.

I was not planning on cropping the photo, so the file will keep its original dimensions (3872x2592 240ppi). I use a professional print lab and I am not sure if they use Fuji Frontier. What do you mean by "the smaller the output the better?"

The quoted pixel dimensions do not fit either a 8x10 or 11x14 inch print without cropping (the proportions are not right). Be that as it may, there does appear to be enough pixels in the photo to produce good output, either contone photographic (be it Fuji, Kokak or whatever) or inkjet "stochastic" dots.

What I mean by the smaller the print size the better, is that when an original is less than perfect, then the smaller the reproduction size, such as 6x4 would show less flaws than a larger size.

I have not seen your original, however I think that with some good sharpening you should be able to output the image to a larger size.

Over and above sharpening, there is a technique known as "deconvolution", which is a software attempt to restore sharpness due to lens or motion blur - this is not the same thing as sharpening, however results often look similar. Some commercial deconvolution (blur removal) software are "FocusFixer" and "FocusMagic" to name just two (Google them, don't expect too much though).


Regards,

Stephen Marsh
http://members.ozemail.com.au/~binaryfx/
http://prepression.blogspot.com/

KMW

BinaryFx wrote:

Sorry Kath, I am one of those Photoshop snobs that does not consider that others would use anything else!

One can add the channel mixer option to PS Elements with a free add-on which can be found via Google.

Without the free add on, you would have to follow the previous text instructions, rather than the video. This presumes that you can copy the green channel data and blend it as a layer in luminosity mode (not using PSE, I am not sure if you can do this without the free add-ons to Elements).

If you have good raw capture conversion software, you should be able to render the raw data into a much better version, improving contrast and adding some "depth" to the skintone so that it is not so washed out etc. Prints often suffer from perceived lesser contrast more than the monitor, so it often pays to build more contrast into the print. BTW, adding more contrast will also help with the illusion of better sharpness/focus/detail.



The quoted pixel dimensions do not fit either a 8x10 or 11x14 inch print without cropping (the proportions are not right). Be that as it may, there does appear to be enough pixels in the photo to produce good output, either contone photographic (be it Fuji, Kokak or whatever) or inkjet "stochastic" dots.

What I mean by the smaller the print size the better, is that when an original is less than perfect, then the smaller the reproduction size, such as 6x4 would show less flaws than a larger size.

I have not seen your original, however I think that with some good sharpening you should be able to output the image to a larger size.

Over and above sharpening, there is a technique known as "deconvolution", which is a software attempt to restore sharpness due to lens or motion blur - this is not the same thing as sharpening, however results often look similar. Some commercial deconvolution (blur removal) software are "FocusFixer" and "FocusMagic" to name just two (Google them, don't expect too much though).


Regards,

Stephen Marsh
http://members.ozemail.com.au/~binaryfx/
http://prepression.blogspot.com/

Thank you for your help! I will take your advice and I will do my best to recover this photo. I think the family would like this one.

Thanks again!

Kath

ziggy53

BinaryFx wrote:

...

Over and above sharpening, there is a technique known as "deconvolution", which is a software attempt to restore sharpness due to lens or motion blur - this is not the same thing as sharpening, however results often look similar. Some commercial deconvolution (blur removal) software are "FocusFixer" and "FocusMagic" to name just two (Google them, don't expect too much though).


Regards,

Stephen Marsh
http://members.ozemail.com.au/~binaryfx/
http://prepression.blogspot.com/

RAW Therapee also has deconvolution as an alternate sharpener, and it's pretty capable. The main caveat with deconvolution is that it cannot know more that the 2 dimensional data of the image, meaning that it will sharpen everything rqually. To use it properly you really need to mask the subject you want "deconvolved", and allow the rest (background and foreground elements) to remain untouched. You could also use deconvolution on the entire image and then stack that as a layer on top of the original and then delete around the subject, allowing the background and foreground to show from the untouched image.

Smart Sharpen is a also a simple deconvolution engine, but I like RAW Therapee better.

RAW Therapee is freeware for Windows and Linux platforms and it can also open 16 bit TIF and 8 bit JPG files. (It's not limited to RAW files.) It lacks layer and mask capabilities, but you can use the GIMP if you want something in freeware that has layer and mask capabilities. (GIMP is 8 bit only and sRGB, but otherwise fairly capable.)

http://www.rawtherapee.com/
http://www.gimp.org/

TheSuede

Hi Kath...

As far as I can see, there's very little defocusing in this photo. What there is tough, is quite a lot of subject movement subtly masked by the flash "freeze-motion" effect. There's no EXIF in the pictures, but I'm guessing at 1/80 to 1/160s at F/5.6? With rear-synch fill-flash?
Can you see the "double image" formed in the blurry parts, like the zipper on his jacket, in the littele girls eyes and so on?

This makes the picture VERY hard to "sharpen", as some parts ARE very sharp, just the part that were moving was blurred. And hose blurred parts still have a very sharp overlay picture on them, made by the fill-flash. :-/

Anyway, there's no way around that you need to do quite a lot of manual work on this one, if you want to print big.

I'd say it's perfectly usable up to 6x9, if you want to hold high standards in printing. Here's a small copy of what's doable in PS in 5 minutes. Better quality sample is on the the link here:
http://pici.se/pictures/YBjrzunPB.jpg

KMW

ziggy53 wrote:

RAW Therapee also has deconvolution as an alternate sharpener, and it's pretty capable. The main caveat with deconvolution is that it cannot know more that the 2 dimensional data of the image, meaning that it will sharpen everything rqually. To use it properly you really need to mask the subject you want "deconvolved", and allow the rest (background and foreground elements) to remain untouched. You could also use deconvolution on the entire image and then stack that as a layer on top of the original and then delete around the subject, allowing the background and foreground to show from the untouched image.

Smart Sharpen is a also a simple deconvolution engine, but I like RAW Therapee better.

RAW Therapee is freeware for Windows and Linux platforms and it can also open 16 bit TIF and 8 bit JPG files. (It's not limited to RAW files.) It lacks layer and mask capabilities, but you can use the GIMP if you want something in freeware that has layer and mask capabilities. (GIMP is 8 bit only and sRGB, but otherwise fairly capable.)

http://www.rawtherapee.com/
http://www.gimp.org/

Thank you Ziggy for your advice. I downloaded Raw Therapee but I haven't had a chance to learn how to use it. It looks like a great program. How does this compare to Adobe Camera Raw? It looks like it has more features.

I am also going to take a look at gimp. Is this similar to photoshop? I'm using Photoshop Elements right now. I would like to upgrade to the Professional version but I need to hold off for now. Does anyone use this exclusively for photo editing? This looks like a great program too. I will have take a look at both of these later on this week.

Thanks!
Kath

KMW

TheSuede wrote:

Hi Kath...

As far as I can see, there's very little defocusing in this photo. What there is tough, is quite a lot of subject movement subtly masked by the flash "freeze-motion" effect. There's no EXIF in the pictures, but I'm guessing at 1/80 to 1/160s at F/5.6? With rear-synch fill-flash?
Can you see the "double image" formed in the blurry parts, like the zipper on his jacket, in the littele girls eyes and so on?

This makes the picture VERY hard to "sharpen", as some parts ARE very sharp, just the part that were moving was blurred. And hose blurred parts still have a very sharp overlay picture on them, made by the fill-flash. :-/

Anyway, there's no way around that you need to do quite a lot of manual work on this one, if you want to print big.

I'd say it's perfectly usable up to 6x9, if you want to hold high standards in printing. Here's a small copy of what's doable in PS in 5 minutes. Better quality sample is on the the link here:
http://pici.se/pictures/YBjrzunPB.jpg

Thanks for replying. The picture was shot at 1/80sec @ f4.5. I probably should have used a faster shutter speed. I didn't use rear synch-flash. I used my sb600 TTL-BL standard fill. I do see the little girls eyes seem slightly double and blurry.

The edit you did looks great. I'm going to work on it this week. I'm hoping this will print well if they order an 8x10 or 11x14.

Thanks,
Kath

ziggy53

KMW wrote:

Thank you Ziggy for your advice. I downloaded Raw Therapee but I haven't had a chance to learn how to use it. It looks like a great program. How does this compare to Adobe Camera Raw? It looks like it has more features. ...

RAW Therapee (RT) is both a RAW converter and an image editor. You would compare it to both Adobe Camera Raw and the basic color manipulation of PhotoShop.

KMW wrote:

... I am also going to take a look at gimp. Is this similar to photoshop? I'm using Photoshop Elements right now. I would like to upgrade to the Professional version but I need to hold off for now. Does anyone use this exclusively for photo editing? This looks like a great program too. I will have take a look at both of these later on this week.
...

The GIMP is very close in basic feature set to PhotoShop, but it lacks the ability to process 16 bit image files (it is 8 bit only) and it is limited to sRGB color space. The GIMP has a very different user interface than PhotoShop, but with the use of "GIMPShop" it is reasonably similar in terminology and vague appearance. It has enough functionality that, along with RAW Therapee, I was able to do real work on a remote machine for several years when I was away from home and the owner of that machine (my "lady friend" who lives 3 hours away) was unwilling to purchase and install (much less learn to use) ACR and PhotoShop.

For free software The Gimp is very nice software, but eventually you will probably want PhotoShop just so you can tap into a relatively vast support structure on the Internet built to support PhotoShop. PhotoShop is also considerably faster to use.