My World: Untitled -- violin in the dark

[Bob Rogers]

It was on a houseboat, after dark. There was only one place to stand to take the photo (because the rest of the band was on my left), and the boat was turning. In the actual photograph, the moon is to her right (i.e., behind her) so I moved it. The only light source is a 150 watt flood, about 20 feet away. The camera is a Canon s95 (compact P&S), so it's a got more than a little motion blur and a lot of noise. I used the drybrush filter in photoshop. I think it makes a nice print at about 5".

Untitled
Bob Rogers​

 

[Doug Kerr]

Hi, Bob,

It was on a houseboat, after dark. There was only one place to stand to take the photo (because the rest of the band was on my left), and the boat was turning. In the actual photograph, the moon is to her right (i.e., behind her) so I moved it. The only light source is a 150 watt flood, about 20 feet away. The camera is a Canon s95 (compact P&S), so it's a got more than a little motion blur and a lot of noise. I used the drybrush filter in photoshop. I think it makes a nice print at about 5".

Untitled
Bob Rogers​

That is curiously effective!

Thanks.

Best regards,

Doug

 

[Asher Kelman]

I like the story and how you set it up! How many images did you get? I'd love to see them all!


Asher

 

[Michael Nagel]

I like this one.

How does the original look like?

The effect has its merits, but I think the original wasn't that bad either. Some time ago I had the S90 and I liked the little camera.

The mood is more imortant than technical brilliance in these situations from my point of view.

Best regards,
Michael

 

[Bob Rogers]

I like the story and how you set it up! How many images did you get? I'd love to see them all!


Asher

Depends on how you define "get" ;-)

I only got one of this scene that I could work with -- all the rest had either way too much blur or poor poses.

Here is what I was working from:

​

This is about a quarter of the original frame, with noise reduction applied.

 

[Asher Kelman]

Depends on how you define "get" ;-)

I only got one of this scene that I could work with -- all the rest had either way too much blur or poor poses.

Here is what I was working from:

​

This is about a quarter of the original frame, with noise reduction applied.

Bob,

In the first post, did you take the moon and its reflection and clone it to the left side to get the picture you first showed?

In any case, I prefer that composition and I believe in that re-arranging, before or after the shot, to express what delights you, brings us a better image.

Thanks!

However, the filter is big question. I'd rather look for deconvolution solutions for undoing blur and Bart has mentioned a few in the past here.

Asher

 

[Bart_van_der_Wolf]

Thanks!

However, the filter is big question. I'd rather look for deconvolution solutions for undoing blur and Bart has mentioned a few in the past here.

Hi Asher, Bob,

Indeed deconvolution should be able to bring some relief, but from what I can see now it won't be enough to completely solve the motion artifacts. Maybe with a better source image than a small JPEG, the improvements can be better.

A nicely performing contender for such repairs, is a program called DeblurMyImage, and you can try it without having to pay, the Freeware version just gets slowed down until you do. It's simple enough to use for most, just a slider and some selections for amount of noise and smoothing.

Cheers,
Bart

 

[Bob Rogers]

Bob,

In the first post, did you take the moon and its reflection and clone it to the left side to get the picture you first showed?

I moved it although it's not exactly a clone. I copied and pasted that whole slice of the image and moved it. Then I masked it and changed the mode to overlay or hard light or some such.

I will give the deconvolution a shot and see how that goes.

 

[Asher Kelman]

Bob,

I knew Bart had good choices for this issue.

Thanks, Bart!

Asher

 

[Bart_van_der_Wolf]

Bob,

I knew Bart had good choices for this issue.

Thanks, Bart!

You're welcome Asher.

The tool does an okay job, but the result is far from perfect. After all, it's impossible to 'un-stir' an omelette ...

The better, more pure, the input data is (16-bit/channel, unsharpened, no corrections), the better the result can be, but some of the original signal will be lost beyond repair. What we occasionally see on TV-shows like CSI, is sometimes pure fiction, but entertaining.

I tried the same type of restorations on the video captures of the Boston bombers, but not much could be made of those either, although more could perhaps have been done with multiple sequential frames had they been available.

Cheers,
Bart

 

[Asher Kelman]

You're welcome Asher.

The tool does an okay job, but the result is far from perfect. After all, it's impossible to 'un-stir' an omelette ...

The better, more pure, the input data is (16-bit/channel, unsharpened, no corrections), the better the result can be, but some of the original signal will be lost beyond repair. What we occasionally see on TV-shows like CSI, is sometimes pure fiction, but entertaining.

I tried the same type of restorations on the video captures of the Boston bombers, but not much could be made of those either, although more could perhaps have been done with multiple sequential frames had they been available.

Cheers,
Bart

Bart,

I wonder whether one could use other, well defined, pictures of the person/suspect and know when it was a perfect fit and then apply those corrections to the rest of the image to discover heretofore hidden evidence?

Asher

 

[Bart_van_der_Wolf]

Bart,

I wonder whether one could use other, well defined, pictures of the person/suspect and know when it was a perfect fit and then apply those corrections to the rest of the image to discover heretofore hidden evidence?

Hi Asher,

Yes, such techniques are being used. But if the original image leaves too many things unclear, it is hard to conclude with certainty that it is the actual person. Only probabilities can be given, and that might not hold in court.

The main drawback of most of the current systems is that they need frontal images to match up with the database which also has the various facial measurements (eye distance, width of nose, cheekbone positions, positions of the ears, width of mouth, thickness of lips, etc.), but things get more accurate and 3D all the time.

Of course the person has to be in the database already, or they can only do a modern version of a composite, by combining parts of faces, but that's often not very accurate.

Given Bob's difficult shooting conditions, it's still a nice memory frozen in time. He can probably improve the quality a little because he has the original file, but no miracles should be expected.

Cheers,
Bart