Need a little guidance..

[Lazlo]

I consider myself fairly well versed on photoshop. I specialize in digitizing old glass negatives and have done about 500 in the last year.. That part I'm pretty good at. Had a friend ask me tonight if I could take a small color print of his uncle and make an 8 x10 out of it. I scanned the picture and don't know an easy way to fix all the small white spots. The background isn't a problem but not sure about the suit Any advice would be appreciated.

 

[dv8_fx]

I don't know what the others might do but if I get to do a restoration like this, OIIIII...... lol.

With image work like this, there are no shortcuts. You'd have to rebuild it a pixel at a time primarily with the clone tool on a new layer.

But first I'd duplicate the image and play with the brightness/contrast to bring out the details - fold, creases, etc - so in that way, I'd know where arms meet with the chest area and suit seams and use it as a guide in my editing work.

On another duplicate, use the Despeckle filter (In FILTER>NOISE) a number of times then apply Sharpen (then repeat process) to knockout/minimize the grime. Use the result as a clone source to rebuild areas such as the suit pinstripes and hair.

The process will be slow and pain-staking. It's more like a labor of love. But for a serious PS user, time stops and you just keep going at it until its done.

One small area at a time.....

 

[Tom Mann]

As dV8 pointed out, there are several problems in addition to the bright speckles. I'll only deal with the speckles and describe an automated 1st pass to minimize their effect.

1) Using any combination of "find edges", brushes and any other tool that you like, make a mask that identifies the edges (areas that should receive no treatment). It should look something like this:



2) Duplicate the original onto a new layer, set its blending mode to "darken", and add a layer mask that is the inverse of the edge mask you just developed.

3) Run PS's "Dust and Scratches" filter on that layer. Start with the radius around 5 px, and the threshold around 40, and adjust to your own preferences.

4) At this point, you are essentially done.

5) Of course, you can do things like reduce the opacity of this layer to reduce the effect or adjust the opacity of the mask to fine tune the edge effect. You can also modify the mask on a pixel-by-pixel basis by painting in white, black, or anything in-between to deal with troublesome pixels. Most of these typically occur around sharp edges. From there you can attack some of the other problems present in the original.

What I obtained is shown below. Total time about 5 minutes, most of which was spend writing up this post, LOL.




Of course, by far, the best way to deal with this problem is to recognize what causes it and prevent it from happening in the first place.

The cause of problems like this is an uneven surface on the print, some areas of which specularly reflects light from the scanner's bulb right into the CCD array. You've got a couple of choices:

a) If you are good at photography and have the right equipment, instead of scanning, you should rephotograph the print with polarizing gels over each of your lights, and a conventional polarizing filter in front of your lens with its axis 90 degrees to the axes of the filters on the lights. This is the way museums get good quality reproductions of artwork with thick reflective impasto layers or other sources of glint (eg, mixed media with embedded pieces of glass or plastic).

b) If you don't want to get into re-photographing it, re-scan the print several times, changing the angle that the print is laying on the scanner dramatically and randomly between scans. Accurately straighten out and re-position the scans in PS and overlay them. Set the blending mode of each layer above the first to "darken", or put them into a stack with the operation set to "Minimum". This way, the final result for each pixel is always produced by the least reflective orientation of the scan.

Done correctly, either of these methods will be vastly superior to the "dust and scratches" method that I described at the beginning of this post.

HTH,

Tom M

 

[Tom Mann]

PS - it's interesting that the subject is the uncle of a friend. He looks *a lot* like one of my uncles, as well -- a distinguished and kindly looking fellow.

Tom M

 

[dv8_fx]

I'll only deal with the speckles and describe an automated 1st pass to minimize their effect.

1) Using any combination of "find edges", brushes and any other tool that you like, make a mask that identifies the edges (areas that should receive no treatment). It should look something like this:

2) Duplicate the original onto a new layer, set its blending mode to "darken", and add a layer mask that is the inverse of the edge mask you just developed.

3) Run PS's "Dust and Scratches" filter on that layer. Start with the radius around 5 px, and the threshold around 40, and adjust to your own preferences.

4) At this point, you are essentially done.

5) Of course, you can do things like reduce the opacity of this layer to reduce the effect or adjust the opacity of the mask to fine tune the edge effect. You can also modify the mask on a pixel-by-pixel basis by painting in white, black, or anything in-between to deal with troublesome pixels. Most of these typically occur around sharp edges. From there you can attack some of the other problems present in the original.

What I obtained is shown below. Total time about 5 minutes, most of which was spend writing up this post, LOL.



Of course, by far, the best way to deal with this problem is to recognize what causes it and prevent it from happening in the first place.

The cause of problems like this is an uneven surface on the print, some areas of which specularly reflects light from the scanner's bulb right into the CCD array. You've got a couple of choices:

a) If you are good at photography and have the right equipment, instead of scanning, you should rephotograph the print with polarizing gels over each of your lights, and a conventional polarizing filter in front of your lens with its axis 90 degrees to the axes of the filters on the lights. This is the way museums get good quality reproductions of artwork with thick reflective impasto layers or other sources of glint (eg, mixed media with embedded pieces of glass or plastic).

b) If you don't want to get into re-photographing it, re-scan the print several times, changing the angle that the print is laying on the scanner dramatically and randomly between scans. Accurately straighten out and re-position the scans in PS and overlay them. Set the blending mode of each layer above the first to "darken", or put them into a stack with the operation set to "Minimum". This way, the final result for each pixel is always produced by the least reflective orientation of the scan.

Done correctly, either of these methods will be vastly superior to the "dust and scratches" method that I described at the beginning of this post.

HTH,

Tom M

I'm very certain the OP will be pleased with this as it's a great start towards cleaning and enhancing the photo.


It has been said by a lot of us here... we learn new things from each other from time to time if not everyday.....

I tried the high pass filter a long time ago but never liked the results. Either I was doing the wrong thing or there were significant improvements to the filter since then but never tried it again.

I don't often do retouch work on image problems like this but I'll keep your dust and scratches filter tutelage in mind and try it the next time I have the chance.....


Bookmarking for reference.....

 

[Tom Mann]

Thank you very much, Vee.

I'm blushing with praise like that coming from an old hand like you.

T

 

[Tom Mann]

A bit more tweaking gives more pleasant colors, but one has to be careful not to continue to tweak endlessly otherwise you start to reduce saturation and hue changes from point to point on the fact, and that makes it look plastic ... and you're seeing a bit of that creeping in here.

T