RootsChat.Com
Old Photographs, Recognition, Handwriting Deciphering => Free Photo Restoration & Date Old Photographs => Photograph Resources, Tips, Tutorials => Topic started by: Avondale16 on Sunday 07 September 25 16:21 BST (UK)
-
Hi - I'm in the process of putting together a history of my home town mainly for the education and benefit (hopefully) of my grandchildren.
I have a number of newspaper images (mainly buildings & street views) and I'm wondering if someone could recommend a simple program that would enhance these types of images.
I'm attaching a couple of examples. Any suggestions would be greatly appreciated.
-
Hi, Keith. Looks like not many people are looking at this sub-board lately.
To try to answer your question: I guess you’re wanting to convert the images to a clean black & white, and smooth out those lines without losing too much detail. Any free app should be able to do those things. GIMP is probably the best free one around, albeit with a steep learning curve if you want to get right into restoring; but it shouldn’t take too long to learn how to use a few of its basic tools.
To take it further—if you’re wanting to really sharpen the images as well…
Traditional apps can’t effectively both smooth and sharpen the one image without a lot of painstaking manual work. Smoother means less sharp, sharper means less smooth.
Enter AI. Whereas traditional tools manipulate existing pixels, AI tools make up new pixels, so that, potentially, they can both smooth and sharpen an image at the same time. However, to date, it seems most companies producing AI apps design their tools to create images that look great but at the cost of only resembling (more or less) the original rather than restoring it. Only the more expensive ones seem to aim to actually restore images using AI.
So, have a look at some of the free and/or cheap apps online, download one or two and give them a run, bearing in mind that further improvements will probably only come at the cost of more work or more money.
And of course, post any you’re not happy with on here to see what restorers can do with them. ;D
Peter
-
Many thanks Peter for your reply.
I will take your advice and give them a go. I not very technical minded so have replied on the generous restorers (like yourself) here to have improved my family photos.
I have assumed that the sort of images I am now looking to improve (street scenes, building etc) and not really what the photo restoration forum is designed for?
Thanks again
Keith
-
Keith, sorry if I misunderstood your intention. I'm sure the restorers on here would consider any type photo as fair game. :)
Peter
-
I will give GIMP a try and some of the AI programs to see how I fair.
Regards
Keith
-
If you are scanning originals, scan them at as high a dpi as possible.
-
If you are scanning originals, scan them at as high a dpi as possible.
I have to disagree with this statement. Photographs in newspapers and magazines are reproduced by a method called halftone screening. Put simply this reduces the image to a series of regularly spaced dots and it is the size of the dot of ink which converts the level of blackness in the printed image. Scanning also operates in much the same way, by taking measurements of brightness levels at pre-determined distances (the dpi in Dublin's reply). If there is a mismatch between the spacing of dots in the source image (known as lines per inch or lpi) which can range between 85 and 185 lpi, and the scanner (dpi) then this will create interference patterns which shows as lighter and darker bands across the scanned result, and thus actually degrades the image rather than improving it.
The solution is to use software to determine the lpi in the source image and then scan using an exact multiple of the lpi. What is certainly a good idea is save the scanned image in a lossless format such as tiff, and don't use .jpg, as the compression involved can negate the value of a quality scan.
You can read more about the whole process here: Wikipedia (https://en.wikipedia.org/wiki/Halftone#Halftone_photographic_screening)
-
Had a go at the first one of your images, Keith. :)
Peter
-
Wow! that is an unexpected bonus. What a great improvement - thanks so much Peter
Regards
Keith
-
Peter, I'd welcome a detailed description of exactly what you did, and how. It's very good.
Zaph
-
Thanks, Zaph.
Sometimes you can get rid of the lines from newspaper images by running them through a Fast Fourier Tranform (FFT) filter. There’s one in the free app, “Image J”. Fortunately in this case it worked well. (I’ve found that, for me, it works about half the time. I couldn’t get it to help much on Keith’s other picture.)
So, I did an initial clean-up with Photoshop Elements, my primary photo restoring app. Then I did the FFT.
Next I put the resulting image through Topaz’s Photo AI. I tried a couple of different processes. First I used “Super Focus” which did a reasonable job, but in smoothing the image it destroyed some detail I wanted to keep. I saved that image, though, and then tried again with the Topaz app's “DeNoise” and “Sharpen” filters at varying settings. That did a better job of keeping detail—but didn’t smooth it enough.
I went back to PSE and used layers and masks, blending the 3 images (FFT and the 2 Topaz ones) trying to get the best mix I could.
Finally I did some more tidying up with PSE's manual tools.
I tried a similar sort of process with Keith’s other picture but it didn't work as well. Hopefully someone else will have a go at it, or I’ll have another go later.
But, basically that’s it: playing around, trial and error, with the different apps, trying for the best compromise I could get. There's a fair bit of manual work, trying to manipulate the abilities of the different apps.
Hope this makes sense.
Peter
-
I'm very comfortable with computer technology but throughout my career there was never any involvement with graphics, so it is all a bit of a mystery to me. Thanks for your long reply. I will look into the products that you mentioned.
Zaph
-
Had a go at the second one, using similar methods as the first one, i.e., swapping between PSE and Topaz Photo AI, trying to get the best balance of retention of detail and smoothing that I could.
Peter
-
Another great surprise - so much better than the original. I never expected either of these to be restored. Thanks so much Peter.