Sharpening a blurry photo

DataDesign · March 10

I have a photo I scanned from a print ca. 1956.

I do not have the negative of course and the faces are quite blurry.

Can Imerge or for that matter any software sharpen this photo?

Thanks

TheBenNorris · March 10

I'd imagine to some extent there is something you could do in a photo editor, although some of the areas have blurred to just solid colour, so sharpening them wouldn't add any missing detail. It could be worthwhile looking into something called "Super Resolution", which is something that machine learning can do. There may be some software that's worth trying in that area.

DataDesign · March 10

@TheBenNorris Thanks for the reply.

Triem23 · March 10

To agree with Ben, Imerge, most photo editors and Hitfilm all have sharpening, but, as sharpening works by exaggerating contrast at pixel edges, there's only so much that can be done.

This is an area where those new AI resizing tools are working miracles.

If you try for an AI solution I'd recommend also scaling to twice desired size, then scale back down. You'll get a free "apparent sharpen" over what the AI does.

Even if you stick with a Photo Editor I'd suggest scaling to twice the size, OVER Sharpening (getting visible halos) then down scale again.

Stargazer54 · March 11

Absolutely! What @Triem23 said. Resize to at least twice the original size. Then sharpen. Then re-scale back down. No offense, but remember the old adage - "garbage in, garbage out". But you can cheat a bit by increasing the pixel count by resizing before making modifications to the image.

DataDesign · March 11

@Triem23 @Stargazer54

Thanks. But I am not sure how to resize the photo. I just scanned it in on my printer/scanner Epson 845. The results are what you see above, unless the forum resizes it when I upload. To me it is just a piece of paper from the last century (non-digital). Seems to me that if I scan it then resize it larger, I will loose more detail. Sorry I am fuzzy on the concepts.

Triem23 · March 11

Sure... The though here is that a little sharpening can clean up some details, but heavy sharpening can cause visible haloing.

The logic here is that you scale the image up (any photo editor can do this - so can Hitfilm, depending on the resolution you need) and sharpen the heck out it until you're getting some visible halos. Then you shrink it back down. Basically this tries to remove the halos of the oversharpen but keep the benefits. It's an old trick to sharpen at double resolution then scale down to smooth things out a bit while keeping the benefit.

Stargazer54 · March 11

What Mike said. You'll need a dedicated image editing program to re-scale the image - such as Photoshop.

DataDesign · March 12

@Stargazer54

"dedicated image editing program"

I thought Imerge was a dedicated image editing program. Thus this post. ?

Stargazer54 · March 12

Sorry. Got off track and forgot this was an Imerge question

Triem23 · March 12

As did I.

Interestingly, the old "scale up, sharpen, scale down" trick was developed for "destructive" workflows - where edits write directly to the pixel data as applied. Imerge is non-destructive (keeps edits procedural and only writes pixel data on export, which means you'd do a couple of "bounces" to try this workflow... Scale up and export out as a PNG, TIFF, TGA, or EXR, re-import. Sharpen. Re-export. Re-import. Scale down. Re-export. Heh.

Most of the time a non destructive workflow is advantageous, but there are a few use cases where it gets in the way.

SamuelMorris · March 12

@Triem23 you can probably skip one step by scaling up and then sharpening on an adjustment layer. Then export/reimport and scale down. Still agreed it's not a great workflow for this situation! And probably won't be able to save much from the original scan. Sounds like ML really is the best bet here.

Triem23 · March 12

@SamuelMorris you are correct, sir! This is what I get for answering questions before having breakfast.

DataDesign · March 12

@Triem23 @Stargazer54 I see we share some things in common. :-)

Thanks again, and thanks to @SamuelMorris also.

Sharpening a blurry photo

Comments

Howdy, Stranger!

Categories