There are situations where I would reluctantly have to turn the ISO to the larger extreme. Nevertheless I can get some photos that I pretty much like under these circumstances.

Here is an example.

enter image description here

(Well, It might not be a masterpiece, but I do like it at least.)

And I'm trying to remove the white dot noise in the dark background.

I tried to copy the layer and blur with gaussian blur, then use a mask to leave the foreground out.

It turned out to be like this:

enter image description here

It changed a little bit. But not perfect.

So How can I further enhance this ISO corrupted photo?

  • 1
    \$\begingroup\$ Honestly I think that the corrected image looks worse than the original, as the background noise becomes like larger "blobs" that result more noticeable. And in the hair you can see the edge of the correction. \$\endgroup\$
    – clabacchio
    Jan 17, 2014 at 10:09
  • \$\begingroup\$ @clabacchio honestly you are perfectly right on the first part. However I didn't touch the hair in my retouching. That edge shouldn't be the resulted from correction. Thank you for noticing this after all. \$\endgroup\$ Jan 17, 2014 at 13:36

1 Answer 1


Masking the foreground and denoising the background is a good technique when the background is out of focus.

You would get a slightly better result using a dedicated noise removal plugin (noiseninja or neatimage, or one of the inbuilt Photoshop tools). Gaussian blur is a bit of a blunt instrument, whilst it's good at removing pure random noise it is not as good at removing banding, which is visible in this image. It's also bad at preserving hard edges. An edge aware noise filter would let you mask much closer to the edges of your in-focus subject without risk of blurring it, which in turn would avoid the halo of grain you have around the bottle.

Blending a little (25% or 33%) of the original noisy background back in helps the result look less fake, whilst still allowing a low overall noise level.

Finally when you have a noisy image, save using the highest quality option if using JPEG (or use PNG). JPEG's attempt to compress noise often looks worse than the noise itself!

  • \$\begingroup\$ You mentioned quite a few very useful points that could helped me a lot. I've tried neat image and it worked out perfectly. And luckily I saved the raw for this photo. BTW you are analyzing the problem in an "image processing" way so I'm wondering if you have a CS background? \$\endgroup\$ Jan 17, 2014 at 14:24
  • \$\begingroup\$ Add Topaz Denoise to that list. \$\endgroup\$
    – Robin
    Jan 20, 2014 at 21:05

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.