I'm not a photographer but a graphic designer.

In GIMP, I only discovered two methods of improving the quality of an photo:

Filters > Enhance > Unsharp Mask

Noise > HSV Noise

Basically, I would like to remove some blurriness and 'fake' high resolution, if possible (I don't mind adding some noise).

I like this final result: http://www.blog.spoongraphics.co.uk/about (the guy in the picture).

Any other suggestions?

  • 2
    \$\begingroup\$ You should say more about what you mean by image quality. Composition->crop, contrast->levels/curves, color balance, removing distractions, blemishes... Also what are the pictures of? People, landscape,... \$\endgroup\$
    – mmccoo
    Feb 24, 2011 at 15:17
  • 1
    \$\begingroup\$ Please be less vague if you want useful responses. You should indicate what issues you are having (different problems require different solutions & some have no solution). Also, what is your medium (different processes work best with certain media: matte printing, glossy brochures and web use are all a little different). \$\endgroup\$
    – Itai
    Feb 24, 2011 at 15:20
  • \$\begingroup\$ I updated the question. \$\endgroup\$
    – wyc
    Feb 24, 2011 at 15:27
  • 2
    \$\begingroup\$ more less-vague would still be helpful even after the edits. \$\endgroup\$
    – mattdm
    Feb 24, 2011 at 16:08
  • \$\begingroup\$ @mattdm I'm pretty sure the questioner is talking about upsampling images, doesn't seem too vauge to me \$\endgroup\$
    – Matt Grum
    Feb 24, 2011 at 17:23

2 Answers 2


There are many ways to improve an image, and sharpening and noise reduction are just the very start. These techniques are almost completely program-agnostic, and can be done using a variety of different tools, Gimp included.

In photography, this is a major portion of the work that is done, and has become known as "Post Processing" or simply "Post".

Post-Processing Techniques:


  • Cropping
  • sharpening
  • white balance correction
  • Contrast enhancement
  • spot touchups/dust removal


  • noise reduction
  • cross-processing
  • halftone/duotone
  • HDR

These lists are by no means exhaustive, but merely show a few of the more common techniques.

There are quite a few tutorials around that may be helpful for you.


It sounds like what you really want to do is increase the resolution (size) of the image whilst retaining as much quality as possible (and avoiding pixilation).

What you need for this is a fractal based resize algorithm, Genuine Fractals is a popular commercial solution, there doesn't seem to be a free alternative, so your best bet is to scale up using bicubic resampling (already supported in Gimp) and add some noise to give the appearance of detail.

At the end of the day it is impossible in general to replace detail that was lost at the time of capture.

  • \$\begingroup\$ Liquid Rescale isn't primarily an upscaling plugin -- it's a "content aware" resizer, designed for photo manipulation. Pretty cool for what it does, but not the same as Genuine Fractals. \$\endgroup\$
    – mattdm
    Feb 24, 2011 at 16:12
  • \$\begingroup\$ Yeah I just realised that and had to edit! \$\endgroup\$
    – Matt Grum
    Feb 24, 2011 at 16:13
  • \$\begingroup\$ So, I looked around a bit, and the only implementation of fractal image encoding in the open source world appears to be based on some y2k-era work called Fiasco. It's implemented in netpbm. Unfortunately, it's a) not very developed and b) entirely focused on high compression levels, not on rescaling. So, for this purpose, it's basically useless. \$\endgroup\$
    – mattdm
    Feb 24, 2011 at 19:38
  • \$\begingroup\$ But see: theonlinephotographer.typepad.com/the_online_photographer/2008/… and theonlinephotographer.typepad.com/the_online_photographer/2008/… — the upshot being that the proprietary scaling algorithms aren't a clear win over bicubic. \$\endgroup\$
    – mattdm
    Feb 24, 2011 at 19:46

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.