I'm trying to salvage a photo that was taken in low light with fast shutter speed. No matter what I do, it is always grainy. I'm interested in how to brighten the image without worsening the noise. I use GIMP. Attached is a JPG file. The TIF looks lighter and grainier.



4 Answers 4


I would suggest from my experience with GIMP starting by upping the brightness and the contrast, as that will substantially brighten the photo without too much painful noise.

Secondly, if you have more noise than you would prefer, I would suggest giving the plugin "wavelet denoise" a try, as it's a powerful denoising tool that -- if used correctly -- can reduce noise without the loss of too much detail.

Additionally, for an all around post processing boost, I would suggest GMIC which stands for GREYC's Magic for Image Computing, and is a plethora of plugins that can handle anything from sharpening to denoising to providing filters for your images. It's something that (in my opinion) GIMP should have by default.

Hope this helps you out!


Most of my hummer shots are underexposed because I crank up the shutter speed on my Powershot SX160IS and I've found that LightZone (now free) does a fantastic job of recovering them.

One of LightZone's most powerful and useful tools is the ZoneMapper. So I used a Style (filter) that uses ZoneMapper to increase exposure. (I've tried making exposure adjustments with ZoneMapper myself, but I don't have the experience with it to get the best results.)

Then I added a Tone Mapper from the Style sidebar under the High Contrast category and in the Styles Menu under the same category. I increased the Gamma, added a bit of Vibrance and Luminosity with the Hue/Saturation Tool, and finally added a High-Pass Filter to sharpen the midtones and highlights.

I don't notice a lot of noise, and I've read that noise can be mitigated with printing.

LightZone recovery

  • \$\begingroup\$ Finally getting back to this. Under Styles . High Contrast I have Polarizer,Soft Wow!, Soft Wow! 2, Wow!, no ToneMapper. I am using LightZone version 4.0.0 \$\endgroup\$
    – numberwang
    Jun 7, 2014 at 14:45

I also use RawThereapee to process raw files. You can experiment with the different settings, including exposure and noise reduction.

The sample image can be processed from within GIMP.

  • Personally, I don't mind some "grainy" luminance noise. To convert chroma noise into luminance noise, I use median blur in copy of the original in a new layer with color blending. (In this case, skipped this step.)

  • To bring up the brightness, I used multiple copies of the original in layers with screen blending.

  • I tend to avoid using the contrast and brightness tools.  I prefer to use the levels and curves tools. In this case, I did not use either.

  • Because the sample image contains mainly JPEG artifacts, I reduced blocking with jpeg2png.

  • I further cleaned and enhanced the image with Anime4K. Other algorithms to try include AMD FidelityFX Super Resolution and Nvidia Image Scaling.

  • If desired, the image can be enlarged with waifu2x.

Here is the result:


  • \$\begingroup\$ How did you draw out feather detail that wasn't in the original image? It's almost like you started over with a RAW image. \$\endgroup\$ Sep 29, 2022 at 2:05
  • \$\begingroup\$ The tools that are most likely to be of interest to you are jpeg2png (reduce blocking artifacts), Anime4K (restore and enhance), and waifu2x (resize). Anime4K and waifu2x are neural networks trained on anime, but can work reasonably well on photos. \$\endgroup\$
    – xiota
    Oct 1, 2022 at 16:46

One thing you could try is converting the photo in RawTherapee using either of LMMSE or IGV, its two demosaicing algorithms optimized for noisy images. When I tested them against other RT algorithms and Adobe Camera RAW, they were clearly superior. I didn't see a significant difference between the two.


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.