Shooting in RAW format to reduce noise

I've recently came back from a volunteering mission in Morocco. I needed to pack light so didn't bring my full camera gear and didn't bring my flash. Now not having a flash means that many of the indoor lighting conditions I had to shoot at high ISO which normally translates to more image noise. However, I'm going to share a technique I've learned from another camera forum in the past which goes over how to reduce noise in low lighting shooting at high ISO which I think you'll find useful.

First of all you need to shoot in RAW for this technique to work. First of all lets define what noise is and this way you will understand the technique I'm going to explain easier. Noise occurs in the region of low signal of your camera sensor or image. i.e. occurs where things are dark. Here is the wikipedia definition, NOISE. Essentially in the dark regions where the light hitting the sensor is weak, the random firing of the sensor in that region may over power the actual signal of the low intensity of light hitting those pixels. Thus translating to what we see as noise. Now at higher ISOs the camera sensor has a higher amount of this random firing secondary to the higher sensitivity. However, if you look at a dark region of a photo shot at ISO 100, you'll still see some noise.

So here's the trick in order to reduce the amount of noise. Have less dark areas in the photo. Do this by over-exposing you're photo by 2/3 to 1 stop. This will brighten the dark areas and hopefully over power the random firing of the sensor, decreasing the noise. But wait, the photos are now too bright. Now what you do is in your RAW editor bring back down to the correct exposure by 2/3 to 1 stop. And presto a photo with less noise than if you shot with correct exposure. Now run your normal noise reduction software on this adjusted photo.

A few things to keep in mind:

What your shutterspeeds are. Since you are over exposing your photo, the shutterspeeds are slower. Higher potential for camera shake or motion blur. This doesn't work well if you HAVE to shoot at your highest ISO in order to stop motion. I use this technique in this way. And I'll use an example to illustrate.

ex. scene shot at ISO 1600 1/100 +1 EV and brought back down in RAW editor will have less noise than same scene shot at ISO 800 1/100 EV 0

Other thing to keep in mind are blow highlights because of the +1 EV. So may not work for all scenes if you need the info from the highlights or brightest parts of the scene (ex. candlelights, lace pattern on wedding dress, etc.)

So try it out. You can try a side by side comparison or test at home to see the differences. I theory it should work for all cameras but it will depend on the noise performance at high ISOs for you given camera model. I hope this helps in reducing the noise in your photos and increasing your acceptance ratios.

Photo credits: Karoline Cullen.

Your article must be written in English

Publish
June 24, 2008

Retina2020

Arindom, you are totally correct. So you definitely need to watch your histogram. However, at higher ISOs the dynamic range of the sensor is much greater than lower ISOs. Therefore more of the highlights will be preserved as long as your histogram is not over the right limit of the graph.

June 21, 2008

Retina2020

Lee. I guess the only way to know is to try it out. In theory +ve exposure compensation or over exposing the photo deliberately should reduce the total amount of noise.

June 21, 2008

Suntrekker

Is this gonna work on my Fujifilm Finepix S9100/S9600?
Thanks anyway.

June 12, 2008

Titania1980

thank you very much for sharing this tip!! Since I bought my canon G9 I always shoot in raw, but I have still noise

June 07, 2008

Retina2020

Hey Patricia. This technique is used during the photo taking process. I.e. shooting +2/3 of stop. And afterwards bringing the exposure back to normal in the raw editor. This technique reduces the amount of noise to begin with.

May 28, 2008

Azathoth973

it works!! I usually do this on my nik D300... the results at ISO 3200 +2 EV are often SURPRISINGLY good.

C.

Related image searches
Noise related image searches