April 2016

Saturday, April 30, 2016

How to use the histogram to reduce noise in your photos


Description

Noise in photos can be a problem at higher ISOs, particularly in older cameras. In the episode, Harley talks about how shooting to the right of the histogram and then adjusting the exposure in post processing can help minimize noise. Topics covered include the reason this technique works, how to make the image in camera and how to adjust it in post-processing.

Special thanks to Rich Legg at Studio o2o for letting me use his remodeled lobby for the example shots. Studio o2o has great photography studios available for rent in the greater Salt Lake City, UT area. See their web page for more details.

Histogram playlist

For a written transcript, go to How to use the histogram to reduce noise in your photos

Music under Creative Commons License By Attribution 3.0.
Intro/Exit: "Hot Swing" by Kevin MacLeod at http://incompetech.com
Sound effect: living-room-light-switch by alienxxx at http://freesound.org

Transcript

Perhaps you’ve heard you should "shoot to the right."

What does this really mean?

And why should you think about doing it?

I'll try to answer these questions today at the House of Hacks.

[Music]

Hi Makers, Builders and Photographers. Harley here.

To "shoot to the right" refers to the right side of the histogram.

If you're unfamiliar with the histogram and how to use it, I have other episodes in this playlist where I talk about these topics.

In general, with an average scene, a proper exposure looks something like this one on the histogram.

The black side starts small, it increases through the grays and then drops off again approaching white.

The camera’s computer will try to make every scene fit somewhere in this curve.

The problem comes when you have large dark areas in your images.

These areas are more susceptible to having problems with noise coming from the sensor.

Here's an example of noise. See these odd blotches in what should be a smooth dark area?

Noise actually happens over the whole image, but it's mainly visible in the dark areas. Let’s see why.

Noise causes small random changes in the value the sensor detects for a given pixel.

Because the changes are small, and dark areas have small values, the noise is a greater percentage of the actual image data in the dark areas as compared to the lighter areas.

Just for the sake of argument, let’s say the noise is variations between plus and minus 10 values from the actual, ideal image pixel value.

Let’s say this area of the image ideally should have values around 20. When we add noise, a swing of 10 represents a change of 50% from ideal of 20.

If you have a two adjacent pixels that both ideally should be 20 but one has +10 noise and the other as -10 noise you end up with a 100% difference from the ideal between the two pixels.

Conversely, if we look at a bright area of the image where the values are around 200 a change of 10 represents only 5%, a much smaller relative change.

And so, because the noise represents a large percentage of the dark area’s values but only a small percentage in the bright area, it’s much more noticeable in those darker areas.

So what does this mean for the topic of “shoot to the right?"

The idea is that you should adjust your exposure compensation to shift the "ideal" exposure for gray to the right as much as possible but without clipping.

If you're in an automatic mode, you do this using the exposure compensation feature of your camera.

And if you're in manual mode, just simply adjust one or more legs of the exposure triangle to get what you want.

Use the histogram to determine when you have arrived at the correct point.

First, take a test image at what the camera says is the "correct" exposure and look at the histogram to see how much room you have.

Then dial the exposure compensation up to increase the exposure. The idea is to move the peak of the image to the right of the histogram.

Take another test image. And then repeat this process until the image is as bright as possible without clipping.

When you do this, the dark areas of the image move up into brighter areas where noise isn’t so much of an issue.

The problem now though is the dark areas will no longer be their proper dark color, they will be a washed-out grey. And the medium areas will be much too bright.

It just simply won't look right, but this can be fixed in post-processing.

Here's our image in Photoshop, a bit too bright since it was shot to the right. There are several ways of adjusting this.

The curves adjustment layer can be used to move that area that's too far to the right back into the center.

Just drag the black point at the bottom left of the curve over to the right but don't go so far that you start clipping your blacks.

You may also need to adjust the curve down to darken the overall image.

Another option is to use a levels adjustment layer to move the black point to the right, causing the overall exposure to be lowered and evening out the curve.

A third option is to use an exposure layer to simply lower the exposure.

Regardless of the method used, since the dark part of the image was captured outside the noise range, when the image is adjusted, the area the noise is in will be crushed and made much smaller, resulting in significantly less noise in the dark areas.

Here's the example scene shot with the exposure as the camera decides it should be.

And here it is shot to the right.

Finally, here's the shot to the right example with post processing.

Notice the difference in noise in the dark areas.

In conclusion, I’d love to hear in the comments if you have any opinions on using this technique. I know in some places it’s a bit controversial and in newer cameras noise is much less of a problem.

If this is your first time here at House of Hacks: Welcome. I’m glad you’re here. We’d love to have you subscribe.

Through these videos I hope to inspire, educate and inform makers in their creative endeavors.

Usually this involves various physical media like wood, metal, electronics, photographs and other similar types of materials.

Thanks for joining me on our creative journey. So subscribe and I’ll see you again in the next video.

Now, go make something. It doesn’t have to be perfect, just have fun!