Test how far your camera can go at ISO

Monday 21 July 2025, 16:32 by Nando Harmsen | 5,377 views | 5 comments

This article was translated from Dutch with the help of AI. Read the original.

Modern cameras let you choose extremely high ISO values. Sometimes up to ISO 51200 or even higher. Are these values still usable? More importantly, how do you find the maximum ISO value you find acceptable?

The highest ISO value a camera can handle rarely produces a good-quality photo. Both luminance noise and colour noise make the photo unusable. But up to which ISO value is a photo still usable?

The ISO settings on a Canon camera. You can set how high the camera is allowed to go. Do you know which ISO value is no longer acceptable to you?
There is always a limit up to which the ISO value still produces a usable image. That limit depends on the camera, but even more on the photographer's personal preferences. What you find acceptable, someone else may already find unacceptable.

So it is not wise to rely on what you read or on preconceptions. Test for yourself how far you want to go with the ISO on your camera. This way, you can make optimal use of the possibilities your camera offers.

Sometimes you simply have to go beyond what is acceptable. The difference is having a photo or not. Here, I went beyond the limit I find acceptable. But I did capture the memory.
You can do the test in two stages. The first is based on a correct exposure. This is a real ISO test. The second tests how far you can correct underexposure. In effect, this tests the dynamic range of the camera.

An ISO test of the Nikon Z6 II. Which ISO value do you still find produces a usable result? For a good comparison, assess the images at 50% or 100% size.
I recommend carrying out the test in the file format you always use. That is jpeg for photographers who do not edit their own photos, and raw for photographers who do.

ISO value test

Choose a subject in your living room and light it with a lamp. It does not matter which lamp, as long as the light does not change. Put your camera on a tripod and set the exposure manually at ISO100. Choose a fixed aperture, for example f/5.6 or f/8.

Take a photo at ISO100. Then raise the ISO by one stop, to ISO200, and make the shutter speed one stop longer. Take another photo. Then choose ISO400, make the shutter speed one stop longer again, and take another photo.

The test I carried out with the Sony ZV-E10. Make sure the light remains constant, as this is important.
Repeat this process up to the highest ISO value you can set on your camera. Take one photo for each ISO value, adjusting the shutter speed to keep the exposure the same. Then load the photos into your favourite editing programme.

Check the noise levels in the photos. Do this at 50% and 100% magnification. Look at them calmly, one by one, and determine which photo still has no distracting noise. That is the maximum ISO value you find acceptable.

Assess the photo at 50% or 100%. This is ISO102,400 on the Sony A7R V, which is completely unacceptable to me.
You can also check to what extent your software's noise reduction can 'save' a photo taken at a higher ISO value. But be careful, as this may also depend on the level of detail in your subject. Sometimes it is usable, sometimes it is not.

Underexposure test

You can also test how far you can recover underexposure. For example, by brightening dark areas. You can do this with the same setup described above. I always do this test at ISO400, but you can do it separately for each ISO value.

Take a correctly exposed photo at ISO400. Make the shutter speed one stop faster and take the photo again. This one is then underexposed by one stop. Then repeat this until you have about five or six stops of underexposure.

Take a series of photos, each one underexposed by one more stop. You can do this for any ISO value you want.
Open the photos in your favourite software and correct the underexposed photos. Brighten the photo with one stop of underexposure by one stop. Brighten the one with two stops of underexposure by two stops. And so on.

Lightroom can go up to a maximum of five stops. If you want six stops or more, use the slider for five stops and a graduated mask for the remaining stops. See how far you can correct the photo while it remains acceptable to you.

How many stops of underexposure you can correct depends on the ISO value. Test it to see what you find usable with your camera.
This shows you how much underexposure you can still recover in a photo. It is useful for the Exposure to the Right technique. Here too, you can supplement the test with noise reduction.

One extra important point: the photos you see in this article have of course been reduced in size for the internet. You cannot properly assess the noise levels at this size. Do this in your photo-editing programme at 50% and 100%.

Get to know your camera

I think this test is essential for getting to know your camera. You will know how far you can raise the ISO without the noise detracting from the photo. That limit is different for everyone. Test it so you can make optimal use of your equipment.

Pushing the ISO value extremely far may produce that one photo that would otherwise not have been possible, or simply fail. This photo was taken at ISO12800, which I knew still produced acceptable noise.
However, do not be too critical either, because a noisy photo is often better than no photo at all. In addition, there are many effective noise-reduction methods you can use.

You might also like

• 

  Removing noise? Use it in moderation

  by Nando Harmsen

• 

  Review: Platypod Delta and Extreme

  by Nando Harmsen

• 

  A panorama should add value

  by Nando Harmsen

• 

  Adobe to acquire Topaz Labs and add its features to Photoshop

  by Elja Trum

• 

  Review: Canon RF 24mm, 35mm and 50mm VCM lenses

  by Nando Harmsen

• 

  Forget full-frame as the holy grail

  by Nando Harmsen

5 comments

Have your say

Please note: Articles older than a week can only be commented on by signed-in visitors.
Would you still like to comment? sign in or create a free account.