Fractional ISOs–Should you use them?

This blog post comes with a warning: it’s kind of boring and technical. And the photos that accompany it are also very boring. That being said, I think this subject is very important, and I really wanted to share it with fellow photographers.

The other day, while reading some online photography forums, I stumbled across an intriguing bit of information about camera ISOs and noise levels. First things first–ISO levels on digital cameras are supposed to be similar to the old film speeds: 100, 200, 400, 800, etc. They play a role in determining exposure, but often the higher the number, the more “noise” you’ll see in the photos. Noise is basically graininess. Sometimes grain looks cool, but I prefer clean and crisp photos. Of course, dark environments often necessitate using high ISOs, so it’s important to know which levels look less noisy.

I always believed (I mean up until a couple of weeks ago!) that the scale of 100 to 1600+ was pretty even. In other words, 100 was the cleanest, 125 was next, 160 after that, and so forth. It looks like I may be wrong. I read an interesting note about “fractional” ISOs (125, 160, 250, 320, 500, 640) being noisier than their traditional film counterparts (100, 200, 400, 800). It had something to do with how the camera digitally computed the difference in ISO levels…and it soon got too technical for my right-minded brain. But, I wanted to see for myself whether there was any truth to this, so I pulled out my Canon 5D and my Canon 5D mark II cameras for a little test run.

And the results? The effect is true–but the two cameras differ slightly. See for yourself. Note: all samples have been cropped to 100%, and I’ve maxed out all exposure settings in lightroom to show you the noise. Also, beyond 1250 noise just continues to increase, so I’m not showing those results:

Canon 5D classic ISO levels 100-200

ISO levels 250-500

ISO levels 640-1250

Canon 5D Mark II ISO levels 100-200 (note, for the 5DII, I changed the aperture, so all these shots look brighter)

ISO 250-500

ISO 640-1250

From the tests, I conclude the following for my 5D camera: 200 ISO is better than 125 and 160, so is 400 better than 250 and 320, as well as 800 better than 500 and 640. So, if I need to move up an ISO on my 5D, I’m just going to go to the next “film” equivalent, instead of using the fractional ISOs.

For the 5D Mark II camera, the results were a little more mixed, I thought. At 125 ISO, things were much uglier than 160 and 200, but I thought 160 looked similar to 200. ISO 250 was nasty, but 320 and 400 looked OK. 640 looked similar to 800 and both I thought were slightly better than 500. Overall, I think on my 5DII I’ll avoid using 125, 250, and 500. Hope that’s not too much to remember!

I’m curious as to what other folks have noticed on their own cameras. I have no idea if this is also a Nikon issue, as I only had Canon 5Ds to test on. I would like to see what a 50D or 7D does. Interestingly enough, my good old trusty 20D only has ISO levels of 100, 200, 400, 800, and 1600. Maybe Canon had a good reason for that. Anyway, I hope this bit of info is helpful to others out there. It definitely was to me!

Rachel Durik - Wow, I had no idea bout this. I always thought the higher you go, the more noise. Very interesting!

caroline - I’ll have to test this out on the Nikon. I tend to stick to ‘film’ speeds when I shoot, I guess out of habit, but I’m curious now.

Elaine - I remember reading about this awhile back, and since then I’ve tried to stick with the normal ISO numbers. I’ve never tested it though (I have a Canon 40D), so this was interesting to see. Thanks for sharing!

Christina Montemurro - Weird, I was just reading about this somewhere, although now I can’t remember where. Wherever it was, they drew the same conclusions you did. Interesting stuff and something to look out for – I know I use ISO 640 often and will have to rethink that and instead just bump up to 800.

Milla - Wow this is so interesting… I’ve NEVER heard of fractional ISOs. Wow. I’ll have to try this out as well and let you know what I find. Interesting…

Your email is never published or shared. Required fields are marked *

*

*