Quick question about natural ISOs

Finnegan

Finnegan

Well-known member
So in DSLR video I know it's a rule of thumb to shoot at natural ISOs (160, 320, 640...etc.) since they're noticeably less noisy. My question is does this same principle apply to DSLR photography?
 
Not really. Just the higher you go, the more noise, so you can more or less eyeball it. Also easier to remove in say, lightroom, than it is in video.

Full disclosure: I am not a photographer.
 
I never was completely convinced by the 160x thing: the most common proof of this idea is a test with the cap on, which is absolutely pointless. It seems that 320 is 400 pushed down 1/3 stops in "in-camera-post". It is the fact that you use more light when shooting ISO 320 than when shooting ISO 400 which explains the reduction in noise (you're overexposing a bit, then pushing down a bit). This added light does indeed generate less noise in the final footage, but there's a price for that: 1/3 stop less DR (not a big deal), and harsher highlight rolloff (this can be an issue). So I'd say: don't use 160x if you're having any highlight clipping in the frame, or if you're not able to increase the amount of light hitting the sensor to compensate for the 1/3 stops lower ISO.
 
DPPro had an article on how to test your camera for the lowest noise ISO settings. Basically you just shoot a black frame in manual with the lens cap on at a slow normal shutter speed like 1/30sec for each ISO speed. Bring up the brightness in photoshop until you can see the noise at the lowest ISO, then compare the noise at various ISO settings. Though higher ISO's will obviously be noisier, due to differences in gain stepping and processing the increase is not necessarily linear with each step up. Different brands and models of cameras may have different sweet spot settings. One may show less noise at 400 vs 320 for instance. Test your own cameras.
 
Razz16mm said:
DPPro had an article on how to test your camera for the lowest noise ISO settings. Basically you just shoot a black frame in manual with the lens cap on at a slow normal shutter speed like 1/30sec for each ISO speed. Bring up the brightness in photoshop until you can see the noise at the lowest ISO, then compare the noise at various ISO settings. Though higher ISO's will obviously be noisier, due to differences in gain stepping and processing the increase is not necessarily linear with each step up. Different brands and models of cameras may have different sweet spot settings. One may show less noise at 400 vs 320 for instance. Test your own cameras.
Click to expand...

But such a test completely ignores highlights and the clipping thereof - as important a consideration as shadow noise.

Here is what I wrote a few days ago at the other forum, follow the link for a1ex's response:

Those were popular opinions on the internet a while ago, but they were based on flawed analysis of incomplete tests.

Testing with the lens cap on may give you an idea of the shadow noise level of a setting, but there is also the issue of highlight retention to take into consideration. The 100 divisible ISOs actually have higher dynamic range in RAW mode than the 160 divisible ISOs when highlights are considered as well.

The 100 divisible ISOs only adjust sensor gain to achieve final exposure. All the other ISOs over or underexpose and then adjust in software to the final exposure. In the case of the 160 divisible ISOs, the image is overexposed and then dimmed by the camera software - which results in less shadow noise, but also means highlights clip sooner, decreasing dynamic range. The 100 divisible ISOs are the true native ISOs with more highlight headroom and total dynamic range in the RAW image - although all of that extra highlight information doesn't make it into a jpeg or movie file, much is discarded in the conversion from RAW. The end result of this is less shadow noise at 160 divisible ISOs, with less effect on the highlights in jpeg or video mode than in RAW mode, but still some additional blown highlights.

By any definition of "native", the 100 divisible ISOs are the native ISOs, although when shooting jpeg or movies of dark subjects the 160 divisible ISOs may give better results.

Some prolific bloggers ran with the early analysis of the lens cap tests and propagated this misinformation quickly, with the truth being slower to disseminate.
 
If ISO 160 would have been ISO 200 pulled by 1/3 EV, the blown highlights would have been identical in both cases.

But ISO 160 does have more highlight detail than ISO 200, which makes me think that 200 is actually pushed by 1/3 EV from 160.

Moreover, going down to around 150 or 140 (still based on ISO 200), white remains white (at 255) - with Neutral 0,-4,-4,0. This makes me think that ISO 160 is also pushed a tiny bit, by around 0.2 EV (maybe it's just a little headroom to make sure white is white without calibrating each sensor individually).

ISO digital gain is linear (it's equivalent with "dcraw -S" parameter). Sensor output may become nonlinear just before clipping, though.
 
a1ex said:
If ISO 160 would have been ISO 200 pulled by 1/3 EV, the blown highlights would have been identical in both cases.

But ISO 160 does have more highlight detail than ISO 200, which makes me think that 200 is actually pushed by 1/3 EV from 160.

Moreover, going down to around 150 or 140 (still based on ISO 200), white remains white (at 255) - with Neutral 0,-4,-4,0. This makes me think that ISO 160 is also pushed a tiny bit, by around 0.2 EV (maybe it's just a little headroom to make sure white is white without calibrating each sensor individually).

ISO digital gain is linear (it's equivalent with "dcraw -S" parameter). Sensor output may become nonlinear just before clipping, though.
Click to expand...

So basically if you are shooting day time with plenty of light stick to 100 iso, when you are in low light situations go with 160. At the end of the day, keep your image below 1250 iso and properly expose and you will have a great image regardless.....
 
You should test to find out specifically how your camera responds. What if 400 ISO has less shadow noise than 320? This was the point of the article, that in some cases particular ISO's may actually show less noise than either the the next higher or the next lower value setting, indicating a setting that does produce the maximum DR within the range of +/- 1/3rd or 1/2 stops choices.
ISO choice is not as straight forward for digital cameras as it is for film stocks. For some cameras you get better highlight rolloff and DR under exposing a stop or so at 400 ISO than exposing normally at 200.

To respond to the second part of the question, it shouldn't make a difference if you are shooting video vs stills except for the possible effect of typically higher shutter speeds on stills.
 
Last edited:
You must log in or register to reply here.
Back
Top