Below base ISOs (e.g. ISO 50, 75)

[Sifor]

Hi all,

I was watching a video by TNG (http://www.youtube.com/watch?v=MA_8UfySoIQ&feature=g-u-u) where he discusses ISO in detail. Specifically, he talks about the dynamic and tonal ranges for example on the D4 are widest at around ISO 75 despite base ISO being 100.

He also suggested in a comment the only disadvantage to using an ISO below base is the affect on shutter speed. I've done some research and apparently others believe ISO 50 for example is created artificially (apparently the same as ISO 100 +1 compensation, then artificially reduced) which reduces the ability to recover blown highlights in PP.

Eh I'm not following to be honest. DxO Mark (http://www.dxomark.com/index.php/Publications/DxOMark-Reviews/Nikon-D800-Review/Sensor-performance) seems to suggest for the D800 the best dyanmic range, SNR and colour depth is around ISO 75. So, should I where ample light is available use ISO 75 rather than ISO 100 (base)? What's the disadvantage in doing this besides the affect on shutter speed?

Cheers

[ricktas]

These non-native ISO's, generally defined in-camera as things like Lo1, Hi2, LO 0.3 etc are digitally enhanced ISO's. IE the sensor does not have a native ISO of that setting and the ISO is achieved by applying some digital magic to the image file to achieve it.

Having said that, with my D3, as you posted, the benefit if using ISO50 (native lowest on a D3 is ISO 200) was a slower shutter speed, but even pixel peeping I could not see any less noise than an ISO 200 photo. And the ridiculously high ISO's are generally not all that usable other than as a happy snap.

As for the disadvantage, well the extreme hi ISO's are usually not usable, the low ISO's (and I do not call them extreme cause you can get ISO 50 film) offer you slower shutter speeds, but that is about it, in real life use.

[arthurking83]

Hi all,

...... I've done some research and apparently others believe ISO 50 for example is created artificially (apparently the same as ISO 100 +1 compensation, then artificially reduced) which reduces the ability to recover blown highlights in PP.

Eh I'm not following to be honest.......


ISO50 is shot at ISO100 and then -1Ev compensation is added in camera processing. ISO100 -1Ev is equivalent to ISO50.
This is the same effect as you shooting at ISO100 and adding -1Ev in PP via your favoured software. Where the problem lies is in the shutter speed.
If you need a longer shutter speed for effect(eg. the classic example is for flowing water to look misty!), exposure compensation on the computer is useless! The shutter speed has already been set by the camera, and your long shutter speed is less long.

The problem with using non native ISO values is that the camera is already processing in some exposure compensation, and in general the camera is not usually as good at processing as is a quality image processing software.

I found that the best highlight dynamic range was achieved at ISO200 on the D300, where in many instances I could easily retrieve 2Ev of recovery in highlight detail, as well as shadow detail.
In fact very good dynamic range was achieved when shooting at ISO200, setting ADL to high and shooting for the highlights.
But ADL has it's pitfalls as well, and I found that Dynamic Lighting(D-Lighting) in CaptureNX2 achieved better results in most situations.

Sometimes tho, where dynamic range may have been a problem with a particular shot, I'd shoot up to +1.7Ev over for the highlight detail(at ISO200) and then use exposure compensation(on the PC) to recover that detail back.
If shot at any of the Lo ISO's, the highlight detail was less recoverable on the PC .. simply because the camera has already done some itself.
I'm guessing that processing an already processed image is where the problem lies.

One thing I did sometimes do, was that if I shot at lower than base ISO, I'd shoot to max out the highlight detail by not losing any, or maybe 0.3Ev over and easily recover that minimal amount of lost detail. And in doing this, I found that recovering shadow detail(excluding colour) gave better results in terms of noise when shot at the Lo ISO values .. as long as there wasn't any important colour info in those shadows.

In the end tho, I found that it was just easier to stick with base ISO unless a specific situation called for lower values.

I'm yet to check to see how the D800 handles Lo ISO values tho.

[Sifor]

Thanks both of you for your comments, your responses are consistent with my research (although more clearly articulated!). I'll continue to shoot at base ISO unless I want to play with the shutter speed.

But one question remains - why does Dx0Mark tests suggest the best DR, colour tone, SNR etc is at ISO 75, given it's artificially produced? I thought their tests are raw sensor data, and any artificial manipulation by the camera is excluded? Some people suggest shooting at ISO 75 for landscapes etc with the D800 for this reason (to push it to the max), but that defies the responses you two have outlined above..?

[arthurking83]

You have to be careful when reading their results I think.

(now I'm not 100% sure on this but this is the way I'm reading them).

Firstly check to see how accurate the camera's ISO sensitivity actually is.
DxO do an ISO accuracy test, so that is base ISO is ISO100, then they check to see how close this actually is to the film based ISO standards.
(in reality this is a meaningless test in that for the majority of photographers this makes no difference to everyday shooting! .. I'll explain why this is really irrelevant in a sec)
But!.... they test ISO for accuracy and if you look at the graph, the D800's ISO100 is closer to the reference ISO75 in real world performance.
if you also look at the ISO50 marker in the ISO graph, again it corresponds with ISO75 in the 'real world'.

So ISO50(Lo1) and ISO100(base) both give real world results that are equivalent to ISO75(as if you were shooting film).
I'm guessing that this ISO75 corresponds to ISO100 on the camera.

Why is this irrelevant(for some, and not others).
If you choose to use DxO's software, then this may be relevant, but if you decide you prefer other software for your raw workflow, then DxO's results(for camera reviews) are really only relevant for their software.

The nature of raw files is that they must be interpolated into a usable image from a collection of data values which must be manipulated, and hence 'open to interpretation'.
The fact that the data must be manipulated into a usable form is the key here.

I'm guessing that Nikon calibrate their measurements to suit their software(CaptureNX and ViewNX) and not DxO's software, not Adobe's software.

This is easily seen if you open the same raw file in ViewNX2 and LR4, where you're almost guaranteed that the same image will be 'exposed differently' both in actual look(the image) and in technical correctness(the histogram).

back onto the topic of max dynamic range from the camera/sensor.
It's been written a few times by different folk that maximum dynamic range can be, or should be, achievable when using 1/3 to 1/2 stop below base ISO.
So in the case of the D800, this would be either Lo.07(ISO80) or Lo0.5(ISO75 )depending on how you set up the camera.

What the theory is supposed to be based on, is that at these ISO values, the camera is only minimally pre processing the highlight data, so there is a lesser destructive effect when the camera's processing is made onto the highlight data, and yet at the shadow end, it still allows the pushing of the shadow detail, which should give slightly less noisy results as they are pushed.
So in minimally processing the image in camera, and keeping ISO at it's lowest level with minimal destructive effects on the image, in theory you should be able to best capture all the data you require(in this case dynamic range).

I tried this for a short while on my D300 too, very early on in the scheme of things, but again, I just found that using base ISO with a mix of AutoISO when it suited, I got results I wanted to see.

note: I've tried DxO's software, and while it seemed to work well, I found it as annoying as all the others in that I wasn't seeing on the PC, what I was capturing in camera.
Using the Nikon software meant minimal processing effort.

While it's important to have information like this available, it's not something to live by!
From what I've read of others thoughts on the topic of DxO's testing, is that unless they release their actual testing methods, their results are not scientific data, just a matter of opinion from them using their own software.

What this could mean: they may have a better understanding of reinventing the Nikon raw file, than they do of the Canon or Sony raw file .. etc, etc.
DxO's results aren't to 'live or die' by. Use them as another resource to confirm or deny your findings.

[Sifor]

Thanks for your detailed response, I understand it better now.

I think I'll just stick to base ISO like I have been doing since day 1, I only use Lightroom.

[Kym]

Try this http://www.ausphotography.net.au/forum/showlibrary.php?title=New_To_Photography:Detailed_explanation_of_ISO