The lowest ISO on the D90 is 200. However, it also has expanded settings, down to Lo1 ( -1 EV below ISO 200). Is there a reason why Nikon didn't design the camera to go down to ISO 100 or lower, rather than having it use expanded modes? And, is there a downfall to using the camera in Lo1 vs. if it had an actual ISO 100 setting?
ISO 200 is the native ISO for the D90. At ISO 200 the images will show the least noise, providing the exposure is correct. The reason has to do with the design of the sensor itself. Using Lo 1 will result in a little more noise than there would be using ISO 200 on the D90. In simple terms; A sensor will work best at a certian light level. When the actual ambient light level is lower or higher, the sensor uses more power resulting in the sensor generating more heat. More heat results in more noise. Good Luck and Enjoy your Nikons!
>At ISO 200 the images will show the least >noise, providing the exposure is correct. > >Not true. The lowest noise is always at the lowest ISO >or "Lo" value. (And Exposing To The Right (ETTR))
Having just gone through the exercise of Noise Ninja profiling my D90 at all ISO settings (including all the Lo and Hi ones as well ass all the numbered ISO values) I can confirm what Bill states - the noise is lower at the Lo settings than at ISO 200.
>What is true is that below ISO 200 you don't get the expected >increase in highlights so you need to pay attention to your >histogram (but you do that anyway, don't you ) > >That's why the curves aren't straight at the "Lo" >ISOs for cameras on the >Photographic >Dynamic Range> Chart
Yes, I agree its in the exif, buried in the MakerNote. I was imprecise. For Lo 1.0, 0.7 and 0.3, Nikon does not write the ISO value *in the ISO field* in the exif. So tools which look for this information to be in the correct place will not find it.
Is there a reason why Nikon didn't design the camera to go down to ISO 100 or lower
The lowest ISO (the "native" ISO) is a function of the sensor. Technically, the limiting factor is the Full Well Capacity (FWC) of the photosite (a/k/a pixel). In other words, there is a maximum amount of light (photons), creating electrons, that can be held at the photosite before it is full. (Just like your bathroom scale can only weight up to a certain maximum amount.) The sensor in the D90 is about as good as it gets for that class of camera at this time; enjoy!
Thanks for the help, everyone! I take a lot of long exposure shots at sunset, so I need to use the lowest ISO possible to reduce noise. I've always used ISO 200 on my D90. So, what I gather is that using Lo 1, for all intents and purposes, is just like ISO 100?
What's the verdict on increased noise using expanded Lo?
I was told that at Lo1, the sensor will overexposre the image and cut it back in processing. Therefore, noise will be lower, but you will have a tendancey to blow hightlights and there will be less contrast and dynamic range than at ISO 200.
LO 0.3–LO 1 The settings through correspond to ISO sensitivities 0.3–1 EV below ISO 200 (ISO 160–100 equivalent). Use for larger apertures when lighting is bright. Contrast is slightly lower than normal; in most cases, ISO sensitivities of ISO 200 or above are recommended.
The test here on Nikonians.org showed that the D90 has a higher dynamic range at Lo1 than at ISO200 and less noise (which is expected).
I was also told that,
"it makes an overexposed shot then pulls down the exposure in processing so I guess it would be more likely to blow highlights or at least have a bit less dynamic range. Don't use it unless you need a slow shutter speed, fast lens in bright light with fill flash (to match sync speed) or something specific like that.
The usual caveat in situations like this is that Lo1 doesn't have enough headroom in the sensor to warrant it as completely implemented...if you do use , the highlights will blow out more readily than if you shoot at ISO 200."
"In Nikon cameras the LO settings (-0.3 ... -0.7 and -1.0 EV) do not change the sensor sensitivity at all. What happens is that the camera stays at ISO 200 and overexposes the shot. After capture the digital image data is scaled back. You can achieve exactly the same end result by staying at ISO 200 and overexposing yourself and pulling back the exposure in Capture NX2 afterwards.
So, overexposing by 0.3 or 0.7 or 1.0 stops naturally means a bit less apparent noise, because you adjust the digital data afterwards. This is in a way the old ETTR principle with a deliberate overexposure/underdevelopment phase added in.
What you gain in noise you lose (and then some) in highlights and dynamic range. "
So, I'm getting different explinations from different sources. Or I am misinterpreting that information.
I could go out and do my own testing. But, I really don't have the time or energy to do that. I really just want to understand the differences in image quality between using ISO 200 and Lo1 on my D90, for shooting long exposure landscapes and waterscapes.
So, I'm getting different explanations from different sources.
The short answer is, that on this subject; you can disregard anything that conflicts with what I have said (really! )
The lower the ISO setting, the lower the noise, the higher the dynamic range; always. Below the lowest numbered ISO, the raw data is shifted toward the right; so there is an increased possibility of clipping highlights.
Mon 23-Feb-09 02:25 AM | edited Tue 24-Feb-09 09:46 PM by kluthage421
I have been experimenting with the lowest ISO possible and I must say I try to shoot at ISO 100 eq. when I can. I was wondering the same thing and the results from a recent trip to Carmel, CA are pleasing. ISO 200 has low noise on this wonderful camera anyway but, I will take as much noise reduction as I can get. Here is an ISO -1.0 ev (100 equiv.) shot straight from the camera.
"The sensor will stay at ISO 200 even with Lo -1.0 (ISO 100 equiv.) but the camera overexposes the image and then in digital domain reduces the exposure of the file by 1 stop.
You can do the exactly same thing by hand: * use ISO 200 with +1 EV * reduce exposure in Capture NX by 1 EV
This is same thing as shooting with Lo -1.0. The analog sensor gain or sensitivity does not change, it is already at the minimum at the native ISO.
Why less noise? Because of 1 stop hotter exposure. The risk is losing one stop of dynamic range in the highlights because of clipping.
Bill Claff's "measurements" in Nikonians regarding "photographic" dynamic range are questionable at best and he explicitly mentions, that his "measurements" are not scientific measurements but "photographic" deductions from the viewed image.
Common sense from a scientist's point of view says that clipping the highlights by 1 stop reduces dynamic range, although the "photographic" result will compensate this by gaining cleaner shadows."
I neither agree nor disagree to those statements. I am simply getting conflicting information from different sources and I'd like to know what each side has to say about it rather than blindly following other people.
I'm not trying to start a "flame war" or anything. But, you say red, another says blue. I'd like to know which it is without having to take sides.
Honestly, I should probably just call Nikon or contact someone in the company who may know for sure.
Yes, that is the chart. As you can see only the D300 and D90 show "real" improvement in the Lo range. But for sure, dynamic range is not lost going to lower ISOs; ever. (Any slight apparent loss for some models is simply due to the fact that values are being displayed as if they have more precision than they really do! Those values are probably "flat".)
BTW, if you run Windows (or perhaps Mono on Linux) you can use NefUtil at my site to measure Photographic Dynamic Range (PDR) and other values yourself. (Some switches are undocumented. Sorry, I'm behind on that!)
Thanks for the chart. It has a lot of info. Two questions - is this data for RAW capture, and does ‘@SNR 20' mean, at 20:1 signal to noise ratios? Following on the last question - how does this constant apply to the chart?
Here's something else that bewilders me. How much wood, could a woodchuck chuck, if a woodchuck could chuck wood? Even this doesn’t bug me as much as my ignorance on applying your fine graphic. It’s like termites in the wood to be chucked. How much additional wood could a woodchuck chuck, if termites lightened the load?
We can deal with ‘wood to tunnel’ ratios, and their impact on quantum chucking later. I'll be content with your reply on RAW capture, and the ‘@SNR 20' metric. Where was this graph found? Seriously, it’s a fine piece of work. I'd like to know its context.
I would like to chime in here, not to create an epiphany for anyone but to hopefully expand my own knowledge and correct any misconceptions I may have.
For film the ISO was calculated by analyzing density versus log of exposure. With digital the sensor sensitivity is calculated by looking at the output level of the sensor over a range of exposure levels close to the saturation point, and the amount of light falling on it at that point tells you the ISO level. And that is the base ISO as reported by manufactures. Well sort of in that the reported ISO is not always the actual ISO on all cameras.
Now by manufacture design film as well as the sensor has a set unchangeable sensitivity. But the difference being you can use different film, you’re stuck with one sensor. A way around this is that you can increase the gain on the single from the sensor effectively increasing sensitivity but not really. By decreasing the amount of light falling on the sensor you are decreasing the single and then by amplifying the single you bring it back to a level it would have been at the expense of noise and dynamic range. But that allows for the increase of ISO, but if you cannot change the sensitivity of the sensor then how do you artificially create a low ISO. Would you de-amplify or would you bit shift.
That might explain why manufactures such as Nikon use Hi1, 2, and 3 or Lo1, 2, 3. Well it appears that there is more then one way of effectively increasing sensitivity. One way is through analog amplification and the other through bit shifting. So perhaps Nikon reserves the actual ISO numbers for the analog amplification and then uses the Hi and Lo designation for the bit shift.
DXO Mark on their site list ISO 145 as the measured ISO for both the ISO 200 and Lo1 setting for the D90. And DPReview lists a decrease in dynamic range of .7EV between ISO 200 and Lo1 (you gain slightly on the low end but loose on the high end). So to me this indicates they may have used bit shifting and not de-amplification.
>Pete, > There's no need to speculate, I (and others) have >measured gain (analog amplification). >"Bit shifting" is used for Hi but Lo is NOT >"bit shifted" it is done with lower bit non-linear >amplification. > > >Bill >
Ah, I see. Do you also find the .7 loss in dynamic range with the Lo setting that DPReview reports?