Why don't DSLRs go low ? ASA 25 equivalent

[sanjosedave]

I'm getting ready to move from film to DSLR. I've always enjoyed shooting ASA 25, ASA 64 film/slides, but, I'm not finding low ASA equivalents in the DSLR world.

It seems all the talk/reviews is about high ISO and noise performance.

What am I not getting?

I shoot mostly on tripods

 

[bvukich]

Some have ISO50, but it's not really ISO50, it's just the digital equivalent of attenuation.

You're probably only the second person around here that's asked about low ISO, but I personally can't think of a single reason to need it (not saying reasons don't exist). So if you'd be so kind, why?

 

[sanjosedave]

In the film world, it was my understanding that ASA related to grain and light capture ability. For ASA 25, for example, grain was almost non-existent, but you needed a lot of light and/or long exposures. While ASA 400 was used by news togs because it was fast and allowed them to shoot with a smaller fstop.

My beginning understanding is that grain is to film as noise is to digital. So, if I could shoot using a low ASA, I could minimize grain(noise) before post production began.

 

[willhuff.net]

It would be really nice to have that really low iso for landscape work that needs a ND filter. It would also be great for balancing daylight with flash.

 

[benperrin]

I have to say that I have asked myself the same question. I would like it when trying to get longer shutter times (8 seconds or so) when shooting things like waterfalls during the day. The answer seems to be that you just have to invest in a good nd filter for your lens. But even with a nd filter and at f16 and iso 50 sometimes you still can't get the exact shutter speed that you want. I am using a B+W 8x ND Filter.

It is a rare occasion that I need this though and since nd filters are out there already most camera manufacturers probably think it is not worth pursuing. Some don't even do iso 50. Also iso 50 does not look as good as iso 100 so maybe that is another reason.

 

[Caps18]

At ISO 100, with a high resolution digital camera, in RAW I think you should check it out to see how it compares. I have used the L (?) setting on the 5Dm2 to get ISO50, but it isn't 'really' much different. I would have to go back and look at my Yosemite pictures again to compare them again though to make that statement.

But, yes, I agree that I would like more discussion about low noise in ISO 100 or possible 80, 50, etc modes.

 

[dr croubie]

Of course, my first question is, why would you want to?

In the film world, there's 2 main reasons to go slower film.
- Longer exposures. You can get longer exposures with slower film, without using ND filters or stopping down too far (if you can). Good for blurring waves into mush and star trails.
- Higher quality images, smaller grain size means less grainy, better "resolution" (and I'm not so sure about Dynamic range and tonal-response). Either way, the lower ISO/ASA you go, the better your prints.

Now, in the digital world, you have the same arguements, but it's a bit different:

- Longer exposures, yes. But then noise comes into play, unless you get some fancy fan-cooled Medium format back, the longer you're reading the sensor, the hotter it gets, and the noisier it gets (I'm not sure, but it could lead to more or the same noise as a shorter/higher-iso exposure. Depends how long you're talking). There's a school of thought concerning star-trails that it's better to take lots of 30s exposures and merge them all digitally (whether it's better or not, people do it). Of course, with digital you have to worry (more) about diffraction, so you can't stop down too far. But there's always ND filters anyway.

- Higher quality images? Depends what camera you're using. Take a look at this graph. Assuming the link works, it's a graph of the ISO response of the 5D2. Note how ISO50 is practically the same as ISO100?
Another graph here, is the Dynamic Range of the 7D. It doesn't get any better going from ISO200 to ISO100, going to 50 or 25 isn't going to do anything to make your images better.
OK, so I chose those 2 examples on purpose, not all digital cameras do that. But it's kind of indicative, you don't gain much in Image Quality by going to a slower ISO, not like you do/did with film.
Even the Phase One IQ 180 (the best sensor at DxOMark thus far) doesn't do any better at lower-ISO-response, but its Dynamic Range does get better all the way down to ISO25. But then, the price tag is the same as a very nice sports-car ...

 

[qwerty]

It is easy to simulate the effects of low ISO on noise and DR by combining multiple exposures. (You can even do this without getting the otherworldly HDR effects.)

As a good rule of thumb, if you combine 4 images, your signal to noise ratio will be improved by a factor of two, with a similar 1-stop boost in dynamic range.

The only disadvantage of doing this are more storage space, and your frames will be spaced a fraction of a second apart (depending on frame rate); that is not an issue for stationary scenes, but might be if you wanted an effect like motion blur.

For the OP (using a tripod), you should have no problem with just taking a few pictures and combining them later.

 

[Kernuak]

I sometimes use ISO 50 to blur water during the day, partly because I don't have a solid ND filter and partly because, it's one less piece of glass (although the effects of that glass would be minimal).

 

[MikeHunt]

I'm getting ready to move from film to DSLR. I've always enjoyed shooting ASA 25, ASA 64 film/slides, but, I'm not finding low ASA equivalents in the DSLR world.

It seems all the talk/reviews is about high ISO and noise performance.

What am I not getting?

I shoot mostly on tripods

Perhaps it has something to do with the physical constraints of the camera processor, namely; that high ISO is achieved by boosting light sensitivity 'electronically' not by using slower film (in the old actual physical world of photography). Maybe it is stretching the limitations of either the computing power or the end result is too black and shadowy pictures.

 

[pwp]

There is nothing you are not getting. The sensors are designed to deliver optimum results at 100-200 iso.

In the iso realm, comparisons with film have become practically meaningless. If your move from film is to something like a 5DIII you'll be knocked flat by the quality delivered by this camera with good glass at 100 iso compared to your best efforts with 25 iso film. Other posters have mentioned that if you require long exposures then ND filters are the way to go.

I saw digital files comfortably exceed anything I could ever have produced on any type of 35mm film right back when I bought a Canon D60 http://en.wikipedia.org/wiki/Canon_EOS_D60 10 years ago. Pretty soon clients were not asking for drum scanned medium format film any more. There was no need.

All the talk you mention about high iso and noise performance is because this is where the boundaries are being pushed. At low iso settings extremely high quality very low noise files have been available to us for years. There's little further conversation required as the results are already rock solid.

The one area of low iso performance that may need further discussion is highlight clipping. This is the one area where you may find drum scanned low iso film still has an edge. But there are plenty of post-pro techniques to easily manage this.

Paul Wright

 

[Mt Spokane Photography]

Think of the photosites on the sensor as buckets that catch photons. They can only accept a certain number of photons, and they are full, or saturated. That determines the lowest ISO. Anything that makes them less sensitive also reduces the high ISO. Manufacturers use as large of buckets as possible consistent with the number of photosites and the physical size of the sensor.

There is a solution, however, and that is to add a neutral density filter on your lens (like sunglasses) to cut down on the number of photons reaching the sensor so that you can use a wider aperture. Its similar to why you might wear sunglasses on a bright day, ... too much light.

 

[distant.star]

I agree with PWP's comments.

There seems to be no real correlation, or interest in one, comparing digital ISO to film numbers.

My experience has been, at least with B&W film, using ISO 50 gives me what I'd get with digital ISO 400 to 800 with a T2i. Using 400 film is like 3200 or higher on the camera. From a measurement or comparative standpoint, that really means nothing -- just my impressions with images I've made.

My advice -- don't even worry about it.

There is nothing you are not getting. The sensors are designed to deliver optimum results at 100-200 iso.

In the iso realm, comparisons with film have become practically meaningless. If your move from film to something like a 5DIII you'll be knocked flat by the quality delivered by this camera with good glass at 100 iso compared to your best efforts with 25 iso film. Other posters have mentioned that if you require long exposures then ND filters are the way to go.

I saw digital exceed anything I could ever have produced on any type of 35mm film right back when I bought a Canon D60 http://en.wikipedia.org/wiki/Canon_EOS_D60 10 years ago. Pretty soon clients were not asking for drum scanned medium format film any more. There was no need.

All the talk you mention about high iso and noise performance is because this is where the boundaries are being pushed. At low iso settings extremely high quality very low noise files have been available to us for years. There's little further conversation required as the results are already rock solid.

The one area of low iso performance that may need further discussion is highlight clipping. This is the one area where you may find drum scanned low iso film still has an edge. There are plenty of post-pro techniques to easily manage this though.

Paul Wright

 

[AJ]

I'd think blowout (overexposed/oversaturated channels) would be an issue out at 25 iso?

 

[epsiloneri]

The low-ISO limit is basically set by the number of electrons you can keep in a photo-site, as already pointed out. The number of electrons collected by the sensor is independent of the ISO-number, and once you fill the electron well you are saturated. You could artificially define a lower ISO, but that would limit your (numeric) dynamic range, so it would be pointless. Usually, ISO 100 corresponds to the gain when your numerical DR just covers the full electron well.

It works like this: when a photon hits the sensor, it generates an electron, a so called photo-electron, with a probability that is equal to the quantum efficiency (QE, typically around 50%, so every second viable photon generates an electron). These photo-electrons are collected by the photo-sites, but there is a limit to the total number of electrons that can be stored, the so-called full-well capacity, typically 30000-60000 electrons depending on pixel size. Ideally, you would like the A/D converter to count each electron, but unfortunately the bit depth is usually not sufficient for that. To get the best S/N, the best strategy is instead to let a group of electrons correspond to one digital unit (DU), so that the numeric dynamic range corresponds to the full well of electrons. I.e., if the full well is 60000 and you have 14 bits, using a gain of 4 electrons per DU gives you optimal S/N (and DR) for well-exposed images (since 4*2^14 is about 60000). If you are photon starved, on the other hand, so that you are far from filling up the electron well, then it can be advantageous to use a higher gain and let fewer electrons correspond to one DU. There's not much point in going beyond one DU per electron, however, which is called the "unity gain". Unity gain for the 5D2 corresponds to ISO 400; so using ISO 800 can still be advantageous for quantization reasons, but anything beyond is merely convenience and marketing: performance will be the same (except for the lower DR).

But you asked for the lower ISO limit. In that case it goes the other way: there's no point in increasing the number of electrons per DU beyond what you need to cover the full well. The result would only be that you are limited by the well depth instead of the numerical DR.

 

[scottkinfw]

ND filters can fill the gap.

The low-ISO limit is basically set by the number of electrons you can keep in a photo-site, as already pointed out. The number of electrons collected by the sensor is independent of the ISO-number, and once you fill the electron well you are saturated. You could artificially define a lower ISO, but that would limit your (numeric) dynamic range, so it would be pointless. Usually, ISO 100 corresponds to the gain when your numerical DR just covers the full electron well.

It works like this: when a photon hits the sensor, it generates an electron, a so called photo-electron, with a probability that is equal to the quantum efficiency (QE, typically around 50%, so every second viable photon generates an electron). These photo-electrons are collected by the photo-sites, but there is a limit to the total number of electrons that can be stored, the so-called full-well capacity, typically 30000-60000 electrons depending on pixel size. Ideally, you would like the A/D converter to count each electron, but unfortunately the bit depth is usually not sufficient for that. To get the best S/N, the best strategy is instead to let a group of electrons correspond to one digital unit (DU), so that the numeric dynamic range corresponds to the full well of electrons. I.e., if the full well is 60000 and you have 14 bits, using a gain of 4 electrons per DU gives you optimal S/N (and DR) for well-exposed images (since 4*2^14 is about 60000). If you are photon starved, on the other hand, so that you are far from filling up the electron well, then it can be advantageous to use a higher gain and let fewer electrons correspond to one DU. There's not much point in going beyond one DU per electron, however, which is called the "unity gain". Unity gain for the 5D2 corresponds to ISO 400; so using ISO 800 can still be advantageous for quantization reasons, but anything beyond is merely convenience and marketing: performance will be the same (except for the lower DR).

But you asked for the lower ISO limit. In that case it goes the other way: there's no point in increasing the number of electrons per DU beyond what you need to cover the full well. The result would only be that you are limited by the well depth instead of the numerical DR.

 

[sarangiman]

. I.e., if the full well is 60000 and you have 14 bits, using a gain of 4 electrons per DU gives you optimal S/N (and DR) for well-exposed images (since 4*2^14 is about 60000).

epsiloneri: First of all, very nice, intuitive, explanation!

I have a question: I know there are different methods to determine ISO of a sensor, but in your example above, for that system where a gain of 4 e- per DU gives you optimal SNR, is that defined as the 'base ISO'? Since you don't want to fill the wells anymore, you shouldn't really be allowed to go to a lower ISO (which'd cause the metering algorithm to increase the exposure).

You mention that for the 5D2, ISO800 is still useful for quantization, even though ISO400 is unity gain. But aren't higher ISOs still useful IF you're really really photon starved? For example, if after an exposure your fullest well only has 2000 e- on the 5D2, wouldn't you benefit from 1/8 e- being a DU (ISO 3200), thereby making that 'fullest pixel' translate to a DU of 16,000? Because then you have more bits to represent the data (the whole philosophy behind 'expose to the right')?

Thanks!

 

[sarangiman]

Also, then, if the 5D2 had a 16-bit ADC, then with the 60,000 well capacity, unity gain would correspond to ISO 100?

And just out of curiosity -- if you had a full well capacity of ~120,000, and the same 14-bit ADC, and therefore used 8e- per DU (ISO 100), you'd still get a cleaner image than the 5D2 with 60,000 well capacity, just b/c of increased DNR, yes? Though, of course ideally you'd want a higher ADC.

Which makes me wonder why Canon doesn't use 16-bit ADCs with sensors that have full well capacity of ~60,000?

 

[torger]

Some older medium format digital backs have low ISO ratings. But this is because less good quantum efficiency (not all photons are registered), which also makes them really bad in low light.

To get the same effect with a modern sensor you can put on a ND filter.

 

[epsiloneri]

ND filters can fill the gap.

Yes, ND filters allow you to extend the exposure, but they don't improve your S/N (because that is essentially determined by how many photo-electrons you can collect). Another way to artificially "lower" the ISO (not really, of course) with the goal of actually improve the S/N would be to take multiple exposures of the same scene, reading out the collected electrons in between. If you have a static scene that is actually a viable way to improve the S/N (often used in astrophotography).