Ever increasing ISO range

Steve

Steve

Suspended / Banned
Messages
1,685
Name
.... Steve
Edit My Images
Yes
Camera manufacturers seem to be increasing the usable ISO range so that I can now handhold my digital camera in very low light with little noise or quality loss which is great.

But all the ISO expansion seems to be one way ie increasing the ISO. Not much seems to be happening in offering lower ISOs.

I can still buy film with ISOs of 25 and 16 (maybe lower if I searched around), but my digital camera comes nowhere close to such an ISO.

So is it because there isn't the same customer demand for lower ISOs, or are there some technical limitations that make it easier to increase the ISO but harder to decrease it?
 
Not sure about the reasons for the limitations, Steve. My D800E has a standard range of 100 - 6400, but expandable to 50 - 25,600. as far as I understand setting the camera to ISO 50 does not produce a better quality image than ISO 100.
Yes, it would be handy to go lower to aid the use of longer shutter speeds when the shot called for it.

I think the constant increase in ISO range will eventually hit a ceiling (maybe not until we can use 1/8000 at f22 in an unlit coal cellar - ha-ha!)
 
I asked this very question albeit ages ago and there were reasons why going down the way was an issue. I can't recall what it was though.

It would be nice to shoot filterless longish shutter speeds to record movement whilst retaining shallow dof with wider apertures rather than using f22

I guess we can always shoot at night.
 
Steve said:
Camera manufacturers seem to be increasing the usable ISO range so that I can now handhold my digital camera in very low light with little noise or quality loss which is great.

But all the ISO expansion seems to be one way ie increasing the ISO. Not much seems to be happening in offering lower ISOs.

I can still buy film with ISOs of 25 and 16 (maybe lower if I searched around), but my digital camera comes nowhere close to such an ISO.

So is it because there isn't the same customer demand for lower ISOs, or are there some technical limitations that make it easier to increase the ISO but harder to decrease it?
Click to expand...


ISO of a film stock is a measure of how reactive to light a chemical emulsion is.

ISO in digital is made up. It sets the gain of the CMOS amplifier. Higher gain, higher noise figure, worse signal to noise ratio. To go lower in fake ISO, you'd need to attenuate the signal. You can't do it the old fashioned way as resistors are a bad idea in CMOS.
 
Steve said:
are there some technical limitations that make it easier to increase the ISO but harder to decrease it?
Click to expand...

As I understand it, 'expanded' ISO settings are achieved essentially by amplifying or reducing the signal from the base ISO (which is commonly ISO 200 on modern digital cameras).

When you're expanding upward, the result is that the image is under-exposed and the signal boosted (the read values are multiplied) to the match the desired ISO setting. Since the 'roll off' of dynamic range is slower at the bottom end of the exposure range (in dark tones you have only a few photons hitting the senel during an exposure) this results in noisier shadows when the ISO is expanded upward. This has a relatively low impact on image quality as noise reduction techniques can mitigate its effect.

However, when expanding downward, the image is over-exposed at base ISO and the signal values then divided to match the lower ISO setting. If you were shooting RAW it is the same as over-exposing by one stop at 100 ISO and then darkening the image in by one stop in PP to achieve ISO 50.

As the roll off at the top end of the tone values is relatively sharp (once the sensels have captured the maximum amount of light that they can handle, they can't read any more and that is a hard limit) you may then run into problems with the brightest tones in the scene being clipped. Blown tones cannot be recovered and the impact on image quality is highly negative.

Consequently, it is easier to offer 'expanded' ISO range upward to faster ISO speeds than downward to slower ISO settings.
 
Last edited:
The obvious solution to the lower iso issue is to have an inbuild ND filter like I think fuji has made. If you had some kind of lcd material you could make it so that it was even graduated or only on part of the sensor image.
 
arclight said:
Not sure about the reasons for the limitations, Steve. My D800E has a standard range of 100 - 6400, but expandable to 50 - 25,600. as far as I understand setting the camera to ISO 50 does not produce a better quality image than ISO 100.
Yes, it would be handy to go lower to aid the use of longer shutter speeds when the shot called for it.

I think the constant increase in ISO range will eventually hit a ceiling (maybe not until we can use 1/8000 at f22 in an unlit coal cellar - ha-ha!)
Click to expand...

ISO 50 is simulated - it is not "real". I imagine they take two exposures and pretty much average them out or overexpose and recover on chip.

I am more than certain big chip makers can produce ridiculous low ISO cameras (and they may do for specialist applications), but in practices that would mean much lower high ISO. I can't see many photographers buying one of those. It would be near useless for conventional applications.

The way forward maybe advanced software processing and combined exposures. I would watch Samsung's android mirrorless line - due to open nature it may be the first one to have any serious FX applied. Oh, and that will need electronic shutter - I am not sure why we are still stuck with mechanics
 
I could conceive of a technical mechanism to do it without clipping or ND filters.

If it were possible to monitor the read state of all the sensels on a sensor during the image capture stage, the it might be possible to set a clipping threshold at which point the state of the entire image is captured when a significant number of sensels pass that threshold. This partial exposure is stored and the exposure resumed from zero again. This cycle continues until the entire exposure time at the desired (expanded) ISO is complete.

Then the series of partial images are combined and the values of each sensel averaged over the course of the entire combined exposure at the expanded ISO setting. Essentially, it would be the same Image Stacking technique as used by astrophotographers to capture images of faint objects in the night sky but done in camera. Providing the read/store/reset of the imaging sensor could be done quickly enough relative to the total exposure time, it would give a reasonably accurate simulation of a longer exposure at lower ISO.
 
Thanks for the replies. Intereresting stuff. I don't pretend to understand all of the science, but get the overall picture.

It never ceases to amaze me the technical knowledge and expertise of members of this forum and their willingness to share.....thanks all.
 
Musicman said:
I could conceive of a technical mechanism to do it without clipping or ND filters.

If it were possible to monitor the read state of all the sensels on a sensor during the image capture stage, the it might be possible to set a clipping threshold at which point the state of the entire image is captured when a significant number of sensels pass that threshold. This partial exposure is stored and the exposure resumed from zero again. This cycle continues until the entire exposure time at the desired (expanded) ISO is complete.

Then the series of partial images are combined and the values of each sensel averaged over the course of the entire combined exposure at the expanded ISO setting. Essentially, it would be the same Image Stacking technique as used by astrophotographers to capture images of faint objects in the night sky but done in camera. Providing the read/store/reset of the imaging sensor could be done quickly enough relative to the total exposure time, it would give a reasonably accurate simulation of a longer exposure at lower ISO.
Click to expand...


That's roughly how the RED Epic does HDR video, sampling twice per shutter opening.
 
Musicman said:
I could conceive of a technical mechanism to do it without clipping or ND filters.

If it were possible to monitor the read state of all the sensels on a sensor during the image capture stage, the it might be possible to set a clipping threshold at which point the state of the entire image is captured when a significant number of sensels pass that threshold. This partial exposure is stored and the exposure resumed from zero again. This cycle continues until the entire exposure time at the desired (expanded) ISO is complete.

Then the series of partial images are combined and the values of each sensel averaged over the course of the entire combined exposure at the expanded ISO setting. Essentially, it would be the same Image Stacking technique as used by astrophotographers to capture images of faint objects in the night sky but done in camera. Providing the read/store/reset of the imaging sensor could be done quickly enough relative to the total exposure time, it would give a reasonably accurate simulation of a longer exposure at lower ISO.
Click to expand...

Wouldn't this need electronic shutter? They are so overdue in dslrs
 
daugirdas said:
Wouldn't this need electronic shutter? They are so overdue in dslrs
Click to expand...

They have already been used in at least two DSLRs that I can think of, the Nikon D40 (not the 40x) and the Nikon D70 which is how they achieved the incredible 1/500 flash sync speed that they have. It was a hybrid mechanical/electronic shutter affair. Subsequent models unfortunately dropped this system though.
 
Neither of those rely on mechanical shutters, only relatively long exposures to allow movement to show. IIRC, the Lartigue effect is where the shutter slit moving shows fast moving subjects to lean in the direction of travel since the slit moves up during the exposure. It was this effect that led cartoonists and others to draw cars etc with oval wheels. The way this effect usually shows these days is in camera phone videos of props - they show as curved!
 
boyfalldown said:
Ot, but they're not. Think of all the techniques that need a mechanical shutter
Click to expand...

Rolling shutter in video and special effects at very high shutter speeds? It would depend on e-shutter implementation and scan sequence but this should be a thing of the past
 
IMVHO this is the biggest issue with the latest batch of CSC. I'm just not interested in a serious camera with a base ISO of 200 and a max shutter speed of 1/4000 and ISO 200 and 1/8000 seems hardly adequate either and in it fact seems like a con when the manufacturer and review site shout about 1/8000 and fail to mention that the base ISO has risen to 200, how's that better than ISO 100 and 1/4000 I wonder.

Whilst it's true that years ago a max shutter speed of 1/4000 seemed like pure luxury and whilst many seem happy to shoot at f8 all day I'm not and these days f8 seems very limiting when we are spoilt with widely available and affordable wide aperture lenses. Although using ND's makes wider aperture shooting in good light possible personally I find it a major PITA and having to repeatedly fit and remove ND's feels like a throw back in time to me... and a PITA.

I blame marketing as they love a number to push and to them ISO 20,000,000 looks so much better printed on the side of the box than ISO 50 :mad: It's pushing me out of MFT as I'm sick of wrestling with camera hoods, bags, ND cases and ND's. Buy our lovely f1.4 lens and use it at f8... No thanks, no sale.
 
Last edited:
Personally, I would rather have the ability to have a very fast ISO but have to use filters to get longer exposures. At least it's possible to filter to get slower shutter speeds if wanted/needed - not an option available if there's not enough light! Higher ISOs have made shots that were all but impossible a few years ago almost commonplace and while it IS a faff having to use filters, (as I've said) at least it's an option.
 
woof woof said:
IMVHO this is the biggest issue with the latest batch of CSC. I'm just not interested in a serious camera with a base ISO of 200 and a max shutter speed of 1/4000 and ISO 200 and 1/8000 seems hardly adequate either and in it fact seems like a con when the manufacturer and review site shout about 1/8000 and fail to mention that the base ISO has risen to 200, how's that better than ISO 100 and 1/4000 I wonder.

Whilst it's true that years ago a max shutter speed of 1/4000 seemed like pure luxury and whilst many seem happy to shoot at f8 all day I'm not and these days f8 seems very limiting when we are spoilt with widely available and affordable wide aperture lenses. Although using ND's makes wider aperture shooting in good light possible personally I find it a major PITA and having to repeatedly fit and remove ND's feels like a throw back in time to me... and a PITA.

I blame marketing as they love a number to push and to them ISO 20,000,000 looks so much better printed on the side of the box than ISO 50 :mad: It's pushing me out of MFT as I'm sick of wrestling with camera hoods, bags, ND cases and ND's. Buy our lovely f1.4 lens and use it at f8... No thanks, no sale.
Click to expand...

Isn't the aperture you want to shoot at, more to do with the DOF you want?
:)
 
Last edited:
simonblue said:
Isn't the aperture you want to shoot at, more to do with the DOF you want?
:)
Click to expand...

Errrr... It's related, of course, and with the smaller sensor systems like MFT and APS-C if you are going for a reduced DoF look you are probably going to be using wider apertures than you would with a larger format system. Do the math. The crop factor works for DoF so if you'd be using f2.8 and 1/4000 sec with a FF camera what would you be using with a MFT? I'll give you a clue... The answer is ND's :D
 
Nod said:
Personally, I would rather have the ability to have a very fast ISO but have to use filters to get longer exposures. At least it's possible to filter to get slower shutter speeds if wanted/needed - not an option available if there's not enough light! Higher ISOs have made shots that were all but impossible a few years ago almost commonplace and while it IS a faff having to use filters, (as I've said) at least it's an option.
Click to expand...

The thing with filters is that they detract from some of the advantages of the smaller systems. Filters add weight (a little...) and bulk as you have to take them with you and they have to retrieved, fitted and removed again and that detracts from the discreteness of the smaller systems.

When out with my MFT camera and a wide aperture lens I find myself needing an ND for maybe 1 shot is 3 and personally I find them a major irritation.
 
Surely you're getting into tripod territory by then so the weight and indiscreetness an ND will add is relatively small. Say you're meter tells you you need 1/30 @ f/1.4 at ISO 1600 (and have no tripod with you) but the subject demands a faster aperture or shutter? No option if your maximum ISO is 1600 but possibilities if you can get cleanish results with higher ISO. In an ideal world we would have the full range from ISO 1 up to 25,600 (or higher) all giving lovely clean results. Until then we have to make some compromises!
 
Rarely used 50 let alone needed lower - I do need to use 4/5k, 6400 tho
 
At our camera club a few weeks ago we had a talk entitled "Art, craft and technology" where the speaker looked at technological developments in the photographic area and how they impacted on what photographers actually do. For example the large capacity of memory cards compared to film has revolutionised underwater photography, and the high ISO capabilities of modern DSLRs allow photographers to capture images which simply wouldn't be possible on film.

He also extrapolated from current trends and predicted that, within 9 years, we'll have DSLRs offering a maximum ISO of 6,000,000 and 16 stops of dynamic range.

I disagreed that it would go that far, and I have a date in my diary for September 2022 when beer will be bought depending on who turns out to be right.

But on a slightly serious note, all sensors do is capture and count photons, and so logically there must be a physical limit where every single available photon is being captured and counted. I've been trying to work out where that limit is - i.e. what ISO value it would equate to. I suspect it's some way short of 6,000,000 for the kinds of pixel densities we're used to - but it's proving to be surprisingly difficult to work out. Does anyone know of this exercise having been performed? Or can anyone point to a good reference about photons, quantum efficiency of sensors, and such like?
 
At a guess, at gigantinormous sensitivities, there may be problems with radiation other than visible light penetrating through the material the body's made of. ISTR reading something about this radiation being the cause of film degradation even when stored chilled (or even frozen) - without thick layers of lead (or similar), x-rays etc cause fogging.
 
Nod said:
Surely you're getting into tripod territory by then so the weight and indiscreetness an ND will add is relatively small. Say you're meter tells you you need 1/30 @ f/1.4 at ISO 1600 (and have no tripod with you) but the subject demands a faster aperture or shutter? No option if your maximum ISO is 1600 but possibilities if you can get cleanish results with higher ISO. In an ideal world we would have the full range from ISO 1 up to 25,600 (or higher) all giving lovely clean results. Until then we have to make some compromises!
Click to expand...

The weight is not really an issue, that was a little tongue in cheek :D but certainly there's bulk and you can't just put an ND in your pocket so it has to be in a case in your pocket or in your camera bag and of course when you want it it has to be removed from your pocket or bag and then removed from its case and in the meantime you have to remove your lens hood and find somewhere to put it and taking the ND filter out of its case will be a two handed operation unless you're Paul Daniels... and when you've taken that shot you may just point your camera at something else and find that your shutter speed has dropped too low or your ISO has risen a little too high and you may decide to take the ND off again and so you recover its case and take the ND off the lens and put it in the case which is all another two handed operation and you put the ND case back in your pocket or bag and refit your lens hood and all this takes time and all this fiddling with bags and cases and lens hoods and ND's may attract attention and personally I'd rather have ISO 100 and 1/8000 available just as I had 8 years ago when I bought my 20D.

I realise that I'm in a minority here but I like to shoot discretely and relatively quickly and for those reasons I do regret the loss of lower sensitivities and the inability to shoot with wide apertures in good light without the palava and pantomime of ND's.

For a scenic shot with the use of a tripod I have no issue with ND's but for walkabout hand held photography involving changing scenes and changing lighting I personally find them a PITA.

Of course there's always the option of shooting at f8 all day... but my compromise is to not buy a camera with a base ISO of 200 and a max shutter speed of either 1/4000 or 1/8000, my compromise is to buy something which will let me shoot how I want to so my money didn't go to Fuji, Olympus or Panasonic.
 
Last edited:
I do take your point and as I said, in an ideal world, we would have all options available! We'd also have 8-1600mm f/1 lenses weighing 28g and measuring 25mm (with corner to corner sharpness, naturally!) On the rare occasions I've needed a lower ISO than 200, I've been able to dial in Lo to replicate 100. Not being a fan of creamy smooth water, I've never needed to go lower than 100 (although I did used to use Velvia 50 in the days of film.) Of course, there is still that option (if your lenses are FF compatible), a good film body would almost certainly be very cheap and a roll of Velvia 50 wouldn't break the bank!
 
Yes, there is always the option of buying something else and that's what I've done.

I have tried to love the latest Fuji, Oly and Panny offerings but the move away from ISO 100 puts them off my radar as I want to shoot how I want to not how a base ISO of 200 dictates. Stuck with that must be like being back in 1952 and after so many years with cameras with lower base senstivities this new base seems very much a backward step to me but fine if you shoot at f8 all day when we remember that with, for example MFT, f4 could give you front to back DoF and yet at ISO 200 f4 could be unusable without ND's! Remarkable... and forces the user to use ND's or head towards apertures which bring you to diffraction.

As I said, IMHO and for me.
 
Last edited:
The thing is, Alan, the cameras of seven to ten years ago that were base ISO 100 often had only 6 stops of dynamic range (the 1Ds Mk1, for example) so you had to treat them like slide film or suffer the consequences. Straying even just a little from the base might produce noise and potentially other problems.

Modern sensors may have a base ISO of 200, but they're achieving even over 14 stops of DR (Nikon D800 gets 14.4) so even if you did have to sacrifice a little of that to use ISO 100 or 50 in 'expanded' ISO ranges, you're very much still well ahead of the game.

DxO don't test down to ISO 50, but looking at a small sample their charts for recent bodies (Canon 70D, 5D III, Nikon D800 and the Oly OM-D) they actually achieve their maximum dynamic range at ISO 100, whatever their 'base' ISO may be.
 
Last edited:
Musicman said:
The thing is, Alan, the cameras of seven to ten years ago that were base ISO 100 often had only 6 stops of dynamic range (the 1Ds Mk1, for example) so you had to treat them like slide film or suffer the consequences. Straying even just a little from the base might produce noise and potentially other problems.

Modern sensors may have a base ISO of 200, but they're achieving even over 14 stops of DR (Nikon D800 gets 14.4) so even if you did have to sacrifice a little of that to use ISO 100 or 50 in 'expanded' ISO ranges, you're very much still well ahead of the game.

DxO don't test down to ISO 50, but looking at a small sample their charts for recent bodies (Canon 70D, 5D III, Nikon D800 and the Oly OM-D) they actually achieve their maximum dynamic range at ISO 100, whatever their 'base' ISO may be.
Click to expand...


I suppose he other question is, what dynamic range can you display on professional paper? on a computer monitor? How many bits do you need in sRGB to accurately store that dynamic range?

At 14 stops it's closer to the limits of the human visual system. Which is good. But I haven't seen any similar increments in display technology and am unsure what the visual effect of compressing dynamic range from say 14 stops to 8 stops will be.
 
To go back to the OP (and btw the question interests me too), I think it's related to the fact that higher ISOs (whatever that may be in digital, StewartR pointed to Clarkvision which is quite a read in that regard) are easier (cheaper) to achieve. It's similar to the MP war (which I hope has calmed down now) vs. dynamic range. I'd rather have a sensor of a given size with proper ISO 50 and the widest DR possible than one with several hundred thousands ISO and 5-10 MP more. But I am obviously a minority.
 
I think the dynamic range of my eyeballs has reduced over the years, I need a lot more contrast now to be able to read without my glasses :D
 
The dynamic range of my other balls has also reduced over the years.
 
st599 said:
I suppose he other question is, what dynamic range can you display on professional paper? on a computer monitor? How many bits do you need in sRGB to accurately store that dynamic range?

At 14 stops it's closer to the limits of the human visual system. Which is good. But I haven't seen any similar increments in display technology and am unsure what the visual effect of compressing dynamic range from say 14 stops to 8 stops will be.
Click to expand...

If we were printing 14bit files straight out of camera I would agree with you. However this is not the case. Some scenes (e. g. sunsets) have incredibly wide DR, and it actually takes far more than 14bit to cook a printable file. That's why we use grad filters, and 16Bit MF cameras look very desirable for landscape and product photography. I could only hope the next big canon will do this.

arclight said:
The dynamic range of my other balls has also reduced over the years.
Click to expand...

Thank you for sharing it with us
 
You must log in or register to reply here.
Back
Top