Forgot Password
Pentax Camera Forums Home
 

Reply
Show Printable Version Search this Thread
08-29-2018, 07:12 AM   #1
Pentaxian




Join Date: Mar 2015
Photos: Gallery | Albums
Posts: 6,381
"I had a dream." What determines the base ISO of a digital sensor?

@Digitalis and other technical experts might want to chime in on this one.

If I wanted a digital sensor with the LOWEST base ISO possible, for slow-speed tripod work (e.g. diffusing water, "eliminating" crowds) or for wide-open super-bokeh work with very fast lenses in bright sunlight, what are the determining factors behind getting it (without having to put a gazillion-stop ND filter in the way and screw my focus up)?

I'm talking SINGLE FIGURES here, certainly not higher than 25, with NR algorithms keeping it usable up to maybe 800 or 1600.

08-29-2018, 07:24 AM   #2
Veteran Member
CarlJF's Avatar

Join Date: Sep 2014
Location: Quebec City
Posts: 1,185
I think that "base ISO" means the sensitivity giving the best signal/noise ratio. That said, with ISO-invariant modern sensors, it's more or less irrelevant. I guess that now the lowest selectable ISO is more a decision of the manufacturer than a technical limitation of the sensor. Although there's certainly is a point under which the sensor isn't ISO invariant anymore, it's almost certainly much lower than ISO100...

Last edited by CarlJF; 08-29-2018 at 07:45 AM.
08-29-2018, 07:43 AM   #3
Site Supporter
Site Supporter
jatrax's Avatar

Join Date: May 2010
Location: Washington Cascades
Photos: Gallery | Albums
Posts: 12,991
QuoteOriginally posted by pathdoc Quote
gazillion-stop ND filter in the way and screw my focus up
Not really an answer to your question but I routinely use live view AF with a 10 stop ND. That is on a tripod and under landscape conditions of course.

Sorry for the interruption, carry on.
08-29-2018, 08:11 AM   #4
Site Supporter
Site Supporter
Digitalis's Avatar

Join Date: Mar 2009
Location: Melbourne, Victoria
Photos: Gallery
Posts: 11,694
In the early days of digital SLRs Circa 2004 there were cameras such as the Kodak SLR/N which had a base ISO of 6 - however, that was in part due to the horrendous inefficiencies of the sensor design and supporting electronics. Sensors these days are very efficient at gathering light - I recall quantum Efficiency figures upwards of 70% for a monochrome sensor and around 60% for a sensor equipped with a Bayer CFA.

These days filters are required to reduce light so super-fast lenses can be used wide open in daylight, Cinematographers use these light attenuating filters quite a bit but for different reasons from us photographers - their use has only fairly recently caught on with photographers and the ever increasing need for subject separation - or use motion blur as an effect. There are some companies that can replace part of the filter stack in front of a camera sensor with a strong ND filter* This eliminates the need for screw in filters but it will throw off the cameras internal AE metering quite badly.



* OR IV/IR pass filter if you so desire. i'm not 100% sure if they are still in business.


Last edited by Digitalis; 08-29-2018 at 08:24 AM.
08-29-2018, 08:17 AM   #5
Pentaxian
photoptimist's Avatar

Join Date: Jul 2016
Photos: Albums
Posts: 5,113
The easiest way to have a really low base ISO is to lower the quantum efficiency of the sensor -- if the photons don't get converted to electrons, the sensitivity is really low. Of course, then the low-light performance suffers.

If you want a sensor with modern quantum efficiency (for good low-light performance) but don't want it to saturate in high light levels, then you need to increase the electrical capacitance of the pixels. That implies building a high surface-area capacitor into each tiny surface-area silicon pixel. For example, one could imagine fabricating tiny tantalum or super-capacitor elements on top of each pixel (to store al the extra electrons generated by longer exposures in bright light) and then flipping the sensor chip for back-side illumination.

P.S. Film has really low quantum efficiency -- about 1/25th that of silicon sensors -- which is why it has such low base ISO and such humongous grain for high ISO films.
08-29-2018, 08:17 AM   #6
Pentaxian
normhead's Avatar

Join Date: Jun 2007
Location: Near Algonquin Park
Photos: Gallery | Albums
Posts: 40,442
QuoteOriginally posted by CarlJF Quote
I think that "base ISO" means the sensitivity giving the best signal/noise ratio. That said, with ISO-invariant modern sensors, it's more or less irrelevant. I guess that now the lowest selectable ISO is more a decision of the manufacturer than a technical limitation of the sensor. Although there's certainly is a point under which the sensor isn't ISO invariant anymore, it's almost certainly much lower than ISO100...

Careful with this. The biggest reason for using base ISO is it gives you the best signal to noise, which by definition gives you the most dynamic range. DR decreases as ISO increases. IN that sense ISO invariance is a complete hoax. The noise and sharpness doesn't vary, but the dynamic range does.

I have no idea why people invented this term.
08-29-2018, 08:18 AM   #7
Veteran Member
Mark Ransom's Avatar

Join Date: Jul 2015
Location: Minneapolis, MN, USA
Photos: Albums
Posts: 498
QuoteOriginally posted by CarlJF Quote
I think that "base ISO" means the sensitivity giving the best signal/noise ratio. That said, with ISO-invariant modern sensors, it's more or less irrelevant. I guess that now the lowest selectable ISO is more a decision of the manufacturer than a technical limitation of the sensor. Although there's certainly is a point under which the sensor isn't ISO invariant anymore, it's almost certainly much lower than ISO100...
I would agree with your first sentence - the base ISO is the point where you've achieved the maximum dynamic range, which is determined by the signal/noise ratio. Assuming the noise floor is constant, it would be the point where you have the maximum amount of signal before overloading the pixel cells. That would certainly be a technical limitation of the sensor.

My understanding of ISO invariance is different, it's all about increasing ISO. It means that increasing the ISO is no different than simply turning up the brightness during the conversion of the RAW image.

08-29-2018, 08:24 AM   #8
Pentaxian
normhead's Avatar

Join Date: Jun 2007
Location: Near Algonquin Park
Photos: Gallery | Albums
Posts: 40,442
QuoteOriginally posted by Mark Ransom Quote
I would agree with your first sentence - the base ISO is the point where you've achieved the maximum dynamic range, which is determined by the signal/noise ratio. Assuming the noise floor is constant, it would be the point where you have the maximum amount of signal before overloading the pixel cells. That would certainly be a technical limitation of the sensor.

My understanding of ISO invariance is different, it's all about increasing ISO. It means that increasing the ISO is no different than simply turning up the brightness during the conversion of the RAW image.
I haven't seen any evidence to suggest that's even possible.
It would only be relevant to an image that would use it's full dynamic range. If an image requires 9 EV DR, and your seniors produces 15 EV, then you can expect a lot of loss of EV before the loss of DR becomes visible. All the images I've seen in support of ISO invariance didn't stress the DR, as almost every sunset image or image with bright sun and depp shadows does.

So in that sense it appears to me to be slight of hand. Use narrow DR image and you can convincingly argue there's such thing as ISO invariance. Using images that are improved by more DR or degraded by the lack of it, and I'm guessing ISO invariance disappears. But as I said, it's always appeared to be nonsense to me so I pretty much ignore it.

I just always shoot at the lowest ISO possible, if that's no longer necessary, someone should let us know. I find it hard to believe a 400 ISO image looks the same as an under exposed 100 ISO image bumped up 2 stops in post.

Has anyone actually done this?

The only examples I've seen showed distinct change in contrast values, which the author conveniently ignored in arguing for ISO invariance. IF it's not exactly the same image, there is variance. Not only that the author didn't seem to understand the effect on contrast of sliding the curves towards the left, making the whole concept meaningless.

Last edited by normhead; 08-29-2018 at 08:34 AM.
08-29-2018, 08:28 AM   #9
Veteran Member
CarlJF's Avatar

Join Date: Sep 2014
Location: Quebec City
Posts: 1,185
QuoteOriginally posted by Mark Ransom Quote
My understanding of ISO invariance is different, it's all about increasing ISO. It means that increasing the ISO is no different than simply turning up the brightness during the conversion of the RAW image.
On paper, it works both ways: you could just as well overexpose a picture and pulling it back to "normal" exposure level. However, in the real world, sensor has a limit on how much light they can capture. And, as we all know (or should know!), overexposing a picture increase the risk of clipping the highlights. Thus, yes,in practice we usually talk of ISO invariance for pushing up the ISO from an underexpose picture and not the other way.

---------- Post added 08-29-18 at 11:40 AM ----------

QuoteOriginally posted by normhead Quote
Careful with this. The biggest reason for using base ISO is it gives you the best signal to noise, which by definition gives you the most dynamic range. DR decreases as ISO increases. IN that sense ISO invariance is a complete hoax. The noise and sharpness doesn't vary, but the dynamic range does.

I have no idea why people invented this term.
No Norm, it's not an hoax. Sure, the same picture taken at different ISO at the same exposure will have different DR. But ISO invariance rather talk about two pictures, at different ISO, but over/underexposed by the same amount of stops. Or, said otherwise, for a same amount of light the DR and noise will be the same (although exposure will be wrong).


In practice, ISO invariance just means that if the aperture and shutter speed stay the same, the resulting image will have the same DR and noise level when brought back to the same exposure in PP.
08-29-2018, 08:42 AM   #10
Site Supporter
Site Supporter
RGlasel's Avatar

Join Date: Sep 2013
Location: Saskatoon
Photos: Gallery | Albums
Posts: 3,225
QuoteOriginally posted by CarlJF Quote
I guess that now the lowest selectable ISO is more a decision of the manufacturer than a technical limitation of the sensor.
It certainly is for values like ISO 64 or 80. If the manufacturer is willing to sacrifice the higher ISO values, it is possible to program the camera to work with lower minimum ISO values. From the sensor's viewpoint, all that matters is it's ability to produce a signal at different light intensities. Everything after that point is mathematics. The sensor's ability to produce a signal depends on the bias voltage supplied to it, which is determined by the camera's metering system. ISO settings are instructions to the camera's processor to set aperture and shutter speed as if there was film with that ISO rating being exposed; it is not a quality of the sensor.

For situations where extremely low ISO settings are desirable, it may be that sensors with less ability to produce a signal from low light intensities will produce better results, but those sensors will not be able to work with the same range of ISO settings as the more sensitive sensors. There is also a limitation on how frequently the camera can sample the signal from the sensor, but my guess is that if the light intensity is high enough to encounter that limitation, the photographer should be wearing welding goggles or operating the camera remotely.
QuoteOriginally posted by pathdoc Quote
I'm talking SINGLE FIGURES here, certainly not higher than 25, with NR algorithms keeping it usable up to maybe 800 or 1600.
Moving the range of allowable ISO settings down by 4 stops is certainly possible, but at what cost? Who is going to buy a purpose built camera that is useless above ISO 800 when you can accomplish much more with ND filters and easily switch back to the camera's normal ISO range when you want to?
08-29-2018, 08:43 AM - 1 Like   #11
Veteran Member
CarlJF's Avatar

Join Date: Sep 2014
Location: Quebec City
Posts: 1,185
QuoteOriginally posted by normhead Quote
I just always shoot at the lowest ISO possible, if that's no longer necessary, someone should let us know. I find it hard to believe a 400 ISO image looks the same as an under exposed 100 ISO image bumped up 2 stops in post.

Has anyone actually done this?
Believe it or not, that's the case, if the sensor is ISO invariant or close enough to it. You can find plenty of articles on this or just try it yourself.
08-29-2018, 08:45 AM   #12
Veteran Member
Mark Ransom's Avatar

Join Date: Jul 2015
Location: Minneapolis, MN, USA
Photos: Albums
Posts: 498
QuoteOriginally posted by normhead Quote
I haven't seen any evidence to suggest that's even possible.
It would only be relevant to an image that would use it's full dynamic range. If an image requires 9 EV DR, and your seniors produces 15 EV, then you can expect a lot of loss of EV before the loss of DR becomes visible. All the images I've seen in support of ISO invariance didn't stress the DR, as almost every sunset image or image with bright sun and depp shadows does.

So in that sense it appears to me to be slight of hand. Use narrow DR image and you can convincingly argue there's such thing as ISO invariance. Using images that are improved by more DR or degraded by the lack of it, and I'm guessing ISO invariance disappears. But as I said, it's always appeared to be nonsense to me so I pretty much ignore it.

I just always shoot at the lowest ISO possible, if that's no longer necessary, someone should let us know. I find it hard to believe a 400 ISO image looks the same as an under exposed 100 ISO image bumped up 2 stops in post.

Has anyone actually done this?

The only examples I've seen showed distinct change in contrast values, which the author conveniently ignored in arguing for ISO invariance. IF it's not exactly the same image, there is variance. Not only that the author didn't seem to understand the effect on contrast of sliding the curves towards the left, making the whole concept meaningless.
Just to make sure I wasn't speaking out of the wrong orifice, I went looking for an article on the subject. Here's a good one, complete with sunset pictures.

ISO Invariance Explained - Photography Life
08-29-2018, 09:02 AM   #13
Site Supporter
Site Supporter
RGlasel's Avatar

Join Date: Sep 2013
Location: Saskatoon
Photos: Gallery | Albums
Posts: 3,225
QuoteOriginally posted by Mark Ransom Quote
the base ISO is the point where you've achieved the maximum dynamic range, which is determined by the signal/noise ratio.
The minimum ISO setting is normally set to where the sensor produces the maximum dynamic range, but it doesn't have to be. ISO is a setting, not a physical quantity. Very small electrical circuits can only handle very small levels of current, which limits the dynamic range of the signal carried by those circuits; the difference in energy between -2 EV and +20 EV is enormous (approximately 2.1 gazillion to 1). Even if the microscopic photosites could produce a different voltage corresponding to an equivalent range in quantity of photons hitting it, the electronics don't have the capacity to read it. That's why a bias voltage is applied to the sensor and the camera's processor calculates the signal by subtracting the bias voltage.
08-29-2018, 09:03 AM   #14
Pentaxian
Fogel70's Avatar

Join Date: Sep 2009
Location: Stockholm, Sweden
Photos: Albums
Posts: 3,062
QuoteOriginally posted by pathdoc Quote
@Digitalis and other technical experts might want to chime in on this one.

If I wanted a digital sensor with the LOWEST base ISO possible, for slow-speed tripod work (e.g. diffusing water, "eliminating" crowds) or for wide-open super-bokeh work with very fast lenses in bright sunlight, what are the determining factors behind getting it (without having to put a gazillion-stop ND filter in the way and screw my focus up)?

I'm talking SINGLE FIGURES here, certainly not higher than 25, with NR algorithms keeping it usable up to maybe 800 or 1600.
I think this will be solved on future sensors, but not with lower base ISO.
Instead it will be by the use of faster readout of the sensor, electronic shutter with much faster shutter speed and combined will averaging multiexposure for adding longer exposure time.

Probably something that cameras will do automatically (optional) if it detects that the exposure settings would lead to a severely overexposed image.

Last edited by Fogel70; 08-29-2018 at 09:08 AM.
08-29-2018, 09:08 AM - 2 Likes   #15
Pentaxian
photoptimist's Avatar

Join Date: Jul 2016
Photos: Albums
Posts: 5,113
For diffusing water and ghosting crowds, interval composite (set to "average") can be a decent solution.
Reply

Bookmarks
  • Submit Thread to Facebook Facebook
  • Submit Thread to Twitter Twitter
  • Submit Thread to Digg Digg
Tags - Make this thread easier to find by adding keywords to it!
amount, apertures, base, base iso, dr, exposure, hoax, iso, iso invariance, light, noise, option, photography, picture, shutter
Thread Tools Search this Thread
Search this Thread:

Advanced Search


Similar Threads
Thread Thread Starter Forum Replies Last Post
I had a dream. robiles Pentax DSLR Discussion 21 08-25-2018 10:43 AM
What determines a Cameras video quality? godwinaustin Video Recording and Processing 8 01-03-2013 10:47 AM
What determines AF speed? body or lens coreyhkh Photographic Industry and Professionals 6 07-08-2012 03:10 AM
Why do you dream about the lenses you dream about? RioRico Pentax SLR Lens Discussion 6 02-02-2012 02:18 PM
Dream Trip - Which Dream Lens? sealonsf Pentax SLR Lens Discussion 8 03-19-2010 05:27 PM



All times are GMT -7. The time now is 10:25 PM. | See also: NikonForums.com, CanonForums.com part of our network of photo forums!
  • Red (Default)
  • Green
  • Gray
  • Dark
  • Dark Yellow
  • Dark Blue
  • Old Red
  • Old Green
  • Old Gray
  • Dial-Up Style
Hello! It's great to see you back on the forum! Have you considered joining the community?
register
Creating a FREE ACCOUNT takes under a minute, removes ads, and lets you post! [Dismiss]
Top