Forgot Password
Pentax Camera Forums Home
 

Reply
Show Printable Version Search this Thread
03-04-2011, 09:28 PM - 2 Likes   #1
Pentaxian
dosdan's Avatar

Join Date: Dec 2007
Location: Brisbane, Australia
Photos: Gallery | Albums
Posts: 1,741
The "ISOless" Sensor

In the Whirlpool.net.au Photographic forum in a discussion about the DxOMark differences in a number of models schmegg lamented the absence of an ISOless sensor.

Quoting schmegg: Until a sensor is produced that is capable of removing ISO as an exposure parameter these gains are mostly of interest to measurbators and fanboys IMHO

The Spanish engineer Guillermo Luijk (and Canon user) has been looking at this very topic. In a new article he has compared the SNR & DR performance of the 5DII, D700 & K5.

http://translate.google.com/translate?u=http%3A%2F%2Fwww.guillermoluijk.com%2Farticle%2Fsnr%2Findex.htm&langpair=es|en&hl=EN&ie=UTF-8

Please ensure that you move your mouse over his Fig. 3 so you can see the overlayed curve showing the improvement in S/N in the deep shadows, if a constant shutter speed and aperture are maintained, as ISO is increased. This is very interesting indeed.

  • With the 5DII (and other cameras with relatively high read noise), the SNR improves significantly (2.1 EV) as the ISO is raised 4 stops (ISO100 -> ISO1600).
  • With the D700 (and other cameras with relatively good read noise performance), the improvement is less (hence the need to raise ISO is also reduced), with SNR improving 1 EV as ISO is raised 2 stops (ISO200 -> ISO800).
  • The K5 (and other cameras using the new Sony low read noise "wonder" sensor) have an almost flat SNR improvement curve, with SNR improving only 0.3 EV as the ISO is raised 4 stops (ISO100 -> ISO1600). Actually, since the K5 goes down to ISO80, that's 4.3 stops.

This appears to be the closest yet to an ISOless camera, i.e a camera where a significant level of underexposure when shooting raw can be corrected in PP without much performance penalty at all.

As regards DR (Fig. 5), the FF models (5DII & D700) are better at ISO800 & ISO1600. But by ISO400 the K5 equalises, and at lower ISO surpasses the FF cameras with an amazing straight-line performance to ISO80.

This Sony APS-C sensor marks a new era in performance.

Dan.


Last edited by dosdan; 03-18-2011 at 11:20 PM.
03-04-2011, 10:02 PM   #2
Veteran Member
Smeggypants's Avatar

Join Date: Sep 2010
Posts: 1,536
ISO is simply light amplifier gain.

Like any amplifier the more you increase the gain the more the noise is increased.

I can't see how you can have an ISOless sensor unless you mean the S/N ratio is always greater than the bit resolution of the recording medium. And if it is you might as well increase the recording mediums bit resolution to take advantage of the lower ISO's S/N

note: the K-5 has already done this by increasing the resolution of the recording medium to 14bit from 12bit
03-04-2011, 10:25 PM - 1 Like   #3
Pentaxian
dosdan's Avatar

Join Date: Dec 2007
Location: Brisbane, Australia
Photos: Gallery | Albums
Posts: 1,741
Original Poster
Consider the same scene with constant aperture & shutter speed, while ISO sensitivity in-camera is varied. At low to medium ISO settings this will involve analogue amplification. (I'll leave the matter of fractional-ISO steps via digital multiplication in-camera and digital multiplication at high ISO in-camera out of this discussion.)

The question then becomes "Which is better: adjusting the in-camera ISO (analogue amplification) to properly expose the shot or shooting at base ISO and then applying digital amplification by using EV boosting in PP?"

Guillermo Luijk covers that topic here (please do read his intro first or you'll misunderstand what he's attempting to explain): http://translate.google.com/translate?u=http%3A%2F%2Fwww.guillermoluijk.com%2Farticle%2Fiso%2Findex.htm&langpair=es|en&hl=EN&ie=UTF-8

As you'd expect, it's usually better to use higher ISO in-camera than EV boost in PP, because analogue amplification before ADC can reduce the effect of the contribution of the read noise.

That's shown by the black part of the overlay curve in Fig. 3:





In the black part of the curve, analogue amplification is raising the image's shadows above the read noise floor, so shadows are less noisy.

Once the grey part of curve is reached, the image noise floor is rising at the same rate as amplification, so there is no further improvement.

With the K5 and other cameras using this Sony sensor, the black curve has a small slope, not because it's noisy, but the reverse: it has so little read noise, that applying amplification in-camera (actually on-sensor with this sensor), is not giving much of an improvement.

So, in contrast to normal sensors, where digital amplification in PP is significantly worse than analogue amplification in-camera, with the K5 it is almost as good as doing it in-camera, and it is certainly more flexible to be able to do it after the shot is taken.

The DR will decrease whether a scene requires high ISO in-camera or EV boost in PP to achieve a proper exposure because the full-well capacity decreases.

But with the K5, the EV boost in PP imposes hardly any noise penalty and you get the advantage of more protection from blown highlights because when you bring the exposure level up in PP you can fiddle with the tone response curve to handle any highlights, whereas with high ISO in-camera you can blow the highlights when you're trying to expose the histogram to the right to maximise noise performance.

With the K5 you can expose to the centre or more to the left without much of a noise penalty at all.


Dan.

Last edited by dosdan; 03-07-2011 at 03:21 PM.
03-04-2011, 10:35 PM   #4
Inactive Account




Join Date: Jul 2008
Location: Alexandria, VA
Photos: Albums
Posts: 100
Enter: Exposing to the Left

03-04-2011, 10:49 PM   #5
Pentaxian
dosdan's Avatar

Join Date: Dec 2007
Location: Brisbane, Australia
Photos: Gallery | Albums
Posts: 1,741
Original Poster
QuoteOriginally posted by Slowpoke Rodriguez Quote
Enter: Exposing to the Left
It's really a paradigm shift. There is little/no need for ISO shifting from base ISO. Let's drop the ISO setting. We can trust this camera to get a good exposure noise performance over a 4.3 stops exposure range (equivalent to switching from ISO80 -> ISO1600). We then use just the Aperture for DOF control and Shutter Speed for motion blur control.

Dan.

Last edited by dosdan; 08-12-2013 at 11:20 PM.
03-04-2011, 10:50 PM   #6
Veteran Member
RawheaD's Avatar

Join Date: Dec 2008
Location: MA, USA
Photos: Gallery
Posts: 831
It took me a minute to understand the gist of the article, but this is actually quite cool. What would be cool is if Pentax (or whoever) goes one step further, maybe with the next generation Sony sensor, and (1) get rid of the ISO setting altogether, maybe with the entry-level cams, and (2) incorporate this characteristic into the in-camera JPEG processing.

I.e., instead of having a mode like TAv, where you get to choose your aperture and shutter speed while the ISO changes, you can have a mode where you choose your aperture and shutter speed and the cam will always take the shot at the optimal/base ISO, and process the RAW inside the camera accordingly so you get properly exposed JPEGs (obviously, this will only work well when the setting is "under"exposing at that setting).

I've always thought of TAv mode being cool, because ISO really doesn't factor into what I'm aiming for in an image (unless you happen to be going for the faux-grainy film look), while shutter speed and aperture are a big part of it, but I've been too hesitant to use it, becuase I'm afraid that it'd shoot in high ISO giving me bad pictures.
03-05-2011, 02:44 AM   #7
Inactive Account




Join Date: Apr 2010
Location: Suomi , Finland
Posts: 368
Muuten hyvä, mutta ei Sony. Liikaa harmia, ei Pentax kestä enään uutta pettymystä.

03-05-2011, 11:48 AM   #8
Veteran Member
Smeggypants's Avatar

Join Date: Sep 2010
Posts: 1,536
QuoteOriginally posted by RawheaD Quote
It took me a minute to understand the gist of the article, but this is actually quite cool. What would be cool is if Pentax (or whoever) goes one step further, maybe with the next generation Sony sensor, and (1) get rid of the ISO setting altogether, maybe with the entry-level cams, and (2) incorporate this characteristic into the in-camera JPEG processing.

I.e., instead of having a mode like TAv, where you get to choose your aperture and shutter speed while the ISO changes, you can have a mode where you choose your aperture and shutter speed and the cam will always take the shot at the optimal/base ISO, and process the RAW inside the camera accordingly so you get properly exposed JPEGs (obviously, this will only work well when the setting is "under"exposing at that setting).

I've always thought of TAv mode being cool, because ISO really doesn't factor into what I'm aiming for in an image (unless you happen to be going for the faux-grainy film look), while shutter speed and aperture are a big part of it, but I've been too hesitant to use it, becuase I'm afraid that it'd shoot in high ISO giving me bad pictures.
I prefer a noiser image to one that is blurred or not enough DOF.

You can set limits to the ISO range in the K-5 anyway. You don't need much PP NR to get great qulaity images at ISO6400 either.

I don't see the point of getting rid of the ISO parameter


QuoteOriginally posted by Dosdan:
The question then becomes "Which is better: adjusting the in-camera ISO (analogue amplification) to properly exposure the shot or shooting at base ISO and then applying digital amplification by using EV boosting in PP?"

Analgue amplification is better because it puts the light levels in a range where the analogue to digital conversion is optimal. Analog noise is nicer than quantisation noise.

IIRC ISOs over 6400 on the K-5 are digitally boosted.

Recording everything at ISO100 and digitally boosting for lower light scenes would be awful. The DR would be rubbish. Shooting to the left is never to be recommended for the very reason shooting to the right is recommended. it's about digitising the signal with the optimal number of bits. The less bits the less DR

Justt becuase the K-5 is now a 14bit camera and that means 2 more bits of info in the darks it doesn't mean we should deliberately shoot dark

When cameras are 24bit, giving 144dB of dynamic range then maybe, but even then I'd still like an ISO setting - even just to see ISO6,553,600
03-05-2011, 01:20 PM   #9
Veteran Member




Join Date: Aug 2007
Location: Madison, Wis., USA
Posts: 1,506
Smeggy, you refer to quantization noise. In which ways is analogue amplification noise nicer than quantisation noise?

In a related question, does the ADC use companding? Larger steps at low signal levels would result in a mess if analogue amplification were not used.

In responding, please note that the above is as technical as I get without help. So .... and in advance .... thanks for your help!
03-05-2011, 01:37 PM   #10
Pentaxian
dosdan's Avatar

Join Date: Dec 2007
Location: Brisbane, Australia
Photos: Gallery | Albums
Posts: 1,741
Original Poster
QuoteOriginally posted by Smeggypants Quote
Recording everything at ISO100 and digitally boosting for lower light scenes would be awful. The DR would be rubbish. Shooting to the left is never to be recommended for the very reason shooting to the right is recommended. it's about digitising the signal with the optimal number of bits. The less bits the less DR
The DR, as ISO increases in-camera, also decreases. See:

Noise, Dynamic Range and Bit Depth in Digital SLRs -- Supplemental page 3a

Quotes:
When the ISO is doubled, one more stop or EV of input is pushed past the saturation point of the ADC and is lost; thus each subsequent one stop increase of ISO removes one stop of highlight exposure.

...
Each subsequent stop of ISO amplification lops off one stop from the upper end, while the room on the lower end improves because read noise drops in absolute exposure terms with increasing ISO, until it saturates about ISO 1600.
...
so long as noise sufficiently exceeds quantization step, there is no difference between amplification in hardware during the capture process, and amplification after the fact during raw conversion.

As long as the analogue noise present at the ADC input is random and is at the same or greater level than the minimum quantisation step, the signal will be dithered sufficiently and posterization will not occur.

See: Noise, Dynamic Range and Bit Depth in Digital SLRs -- page 3

Assuming the read noise is very low, as it is in the K5, and therefore there is little of the "room on the lower end improves because read noise drops in absolute exposure terms with increasing ISO", a lSO100 shot underexposed say 4 stops, with sufficient dithering, and boosted 4 stops in PP should have almost the same amount of noise, posterization & the DR as an ISO1600 in-camera properly exposed shot.

Dan.

Last edited by dosdan; 03-05-2011 at 03:33 PM.
03-05-2011, 03:19 PM   #11
Pentaxian
dosdan's Avatar

Join Date: Dec 2007
Location: Brisbane, Australia
Photos: Gallery | Albums
Posts: 1,741
Original Poster
If I understand it correctly, the only problems with shooting raw up to 4 stops under-exposed on the K-5, rather than boosting the in-camera ISO sensitivity, are:

1. The histogram shown after shooting becomes less useful (but also less necessary).

2. The preview image which is embedded in the raw file, and which is also shown on the back
LCD screen, is darker.

3. You suffer some noise increase with extreme under-exposure (+0.3 EV more noise after 4 stops boosting in PP), although this slight increase proportionally reduces with a decreasing level of under-exposure.

Dan.

Last edited by dosdan; 03-05-2011 at 03:35 PM.
03-05-2011, 03:21 PM   #12
Veteran Member
Smeggypants's Avatar

Join Date: Sep 2010
Posts: 1,536
QuoteOriginally posted by glanglois Quote
Smeggy, you refer to quantization noise. In which ways is analogue amplification noise nicer than quantisation noise?
The analog noise is random, while quantisation noise is signal related and thus ot random

QuoteQuote:
In a related question, does the ADC use companding? Larger steps at low signal levels would result in a mess if analogue amplification were not used.

In responding, please note that the above is as technical as I get without help. So .... and in advance .... thanks for your help!
No idea.
03-05-2011, 03:40 PM   #13
Veteran Member
Smeggypants's Avatar

Join Date: Sep 2010
Posts: 1,536
QuoteOriginally posted by dosdan Quote
The DR, as ISO increases in-camera, also decreases.
Yup that's right. My points were pertaining to increasing ISO ( amplification ) in the analogue domain versus increasing ISO in the digital domain post conversion.

As most know shooting to the right is done to maximise the number of bits the signal is digitised by and thus keeps quantisation noise ( the digital noise floor ) down as low as possible.

It's always better to capture the signal with as many bits as possible. Of course if you erroneously grossly underexpose then a 14bit camera liek the K-5 rescue such an image much better than a 12bit camera. And I assume the analogue noise specof the sensor and it's circuitry is also improved over previous models.

But this is a rescue. And i've seen no evidence that shooting deliberately this way is better than properly exposing in the first place.


Anyway you can have an ISOLess camera now if you want. Simply work in manual mode, set your ISO to 100 and then ignore your meter and shoot with preferred shutter and aperture settings. Have fun looking at a filmstrip of dark pics in PP and messing around adjusting their exposure.

Sounds madness to me
03-06-2011, 05:36 PM   #14
Veteran Member
falconeye's Avatar

Join Date: Jan 2008
Location: Munich, Alps, Germany
Photos: Gallery
Posts: 6,871
QuoteOriginally posted by dosdan Quote
In the Whirlpool.net.au Photographic forum in a discussion about the DxOMark differences in a number of models schmegg lamented the absence of an ISOless sensor.
Dan,

I absolutely agree.

For the record though, let me say that I've been thru all this in this forum last year already. It's a done topic. I am too lazy to look up all my corresponding posts. But they exist

ISO isn't (and has never been) a fundamental exposure parameter. Shutter and aperture are as they control the light entering the camera. ISO isn't just like WB or contrast isn't. ISO and EV compensation are the same (and actually were mechanically linked in older analog cameras for a reason).

The confusion comes from the fact that many cameras exists which have (avoidable) amplifier noise and that it then influences the data captured in a RAW file. As has been said here, the effect is already small for the K-5.

The APSC Exmor sensors have a programmable gain amplifier (PGA) for up to 24dB gain (which is 4 EV). So, it can only amplify ISO 100 to 1600 or ISO 80 to 1280. Everything beyond is isoless and there is no reason to use ISO higher than 1600 and risk blown highlights.
03-06-2011, 06:31 PM   #15
Veteran Member
Smeggypants's Avatar

Join Date: Sep 2010
Posts: 1,536
QuoteOriginally posted by falconeye Quote
Dan,

I absolutely agree.

For the record though, let me say that I've been thru all this in this forum last year already. It's a done topic. I am too lazy to look up all my corresponding posts. But they exist

ISO isn't (and has never been) a fundamental exposure parameter. Shutter and aperture are as they control the light entering the camera. ISO isn't just like WB or contrast isn't. ISO and EV compensation are the same (and actually were mechanically linked in older analog cameras for a reason).

The confusion comes from the fact that many cameras exists which have (avoidable) amplifier noise and that it then influences the data captured in a RAW file. As has been said here, the effect is already small for the K-5.

The APSC Exmor sensors have a programmable gain amplifier (PGA) for up to 24dB gain (which is 4 EV). So, it can only amplify ISO 100 to 1600 or ISO 80 to 1280. Everything beyond is isoless and there is no reason to use ISO higher than 1600 and risk blown highlights.
Interesting.

So are you saying that everything above ISO1600 in the K-5 is digitally amplified?

( I'm assuming ISO100 is Unity Gain and thus ISO80 is a gain reduction. )
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!
5dii, camera, cameras, dslr, ev, iso, k-5, k-5 ii, k-5 iis, k5, noise, pentax k-5, sensor
Thread Tools Search this Thread
Search this Thread:

Advanced Search


Similar Threads
Thread Thread Starter Forum Replies Last Post
Post your "Park Bench" "or "Picnic Table" images tessfully Mini-Challenges, Games, and Photo Stories 2196 5 Days Ago 09:29 AM
Please explain "Sensor is Outresolving Some Lenses" Parallax Pentax K-5 & K-5 II 21 11-24-2010 05:05 PM
Sensor shift during "long" exposures eclipsed450 Pentax DSLR Discussion 11 12-18-2009 12:51 PM
K10D "Sensor moving" possible problem? jaitas Troubleshooting and Beginner Help 6 05-07-2009 02:08 PM
"Green Clean" Sensor Cleaning System. NLAlston Pentax Camera and Field Accessories 2 05-16-2007 07:35 PM



All times are GMT -7. The time now is 04:43 AM. | 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