How will advances in digital Technology change how exposures are made ?

[petersmart]

Not so - the output from a photodiode will always be limited by its saturation level - the level at which no further output is possible no matter how much extra light is available.

Of course it is, but we are future gazing and they will have far higher saturation capacities in the future. (all stated previously.)

Precisely how?

It's one thing to "future gaze" but quite another matter to explain how that would be achieved.

There is no film which has never suffered from burnt out highlights due to saturation of the silver halides and there are no sensors diodes which do not saturate.

The only REALISTIC way to achieve this is to prevent the sensor from saturating by limiting the exposure but then you have blocked shadows - assuming a very bright day.

What you would need are LARGER photodiodes with a larger output and there are already cameras available with precisely that - Medium Format - but the costs are too high for most of us.

Medium Format cameras do fulfill some of your dreams, with a larger dynamic range AND less noise, but the majority of us are stuck with a limited dynamic range and noise.

And we will be for a very long time to come, certainly far longer than the 10 years you mentioned.

.

 

[Pookeyhead]

Precisely how?



There is no film which has never suffered from burnt out highlights due to saturation of the silver halides and there are no sensors diodes which do not saturate.

.

Correct, but if the dynamic range of that sensor is such that the saturation point requires so much light as to render it academic, then in reality, it may as well be film (or behave like film in terms of being unable to burn out due to saturation)

 

[petersmart]

Correct, but if the dynamic range of that sensor is such that the saturation point requires so much light as to render it academic, then in reality, it may as well be film (or behave like film in terms of being unable to burn out due to saturation)

I actually said that ALL films suffer from the same problem which is saturation of the material.

In the case of silver halides that point is reached when all the halides are converted to metallic silver.

.

 

[Pookeyhead]

But not in the same way, you can recover highlight detail far easier from film, as while being very dense on the negative still contains information unless grossly over exposed. Sensors become saturated very quickly once they approach their dynamic range limit and behave in a very different way. I assumed you were referring to that as I misread. Film is limited in as much as the amount of tolerance to over exposure is a product of silver density, whereas a sensor's theoretical limit is simply it's ability to translate ever increasing numbers of photons hitting it into an ever increasing signal. That limit is theoretical at the moment, because we don;t know what's just around the corner. Hard drives are an example. No sooner do we assume we're approaching the limits of data density of magnetic hard drives, and some smart alec invents perpendicular recording, and all of a sudden we have cheap 4TB 2.3" hard drives. Wind the clock back a couple of years and people would have said that was unlikely as you couldn't fit enough platters into a 3.5" drive.

 

[HoppyUK]

I actually said that ALL films suffer from the same problem which is saturation of the material.

In the case of silver halides that point is reached when all the halides are converted to metallic silver.

.

Negative film, or rather Type II colour neg to be specific, plays a rather clever cheat with DIR colour-couplers - Development Inhibiter Release, IIRC. Basically highlights reach a certain density and then further developement slows dramatically, suffering relatively little damage to IQ with gross over-exposure, but the shadows continue to build. It's the trick that made disposable cameras turn in surprisingly good results with fixed exposure, and saved the bacon of many a wedding photographer.

You could do something similar with B&W negs, but with much less benefit, with careful developing techniques. Nothing possible with slide film, that is similar to digital in that once highlights are blown, they're gone. I'm wondering though if digital might be made to play the same kind of game, with increased resistance to exposure before it reaches the blowing threshold - some kind of photon over-flow before the pixel well is full?

 

[petersmart]

I'm wondering though if digital might be made to play the same kind of game, with increased resistance to exposure before it reaches the blowing threshold - some kind of photon over-flow before the pixel well is full?

In fact it already does to a very limited extent which is why the characteristic curve of a digital sensor is commonly called an "S" curve and is similar to the curves for films.

Once the output from the sensor ceases to be linear at either the top or bottom tonal compression is taking place.

At the top this means that extra light no longer produces a linear response in much the same way you suggested.

Personally I believe we are already approaching the limit of current technology and the Nikon D800 with its 36Mp sensor shows why.

The resolution of this sensor now outstrips the lenses available inasmuch as it can actually show diffraction softening at levels we can actually see and even a 1 or 2 stop difference from the ideal produces a visible softening of the images.

In addition noise is much higher than on lower MP cameras even if the larger size images do help to hide it.

If that is the case and 36MP is the limit for Full Frame cameras then the only way to improve is to find ways to reduce noise and increase dynamic range outside the camera with techniques such as NR software and HDR or similar techniques.

.

 

[HoppyUK]

Yes, now you mention it the S curve is certainly a feature of digital sensors. I don't know if that's somehow inherant or designed in. Either way, this is such an obvious trick to contain highlights that I'm sure we'll see more of that if it's feasible.

Not sure I agree we've reached any kind of ceiling with sensor development though. Whether we need it or can realise higher resolution is a moot point, but I'm sure we'll see plenty higher than 36mp on full-frame over the next few years. And there are good arguments in favour that I'll leave to one side for now.* Take the Nikon 1 V2 for example, now with 14.2mp, and equipped with lenses that can resolve them. That's the equivalent of 104mp in full-frame terms

But isn't an underlying point that Terry is making based on the fact that what we previously thought were limits, re noise and ISO and pixels etc, have been proved not to be the case? There are no limits, other than perhaps the law of diminishing returns, cost vs benefits, and so on. And if we re-think the way we adjust for exposure, there are new and liberating creative opportunities?

*Okay, just to illustrate Take the D800 with 36mp, that can be cropped down to 1.5x format and still leave a very usable 16mp - ie, same as a D7000, but with 1.5x more 'pixel reach' than the uncropped D800. Also, another upside of that is the D800 can run at 6fps in DX mode so you effectively get two cameras in one - great for super-detailed landscapes, and very handy for action and wildlife too.
On pixel counts, sharpness is about both resolution and contrast so even if resolution is capped by total pixels, what we perceive as sharpness can still increase - as has been proven often enough.

 

[Pookeyhead]

Y
Not sure I agree we've reached any kind of ceiling with sensor development though.

I think it would be foolhardy to think we have. How many times in the past have we assumed we've reached the limit of something, then someone invents a new process or means of achieving something. The hard drive example a few posts up is a good one. It's usually the case that what we think of as limits now, turn out to be no such thing.

 

[Terrywoodenpic]

Precisely how?

It's one thing to "future gaze" but quite another matter to explain how that would be achieved.

That is not the subject of the OP, the scientific How's are the province of engineers.

There is no film which has never suffered from burnt out highlights due to saturation of the silver halides and there are no sensors diodes which do not saturate.

The use of the "double negative" is no help to comprehension.
But in film work compensating developers like Beutler formulations help a great deal as they allow the shadows to develop further once the local agent is exhausted at the highlight. Over exposure is the main cause of poor highlight detail in all photography.

The only REALISTIC way to achieve this is to prevent the sensor from saturating by limiting the exposure but then you have blocked shadows - assuming a very bright day.

This technique is called exposing to the right, and will Always be the best strategy even when very wide dynamic range is available as it utilises the maximum straight line response of the sensor.

What you would need are LARGER photodiodes with a larger output and there are already cameras available with precisely that - Medium Format - but the costs are too high for most of us.
Medium Format cameras do fulfil some of your dreams, with a larger dynamic range AND less noise, but the majority of us are stuck with a limited dynamic range and noise.

The Nikon D800 is already superior in this respect to any medium format available.

And we will be for a very long time to come, certainly far longer than the 10 years you mentioned.

Think back to 2003 ( ten years ago) there was no camera that had an acceptable noise profile at ISO400.

The speed of progress is accelerating many-fold.

 

[ernesto]

I think it would be foolhardy to think we have. How many times in the past have we assumed we've reached the limit of something, then someone invents a new process or means of achieving something. The hard drive example a few posts up is a good one. It's usually the case that what we think of as limits now, turn out to be no such thing.

Agree. 20 years from now we won't believe what we are using and how much better it is than today's technology.

For example, an image could be captured over 1 second but compiled from from shots taken at 1/1000 intervals. So 1000 shots make up the end image and software allows the user to pick and choose they want from any time.
Like a super bracketed option but with more intelligent software and better capability in the sensor and firmware to tie it all together.

 

[Terrywoodenpic]

Negative film, or rather Type II colour neg to be specific, plays a rather clever cheat with DIR colour-couplers - Development Inhibiter Release, IIRC. Basically highlights reach a certain density and then further developement slows dramatically, suffering relatively little damage to IQ with gross over-exposure, but the shadows continue to build. It's the trick that made disposable cameras turn in surprisingly good results with fixed exposure, and saved the bacon of many a wedding photographer.

You could do something similar with B&W negs, but with much less benefit, with careful developing techniques. Nothing possible with slide film, that is similar to digital in that once highlights are blown, they're gone. I'm wondering though if digital might be made to play the same kind of game, with increased resistance to exposure before it reaches the blowing threshold - some kind of photon over-flow before the pixel well is full?

The Introduced S Curve on sensor output helps in a similar way as the shoulder region progressively compresses the tones. Of course the registers eventually fill and tones burn out.
The Raw output is straight line from the extreme shadows to the near saturation point. it is this "line" that is replotted to establish the various S Curves representing the ISO values.

 

[Terrywoodenpic]

In fact it already does to a very limited extent which is why the characteristic curve of a digital sensor is commonly called an "S" curve and is similar to the curves for films.

Once the output from the sensor ceases to be linear at either the top or bottom tonal compression is taking place.

At the top this means that extra light no longer produces a linear response in much the same way you suggested.

Personally I believe we are already approaching the limit of current technology and the Nikon D800 with its 36Mp sensor shows why.

The resolution of this sensor now outstrips the lenses available inasmuch as it can actually show diffraction softening at levels we can actually see and even a 1 or 2 stop difference from the ideal produces a visible softening of the images.

In addition noise is much higher than on lower MP cameras even if the larger size images do help to hide it.

If that is the case and 36MP is the limit for Full Frame cameras then the only way to improve is to find ways to reduce noise and increase dynamic range outside the camera with techniques such as NR software and HDR or similar techniques.

.

There never will be "NO NOISE" It is inherent in the shot noise of photons and is totally random.

Noise reduction is probably as close as it ever will be in percentage terms, though the small improvements might be significant visually.

The sensor response is linear (except at the extreme shoulder) The "S curve" is imposed in Processing.

 

[Terrywoodenpic]

But not in the same way, you can recover highlight detail far easier from film, as while being very dense on the negative still contains information unless grossly over exposed. Sensors become saturated very quickly once they approach their dynamic range limit and behave in a very different way. I assumed you were referring to that as I misread. Film is limited in as much as the amount of tolerance to over exposure is a product of silver density, whereas a sensor's theoretical limit is simply it's ability to translate ever increasing numbers of photons hitting it into an ever increasing signal. That limit is theoretical at the moment, because we don;t know what's just around the corner. Hard drives are an example. No sooner do we assume we're approaching the limits of data density of magnetic hard drives, and some smart alec invents perpendicular recording, and all of a sudden we have cheap 4TB 2.3" hard drives. Wind the clock back a couple of years and people would have said that was unlikely as you couldn't fit enough platters into a 3.5" drive.

I can not conceive of a time when we will be unable to come up with a way of counting higher numbers of photons. Since the days of the abacus we have been improving and inventing new ways to count and measure every thing, Photons are no exception. Even though their position can not be be predicted and they are totally random.

 

[Terrywoodenpic]

I think it would be foolhardy to think we have. How many times in the past have we assumed we've reached the limit of something, then someone invents a new process or means of achieving something. The hard drive example a few posts up is a good one. It's usually the case that what we think of as limits now, turn out to be no such thing.

As of now most imaging sensors are either CMOS or CCD Both are based on Silicon. It seem all medium format sensor are still CCD, and virtually all the rest are now CMOS. (few exceptions)

It seems unreasonable to suppose that silicon will always be the only option, or that those two strategies will be the only ones ever invented.

At one time all but the most basic wheels were made of wood and had iron rims, the world understood and ran on them for centuries...... Times changed.