wdross Loc: Castle Rock, Colorado

Not necessarily a conflict. In the film days, the best side film was about 5 stops, the best negative film was about 7 stops, and the best B&W film was about 10 stops. This is why slide film required the most accuracy in exposure because the print paper was almost the same value. One could easily lose highlights or shadow details when shooting with slide film and then trying to print. It was a challenge. But now with the sensor manufacturers expanding the dynamic range of the sensors, it makes it much easier for any sensor to capture a range of 10 stops and match it the 6 to 8 stops of prints and monitors. Like Burkphoto said: it is easier to scan for the maximum slide image than it is to actually print directly from that slide because of the narrow ranges involved. Because the need for the 6 to 8 stops is easily exceeded by 10 or more stops of capture, the dynamic range of the different sensor is not as important as it use to be.

Now, do all of us photographers want more dynamic range? Of course! If we could get 20 stops, we would just post process it down to the 5 to 7 stops needed for prints with maximum image detail in the highlights and shadows. Would 20 stops always be needed to get a great shot? Absolutely not! In fact, even with the 20 stops desired, my bet is most of them would end up actually being processed out to prevent distracting details from the print.

hjkarten Loc: San Diego, California

Amazing statement on your part. The only part of the message that was not of my authorship was my one-line quote from the internet. I repeated the same line for emphasis at the beginning and end of my posting. The balance of my posting was an abbreviated version the same material that I authored and teach to an advanced graduate class in digital light and fluorescent microscopy. For the sake of my students, please let me know which statements were incorrect so that I can correct my lecture!
Harvey

bclaff Loc: Sherborn, MA (18mi SW of Boston)

I didn't read this through initially because of numerous problems I saw as I scanned it.
Now I see one that mostly it's one very wrong concept repeated over and over.

Dynamic range is never expressed as "bit-depth".
In photography it's typically stops or Exposure Value (EV) and sometimes dB (decibels).
I cannot recall a camera that lists dynamic range in the tech specs.

Continued confusion about bit-depth and how it related to dynamic range.
If you intend to talk about "gray levels" then you're talking about tonal range not dynamic range.

More on bit-depth.
I'm unaware of any consumer camera with a bit-depth of less than 8.

Even more on bit-depth. And confusion with tonal range.
FWIW, 16-bit sensors have been around for a while.

Still following the fundamentally wrong concept of confusing bit-depth with dynamic range.

Right.

Actually that's demoasicing as well as other things like the power function and tone curve.
Quantization is what happens when the Analog to Digital Converter (ADC) converted the analog voltage to a discrete integer value. Perhaps you consider the reduction to 8-bits as a form of quantization.

bclaff Loc: Sherborn, MA (18mi SW of Boston)

Without quote marks it was hard to tell.
To me it looked like a long quote. Perhaps you were quoting yourself.

TucsonDave Loc: Tucson, Arizona

The following is from a book entitled "Real World, Camera Raw":

"Dynamic range in digital cameras is an analog limitation of the sensor. The brightest scene information the camera can capture is limited by the capacity of the sensor element. At some point the element can no longer accept any more photons—a condition called saturation—and any photons arriving after saturation are not counted. The darkest shade a camera can capture is determined by the more subjective point at which the noise inherent in the system overwhelms the very weak signal generated by the small number of photons that hit the sensor—the subjectivity lies in the fact that some people can tolerate more noise in their photographs than others.

One way to think of the difference between bit depth and dynamic range is to imagine a staircase. The dynamic range is the height of the staircase. The bit depth is the number of steps in the staircase. If we want our staircase to be reasonably easy to climb, or if we want to preserve the illusion of a continuous gradation of tone in our images, we need more steps in a taller staircase than we do in a shorter one, and we need more bits to describe a wider dynamic range than a narrower one. But more bits, or a larger number of smaller steps, doesn’t increase the dynamic range, or the height of the staircase."

Should clear up the differences and relationships between bit depth and dynamic range. Next Topic!

hjkarten Loc: San Diego, California

Dear WClaff,
Let's decompress the intensity of this conversation. I meant you no offense in my short "lecture".
I won't try to address all your points in one reply. However, we may be suffering from using slightly different definitions of related physical properties.
I still have access to 4 and 6 bit depth cameras. They provide high speed videos at ca. 25,000 frames per second. We have other cameras in the lab that have frames of 128 x 128 pixels. (Not 128,000) with bit-depths of 4 bits. Bit depth is programmable. This is vital when photons are scarce and time is an important variable.
I still have a PMT scanning galvanometer confocal with a 12 bit detector.
I have been using 16 bit cameras for more than 10 years in the lab. I have been using a 16 bit/channel camera with up to nine different wavelengths in photos in excess of 80 GIGAPIXELS. My first book contained about 80 photos taken on 8x10 inch astronomy film. It didn't have enough resolution for our long term needs.

SONY has ads that speak of the 14 bit capabilities of the A7Riv.
On the A7Riv you can store images in a variety of BIT-DEPTHS (not EV values). Bit depths determine the number of colors than can be captured. Colors are an artifice of how colors are captured and stored. A simple CMOS or CCD may have a frequency/wavelength sensitivity, but it usually only detects a broad range of gray values. A color bandpass filter and software expresses the results as color of a specified spectral range as a color picture.
On the SONY A7Riv camera, images can be stored in a variety of bit depths and file formats/sizes.
JPEG is typically stored at 8 bits. It is further compressed to save storage space.
RAW can be stored at 12 bits (called COMPRESSED RAW)
RAW can also be stored at 14 bits (called UNCOMPRESSED RAW)
One of their video modes is stored at 10 bits.
Much of this may seem superfluous, since typical computer monitors (and TV sets) can barely manage to display 7 bits of depth/channel. Monitors that can display 10 bits of depth/channel cost many thousands of dollars. There is little point in struggling to generate a true 8 bit/channel image if it is only going to be viewed on a laptop. Even more so, if you plan to display it on an older smartphone. (In an attempt to make things look more vibrant to the human eye, they invented a trick called Gamma).

One of the very neatest tricks of the latest few generations of digital cameras is Auto-ISO shifting. Added to automatic shutter speed and automatic aperture adjustment. But even as your own MTF graphs demonstrate, as you shift to ever higher ISO sensitivities, the bit depth of the photos drops off in (more or less) linear fashion. That means that noise levels increase, resolution decreases, limiting the benefit of ISO shifting at dimmer levels. The noise levels become intolerable.

EV is a useful formulation, but I have not used it for anything except in reference to consumer/prosumer cameras. I will confess that I tend not to use it, but acknowledge that it is a very useful concept for helping recognize equivalent levels of illumination at the sensor/film plane. For Birds in Flight photos, I set my camera to Manual Mode, and then bracket my Auto-ISO within levels that limit the noise relative to my acceptable results.

Forgive me for giving a small lecture on binary counting. I'm sure that you are personally comfortable with binary, but most people are confused by it. My Physics and computer sciences students are not, but my neurobiology students develop a blank look on their face when asked about counting in binary. (These students are fluent in PYTHON and mumble MatLab in their sleep.) I thought it might be helpful for photographers to learn about this.
Most photographers know that RAW files require more storage space on their storage media than does JPEG. They do not know why. 14 bits occupies more space than 8 bits. In fact, it requires twice as much space, because file formats are available in bytes (=8 bits). 14 bits requires 1.75 bytes. To simplify things, computer engineers wisely chose to leave the last two bits empty when encoding 14 bits of a single data point.
(And I won't get into what we mean by a pixel when dealing with a Bayer mask.)
Most photographers do not realize why RAW format is so much larger than JPG, or why RAW may be a better format. Astrophotographers appreciate the importance of bit depth, hence my comparison of a dim star vs. a very dim star.
Enuff'
I find your website with lots of MTF charts extremely useful. THanks for the work that goes into that.
Stay safe, stay healthy, get vaccinated as soon as your number comes up.
ANd don't be so quick to disparage people you don't know.
regards,
Harvey

bclaff Loc: Sherborn, MA (18mi SW of Boston)

Harvey,
You still don't acknowledge your confusion between bit depth and dynamic range which concerns me because you apparently teach this to others.
If you feel that someone who corrects your incorrect statements to be disparaging I have no control over that.
FWIW, there are no MTF charts at my site; so that confuses me.

hjkarten Loc: San Diego, California

Dear TucsonDave,
That is worthwhile pondering. I like his staircase analogy.
I have to reflect on your quote that that "...we need more bits to describe a wider dynamic range than a narrower one." Strictly speaking, it is valid to define Dynamic Range as range of sensitivity, not the number of gradations. But then why did the author speak of "...more bits to describe a WIDER DYNAMIC RANGE..." THe author introduces an ambiguity to the definition of dynamic range. Is dynamic range a monodimensional variable, or does it imply a bidimensional variable? If monodimensional, is it the height of the staircase or is it the number of steps. i.e., the spectral range (e.g., 400 - 750 nanometers -short blue to long wavelength red) would mean that there would be no gradient of color/intensity. If bidimensional, then dynamic range both height of the staircase and the number of gradations. Does this then also imply that there are different gradations for different wavelengths? We know that it true for human vision. How did that affect chip sensor design?
regards,
Harvey

bclaff Loc: Sherborn, MA (18mi SW of Boston)

Perhaps the author was trying to say that bit-depth places an upper limit on dynamic range so you (may) need more bits to capture the higher dynamic range.

hjkarten Loc: San Diego, California

I apologize for attributing the MTF charts to you. They were on a different page. I erred in giving you an undeserved expression of gratitude.
I confounded the attribution of your useful chart showing the relationship of bit depth to ISO.
I also defer to your wisdom and competitive spirit.

Harvey

hjkarten Loc: San Diego, California

Dear Bill,
I apologize for attributing the MTF charts to you. They were on a different page. I erred in giving you an undeserved expression of gratitude.
I confounded the attribution of your useful chart showing the relationship of bit depth to ISO.
I also defer to your wisdom and competitive spirit.

Harvey

TucsonDave Loc: Tucson, Arizona

Thanks for your observations. My opinion is that the author is saying there is an indirect relationship between bit depth and dynamic range. Thus, changing the bit depth doesn't change the dynamic range (number of steps), just the height of the steps within the range. It certainly implies that there could be different gradations for different wavelengths.

This discussion has deepened past the point of providing a useful simple answer to the original question. Even so, it is instructional to probe deeper every now and then. Thanks for taking that step.

User ID

The tech discussion is useful and the material is understandable. Some of the disagreement is culture clash, a difference in vocabulary between general photography and scientific photography.

As to hijacking the thread, anything goes once a thread passes three pages. That may be a sarcastic remark, but it’s also very true.

Please critique this distillation of my take on the discussion of bit depth as related to dynamic range:

An increased DR may require greater bit depth to record a fully useful and more “processing friendly” file, but simply increasing the bit depth is not going to increase the DR from a given sensor.

IOW it’s not a reversible or symmetrical relationship and therefore it is not expressable as an equation. A wider gamut of data input benefits from being stored at greater bit depth, but the providing of that potentially more beneficial storage is not going to automatically widen the gamut of the data provided from the sensor via the ADC.

Well OK ... is that a correct summary ?

texashill Loc: Texas Hill Country

no kidding; he has been a treasure for years

bleirer

Works for me as a nonscientist. I can see sensors with the same bit depth when tested having very different dynamic ranges.