-
Welcome to the new forums! Please read this first. For known issues we are working to resolve, click here.
What affects the "Dynamic range"?
- Thread starter Alireza S
- Start date
If camera developers would only focus on base ISO performance and forget other ISO settings, could it improve base ISO dynamic range more than what we have now? High contrast landscapes usually have so much light that even base ISO 10 can be enough. That is what film shooters have done since the beginning of camera technology. Plenty of gorgeous photos from that time.
Is there other ditchable things which could improve dynamic range, if cameras would be designed that way?
Is there other ditchable things which could improve dynamic range, if cameras would be designed that way?
Iliah Borg
Forum Pro
Dynamic range is often abbreviated as DR.
In real life photographically useful dynamic range is the range scene brightnesses over which a camera responds with a good contrast, good resolution, and a good signal to noise ratio (SNR).
Engineering definition of dynamic range usually does not account for contrast or resolution, or any other image quality features. It is calculated from the ratio of the maximum signal (maximum number of signal electrons, that is in extreme highlights just short of clipping) to dark noise (pixel read noise, dark current noise, and any noise accumulated in the signal readout and processing, that is in a dark image captured with lens cap on).
You can read more on what noise sources are in Kodak publication "CCD Image Sensor Noise Sources" http://www.onsemi.com/pub/Collateral/AND9189-D.PDF
The exact delta between eng, dynamic range and photographically useful dynamic range is a matter of discussion, Kodak were putting it at 1 2/3 stops minimum, that is photographically useful DR = eng. DR - 1 2/3
Dynamic range is often given in dB, to convert to photographic stops (EV) you can divide by 6.02. Say, when somebody puts 20dB as a difference between photographically useful DR and eng. DR, it means 3 1/3 stops (20 / 6.02 = 3.3).
This difference between eng. DR and photographically useful DR is to ensure perceived image quality, and it changes with the display / print size and viewing conditions, including viewing distance. The smaller is the image as viewed, the smaller the difference between eng. DR and photographically useful DR needs to be.
ISO setting: improves shadow noise (sometimes substantially, sometimes nearly zero improvement, depends on the sensor and firmware design), clips highlights, overall reduces dynamic range because the gain in shadows is overweighted by highlight clipping. If ISO setting is fully applied to raw data, that is 1 stop of ISO setting up results in doubling raw data numbers, for each 1 stop of ISO bump 1 stop of highlights is clipped. 2 stops ISO bump causes 2 stops in highlights to be clipped.
Noise: see the pdf I linked above, or:
The most important factor is Exposure / photon shot noise: the higher is the exposure, the more light is collected on the sensor, the cleaner is the image. It is explained in detail in many places, including DPR's own https://www.dpreview.com/articles/8...e-shedding-some-light-on-the-sources-of-noise
Dark noise / electronic noise: https://www.dpreview.com/articles/0388507676/sources-of-noise-part-two-electronic-noise
Flare and glare: McCann, J. J., Rizzi, A wrote a lot about the subject, with some very interesting experimental work. Conclusion is that the optical system puts the limit to photographically useful DR at slightly more than 11 stops.
Viewing conditions: size and distance, intensity and "colour" of light, illuminating the image, properties of the paper / display.
The above are not all the factors, but what I consider to be major ones.
edhannon
Senior Member
Not sure I understand the clipping. If I set exposure time and aperture based on base ISO, increase ISO, and then take the image I agree that camera will clip as you state.
But why would I do that? I adjust ISO when I need to need to reduce the exposure from what base ISO would require. So I usually increase the ISO and set exposure by placing brightest value that I need to retain detail as far to right as I can.
What would be the impact on DR in that case?
Iliah Borg
Forum Pro
Suppose you have a high dynamic range scene and relatively low light (theatrical light, for example) or you need a good DoF and stop down. Low ISO results in a "dark" image. You bump ISO setting to get a "brighter" image, and highlights are clipped. But if you know the camera is acceptably ISO-less when shooting raw, you may decide not to bump ISO setting, and not to clip.
Tony Beach
Forum Pro
- Messages
- 11,967
- Solutions
- 5
- Reaction score
- 7,035
Do the math. Take Sunny 16 suggests that a camera with a base ISO of 100 would be at 1/200s at f/11 on a sunny day, and we currently have the D810 with a base ISO of 65, so that's 1/125s on a sunny day. Do you really think being at 1/25s on a sunny day is generally feasible? Add to that, that most of time you are not necessarily shooting on sunny days.
Plenty of constraints on what they could do then too, and there still are constraints, but now DR is far less of a constraint than then, so why concentrate on that over latitude?
While not the same, latitude and DR are not necessarily mutually exclusive. Ditching ISO performance at even moderately high ISO for (at best) modest gains in DR does not benefit most (if not practically all) photographers.
edhannon
Senior Member
To be sure we are talking in same terms, when I say that clipping occurs I mean that the brightest part of a scene that I want to retain detail is clipped by the camera.
In your scenario, clipping will occur in the amount you compute when ISO is raised if the brightest part of the scene is at or above the saturation level at base ISO. In this case, I agree that raising ISO is not appropriate. Other means of capturing the high dynamic range scene must be used.
However, to the extent that highlights are below the saturation level at base ISO, raising ISO (in a non-ISO-invariant camera) will not reduce the dynamic range.
In your scenario, clipping will occur in the amount you compute when ISO is raised if the brightest part of the scene is at or above the saturation level at base ISO. In this case, I agree that raising ISO is not appropriate. Other means of capturing the high dynamic range scene must be used.
However, to the extent that highlights are below the saturation level at base ISO, raising ISO (in a non-ISO-invariant camera) will not reduce the dynamic range.
Iliah Borg
Forum Pro
The question of what affects the "Dynamic range", as I understood it, is about the camera. Which in this context means the highlights are already at the maximum, or so I thought.
rubank
Senior Member
I agree that higher ISO setting doesn´t have to mean clipped highlights - you contol that with exposure.
However, the DR will still be reduced due to excessive noise in the shadows.
bclaff
Forum Pro
No, as has been said, DR is reduced because highlights are clipped, noise in the shadows actually improves.
rubank
Senior Member
Sigh!
Try reading "edhannon"s post again.
bclaff
Forum Pro
Sigh!
Try reading Iliah's post again.
EDIT: There are some flaws, but this may help to explain the general concept well (start at 14:25).
Others have given good answers, but I'll add a few bits.
"Dynamic Range" has several definitions depending on context. A few versions:
- Scene's Dynamic Range: Nothing to do with the camera. The difference in luminance between the brightest and darkest areas of the scene. For example, a frame with both the sun and dark shadows (like a mountain side away from the sun) would have high dynamic range.
- Engineering Dynamic Range: Nothing to do with the scene. Sensor's noise floor to sensor's point of saturation. The lowest number of photons that can be converted to a signal above noise vs. max photons that can be distinctly counted.
- Photographic Dynamic Range: The difference between "black" and "white." Black is the brightest point where an area can no longer be distinguished from surrounding noise; and white is the darkest point where an area is just about to 'blow highlights.' Anything 'darker' than black is indistinguishable (shadow clipping); and anything 'brighter' than white is also indistinguishable (highlight clipping).
As far as what specifically affects PDR, here are a few:
-The scene. This is often overlooked. Many scenes simply do not have high dynamic range. One extreme example: take a picture of a uniformly colored wall. It will have a low dynamic range, and a camera's DR is not as relevant. Now take a picture of a sun in-frame rising over a mountain--and DR is suddenly very important.
-The sensor. Different designs offer different DR responses. Physical characteristics have the most bearing here: sensor size, sensor's efficiency in converting between photons & signal, heat profiles, signal interference, etc.
-The "black" noise floor. This has 2 main causes:
- Electronic noise (which usually decreases with increased ISO; and increases with longer shutter speeds)
- Shot noise (which decreases with increased exposure).
-The "white" signal ceiling. This decreases as ISO is increased, since ISO usually means a signal/noise is amplified more without raising the point of saturation. Why not raise the saturation point as well? Remember that if you raise the point of saturation, everything else gets darker--you've moved your goal posts further. So you'd have to raise this 2x as much as your amplification of the signal for the same brightness. Incidentally, this is also largely why larger sensors have greater DR than smaller sensors. Smaller sensors have less noise; but they also have lower points of saturation.
-The exposure & techniques. (beyond what was mentioned above). A few:
- Lenses can add flare--which decreases DR by diverting light to areas that should be dark.
- Image processing can compress local DR to increase an entire frame's DR.
- Bracketing & stacking can reduce shot noise & increase signal (additive). Dark frame subtraction can remove electronic noise (subtractive). Both expand DR.
- (Reminders):
- Raising ISO accelerates the exposure needed to get to saturation = less DR
- Longer shutter speeds increases electronic noise; but also decreases shot noise
- More light transmission (~aperture) decreases shot noise, but can increase flare
Last edited:
edhannon
Senior Member
I think that we are talking about two different ranges:
Iliah is talking about the potential dynamic range from usable noise level to maximum possible saturation level.
I am talking about the effective range that I can capture from a given scene. If the camera is not ISO-Invariant and there is headroom at base ISO at the exposure time and aperture I need/want to use, then I can capture a greater range of the scene (at the same exposure time and aperture) by raising the ISO setting.
Iliah is talking about the potential dynamic range from usable noise level to maximum possible saturation level.
I am talking about the effective range that I can capture from a given scene. If the camera is not ISO-Invariant and there is headroom at base ISO at the exposure time and aperture I need/want to use, then I can capture a greater range of the scene (at the same exposure time and aperture) by raising the ISO setting.
bclaff
Forum Pro
Stated this way, "and there is headroom", then yes; because you're not using the lost highlights and you are able to get deeper into the shadows.
rubank
Senior Member
If you, as you raise the ISO, adjust exposure so as not to clip highlights - that means reduced exposure, i.e. the sensor gets less light. What do you think happens to shadow noise with less light?
edhannon
Senior Member
No, I adjust exposure at base ISO. If the exposure time needs to be reduced or a smaller aperture is needed for photographic reasons, I put exposure where I need it to be and then adjust ISO up to move highlights to jut below saturation. For a non-ISO-invariant camera, I believe that this provides the best capture range.
rubank
Senior Member
OK, so I may have misread you.
Yours is not my MO, but to each his own
The way I use my camera high ISO means blocked shadows - not blown highlights.
Funny enough, it often - far from always - means my chosen exposure agrees with what the camera in auto would set. Who´da thought
Last edited:
Sensor saturation point is not relevant for consumer cameras.
That's much better.
Black is the lowest sensor readout where minimal necessary resolution is still maintained. It is not a fixed value.
DR is always relevant, at least somewhat. To check that, compare noise on shots with cameras having different dynamic ranges, or with the same camera but setting different ISO and exposing according to camera meter.
Sensor size is relevant mostly when DoF is not an issue.
You meant to say photon shot noise. Electronic noise also include shot noise component.
Not always, depending on the design.
You lost me here. More light is more photon shot noise. Did you mean to say increases SNR?
On the bright side, the noise in the clipped highlights is zero.
Similar threads
- Question
Keyboard shortcuts
- f
- Forum