Not that I see a big advantage in noise at high ISO (low light performances) fromthe cameras well known to have good DR...
-
Welcome to the new forums! Please read this first. For known issues we are working to resolve, click here.
POLL: Dynamic Range - Do we need more?
- Thread starter composed
- Start date
Not that I see a big advantage in noise at high ISO (low light performances) fromthe cameras well known to have good DR...
Rick Knepper
Forum Pro
Maybe he meant shooting in low light at base ISO then lifting the shadows. Dynamic range is all about base/low ISOs. I think of low light performance when the photographer wants to hold his camera as high ISO noise characteristics.
Not that I see a big advantage in noise at high ISO (low light performances) fromthe cameras well known to have good DR...
--
Once you've done fifty, everything else is iffy.
Mark S Abeln
Forum Pro
Decades ago, just about the fastest lens you could get was f/1.2; today, you'd be hard finding anything wider, and so there has been little progress in lens aperture.
Decades ago, many if not most serious photographers used medium format; nowadays, the so called full-frame cameras are the cat's meow, but our ancestors would have dismissed these as "small format"; even the largely unaffordable so-called medium format sensors are rather small compared to the old days, and so we've taken a step back in sensor sizes.
Decades ago, ISO 400 was the most common fast film speed, and some brave folks would push monochrome film to ISO 800 or 1600, but the results were a noisy mess. Nowadays, ISOs in the tens or hundreds of thousands are common, with good full-color image quality. That's where the progress has been found. Good high ISO performance is due to high dynamic range.
Exactly... if I'm in low light I'd rather shot directly at ISO 6400 than at ISO 100 and then lifting shadows by 6 stops...Maybe he meant shooting in low light at base ISO then lifting the shadows. Dynamic range is all about base/low ISOs. I think of low light performance when the photographer wants to hold his camera as high ISO noise characteristics.
Not that I see a big advantage in noise at high ISO (low light performances) fromthe cameras well known to have good DR...
--
Once you've done fifty, everything else is iffy.
John Sheehy
Forum Pro
Not really. DR has nothing directly to do with absolute SNR.
Assuming that all camera noise was non-correlated, then the DxO DR charts would reflect read noise relative to signal, but that is only because they are not DR charts anymore, after they are corrected for "measured ISO".
They data points are about engineering DR (or something close to it). The trends are an inverse proxy for read noise.
DR by itself at an ISO setting tells us nothing certain about high-ISO-index performance. Much of the DR can be in extended headroom, which does not aid SNR at middle gray for the ISO exposure index.
John Sheehy
Forum Pro
How about we stop referring to "high ISO (exposure index) performance" as "low-light performance". Low-light performance can be base ISO with ETTR and a 0.5 second exposure.
That is a lot of it. Shallower DOF is the real association with more light collection at a given exposure time with a larger sensor. If you are stopping down for increased DOF, that benefit vanishes, and any possible optical advantage of using more of the image circle disappears, as diffraction becomes the main blur source, and is 1.6x as strong as on an APS-C 1.6x sensor, with the same FOV, DOF, and shutter speed.
John Sheehy
Forum Pro
This discussion is a big mess.
Low light does not equate to high ISO settings or high ISO exposure indices.
DR does not determine performance at high exposure indices.
A camera at a given ISO can get a stop more DR just because the manufacturer decided to use the analog gain of half that ISO, but chose to leave the RAW data alone instead of multiplying it by 2x and clipping a stop away.
In fact, if Canon had chosen to implement only HTP in their cameras, and didn't label it as such, they would have a stop more DR at high ISOs, without any improvement in noise for a given high-ISO exposure index.
The size of the sensor is not quite right. It is pixel size that makes it more sensitive to low light. Of course, if you compare an APS-C sensor with 24M pixels, and a Full Frame with 24M pixels, the pixel size is going to be automatically larger due to the larger real estate available.
With CCD cameras, we use something called "Binning" to increase sensitivity. For example, 4 pixels can be ganged together to create one really big pixel with corresponding increase in sensitivity. I do this when originally setting up and not knowing how far from actual focus I might be. Of course, when binning, resolution goes down.
alanr0
Senior Member
Agreed.
One could equally say that dynamic range has increased as a side effect of improved low light performance, but this too is a misleading over-simplification.
In crude terms, engineering dynamic range is the ratio of pixel saturation capacity to read noise.
Low light performance is determined by quantum efficiency, read noise, sensor and pixel size, and by the light collection efficiency of the optics.
Much of the improvement in both dynamic range and low-light performance for DSLRs has come from substantial reductions in read noise, which benefits both. Generally, low ISO read noise has improved more than high ISO read noise, favouring low ISO dynamic range.
There have been improvements in quantum efficiency and well capacity per unit area, but these have been more modest, and impact low light performance and dynamic range respectively.
An increase in sensor size is beneficial for both, given similar pixel count and a lens aperture not constrained by depth of field concerns.
John Sheehy
Forum Pro
That's debatable. The D500 has less or equivalent high-ISO read noise per unit of sensor area than many recent FF cameras with much larger pixels. There is a prototype sensor with 1.4 micron pixels (450MP if it were full-frame) with about 0.22 electrons read noise, which, if lowered to 0.16 electrons basically becomes zero read noise, as the histogram hump for each electron value is narrow enough to prevent count errors.
What we're seeing now is that some cameras with large pixels have less read noise per unit of sensor area, months or years before denser sensors catch up. Part of that may be due to the fact that it is generally more expensive, speedy, high-end "pro" action FF cameras that have the biggest pixels, but those cameras have a higher engineering budget for better electronics.
If the pixels are combined in hardware to be read as one charge, yes, some parts of noise are weaker relative to the same exposure. Post-gain noise might actually be worse, though. Post-gain noise tends to be the least with very high pixel counts. That's why image-level DR in CMOS cameras with off-sensor ADCs tends to track with pixel count.
At least you get something in return. Binning after readout only loses spatial data with nothing gained except the illusion of a smoother image.
bclaff
Forum Pro
You didn't specify whether you're referring to pixel level ("Engineering" Dynamic Range (EDR) or a normalized measure such as Photographic Dynamic Range (PDR).
I assume you mean something like PDR (or DxOMark Landscape score).
You also didn't specify low ISO setting (lot's of light and using high dynamic range) versus high ISO setting (low light but requiring some minimum dynamic range).
I assume you mean low light / high ISO setting.
I think that despite the fact that most cameras have more PDR at base ISO than many of us need more PDR is desired primarily because it raises the ISO setting at which you can get reasonable PDR ("a rising tide raises all boats").
--
Bill ( Your trusted source for independent sensor data at PhotonsToPhotos )
I assume you mean something like PDR (or DxOMark Landscape score).
You also didn't specify low ISO setting (lot's of light and using high dynamic range) versus high ISO setting (low light but requiring some minimum dynamic range).
I assume you mean low light / high ISO setting.
I think that despite the fact that most cameras have more PDR at base ISO than many of us need more PDR is desired primarily because it raises the ISO setting at which you can get reasonable PDR ("a rising tide raises all boats").
--
Bill ( Your trusted source for independent sensor data at PhotonsToPhotos )
No doubt... If folks really care about dynamic range they'll be processing from RAW and learning about how best to get detail out of the shadows and highlights. Still, there are going to be certain situations where even if you're pulling the most possible DR from the shot that more is useful... so you can't really have too much.
Personally, I often treat my images so that they have a little less DR, with shadows blocking up and creating dramatic negative space. Still, I like to have the option of going the other way with it as well...
John Sheehy
Forum Pro
If we are getting better DR partly because noise is going down, far below middle gray (it can also come from increased headroom, which we may or may not use), then it's just a gift, whether you understand it or not, because you have less noise in the most noise-prone tonal ranges.
If you understand DR, then you can count on it in your exposure risk assessment. Why shoot at ISO 800, if ISO 100 "under-exposed" 3 stops gives little extra noise, and a smaller compressed RAW file, and room in the highlights to save your image if, for example, a flash of light occurs during your exposure?
gianstam
Senior Member
Thanks for clarifying
I'd be interested to see what you think of using the Lightroom merge feature. Instead of the effect-y results that often come from Photomatix, I find the LR merge to be a bit more natural looking and yet it still retains more detail in the shadows.
gianstam
Senior Member
Tom Axford
Forum Pro
I would agree with that!
This thread is in the Open Talk Forum, yet many people seem to be trying to treat statements with the degree of precision that you might expect in the Photographic Science and Technology Forum. I don't think it is really very helpful to expect every statement in this sort of discussion to have a high level of scientific rigour and accuracy. A lot of people just stop reading these nit-picking arguments.
The only defence of my statements above that I think is worth making is that they were intended to express a correlation that exists in practice, not a strict cause and effect.
Rick Knepper
Forum Pro
Depends on how you define "experienced" photographers. To me, an experienced photographer shooting RAW includes having a healthy knowledge of post processing unless that person is a pro who has a dedicated processor back in the office.
This is a fail in terms of logic. If you are just beginning to understand PPing, how can you make a judgment as to what experienced photographers know or do not know?
I think I understand your 'main' point though, that is, PPing is an important component of the image making work flow, that one can work with the DR one has (until one runs out of that DR).
--
Once you've done fifty, everything else is iffy.
Last edited:
Similar threads
Keyboard shortcuts
- f
- Forum