The knowledge base of the people here is just amazing. I'm trying to optimize my fashion shots and I feel like I have so much toearn about the "science" where just my "art" has carried me for so many years. Man I miss the smell of developer and using and enlarger.... It all seemed so simple then.
Thanks guys!!!!!!
Hi,Hi Rich, My IQ is diminished because of noise and a lack of definition caused by too high ISO. Next show I'll be halfing my ISO and hopefully having greater latitude while editing my RAW'S. I have no control of lighting while shooting runway, perhaps shutter speed and aperture adjustments within reason will get me where I need to be. Thanks
Erik, so a slower shutter speed yields better better mid-tones? Or did I over simplify it? TonyHi,Hi Rich, My IQ is diminished because of noise and a lack of definition caused by too high ISO. Next show I'll be halfing my ISO and hopefully having greater latitude while editing my RAW'S. I have no control of lighting while shooting runway, perhaps shutter speed and aperture adjustments within reason will get me where I need to be. Thanks
Most of the noise depends only on exposure. When raising ISO, we would generally reduce exposure.
In the darkest part of the image, 'readout noise' will come into play. Readout noise can vary quite a lot between CMOS sensor generation, especially at base ISO.
Late generation CMOS sensors can achieve high DR at base ISO. 'Engineering DR' is just full well capacity divided by readout noise.
This compares 'old' Canon with new 'Canon' and modern Sony. Huge difference in DR at low ISO but all are very close at high ISO.
What happens is that the 'pixels' have a very clean signal on all CMOS sensors. But, reading out the pixels is quite noisy on old CMOS.
At high ISOs the gain before readout is increased, at some level readout noise drops to minimum.
But, keep in mind, read noise just affects the extreme darks. Mid tone noise is probably pretty close on modern sensors of similar size. Also, more exposure yields better mid tones.
Best regards
Erik
--
Erik Kaffehr
Website: http://echophoto.dnsalias.net
Magic tends to disappear in controlled experiments…
Gallery: http://echophoto.smugmug.com
Articles: http://echophoto.dnsalias.net/ekr/index.php/photoarticles
You're quite right. I should have said decreasing shutter speed or the more precise 'increasing your exposure time.'
Hello,
. It is extremely ambiguous !!!
Thank you for your help!
Tony 
No, I would say it is OK to say so.Erik, so a slower shutter speed yields better better mid-tones? Or did I over simplify it? TonyHi,Hi Rich, My IQ is diminished because of noise and a lack of definition caused by too high ISO. Next show I'll be halfing my ISO and hopefully having greater latitude while editing my RAW'S. I have no control of lighting while shooting runway, perhaps shutter speed and aperture adjustments within reason will get me where I need to be. Thanks
Most of the noise depends only on exposure. When raising ISO, we would generally reduce exposure.
In the darkest part of the image, 'readout noise' will come into play. Readout noise can vary quite a lot between CMOS sensor generation, especially at base ISO.
Late generation CMOS sensors can achieve high DR at base ISO. 'Engineering DR' is just full well capacity divided by readout noise.
This compares 'old' Canon with new 'Canon' and modern Sony. Huge difference in DR at low ISO but all are very close at high ISO.
What happens is that the 'pixels' have a very clean signal on all CMOS sensors. But, reading out the pixels is quite noisy on old CMOS.
At high ISOs the gain before readout is increased, at some level readout noise drops to minimum.
But, keep in mind, read noise just affects the extreme darks. Mid tone noise is probably pretty close on modern sensors of similar size. Also, more exposure yields better mid tones.
Best regards
Erik
Lets look at a dark part of the image, 5EV below mid tones. 5EV is a factor of 32, so our well usage would 50000/8/32 -> 195.No, I would say it is OK to say so.Erik, so a slower shutter speed yields better better mid-tones? Or did I over simplify it? TonyHi,Hi Rich, My IQ is diminished because of noise and a lack of definition caused by too high ISO. Next show I'll be halfing my ISO and hopefully having greater latitude while editing my RAW'S. I have no control of lighting while shooting runway, perhaps shutter speed and aperture adjustments within reason will get me where I need to be. Thanks
Most of the noise depends only on exposure. When raising ISO, we would generally reduce exposure.
In the darkest part of the image, 'readout noise' will come into play. Readout noise can vary quite a lot between CMOS sensor generation, especially at base ISO.
Late generation CMOS sensors can achieve high DR at base ISO. 'Engineering DR' is just full well capacity divided by readout noise.
This compares 'old' Canon with new 'Canon' and modern Sony. Huge difference in DR at low ISO but all are very close at high ISO.
What happens is that the 'pixels' have a very clean signal on all CMOS sensors. But, reading out the pixels is quite noisy on old CMOS.
At high ISOs the gain before readout is increased, at some level readout noise drops to minimum.
But, keep in mind, read noise just affects the extreme darks. Mid tone noise is probably pretty close on modern sensors of similar size. Also, more exposure yields better mid tones.
Best regards
Erik
The explanation is that most of the noise is coming from photon statistics. Pixels that detect few photons will be noisier than pixels seeing more photons.
When we increase ISO, we generally reduce exposure. That means that there are fewer photons to detect, so photon statistics get worse, so noise goes up, or more correctly Signal Noise Ratio goes down.
The noise coming from photon arrival statistics is often called shot noise.
There is some noise coming from the pixels themselves and from reading out the pixels. That is sometimes called readout noise. That adds to shot noise. That addition is normally 'in quadrature', but we can ignore that in this simple discussion. The readout noise is sometime reduced when increasing ISO.
Some sensors have very low 'readout noise'. Some modern sensors have 'dual gain' where the sensor switches to 'high analogue gain'. Older sensors often reduce readout noise when increasing ISO. That essentially means 'clean pixel and noisy readout'.
Typically both full well capacity (FWC) and readout noise would be measured in electron charges (e-). Typical values for FWC would be around 40000 e-. Readout noise may be around 10e- on older sensors with external ADC-s while modern sensors with per column on sensor ACDs may have readout noise around 2-3 e- at base ISO.
SNR (Signal Noise Ratio) from shot noise is always the square root of the well capacity used.
Assume FWC a 50000 e-, and assume mid tones are 3 EV below saturation. Well usage will be 50000/8 that is 6250 e-. That will yield a signal noise ratio of sqrt(6250) -> 79.
In this case, signal would be 6250 and readout noise would be SQRT(6250) that is 79 e-.
Now, assume a readout noise is 10 e-. That would add to shot noise in quadrature, that is total noise would become sqrt(79*79 + 10*10) -> 79.7, leading to midtone SNR 6250/79.7 ->78.4. So readout would have very small effect on midtone noise.
Best regards
Erik
By "older" sensor, I'll assume it's pre Canon 1DX? I don't see a 3 to 5 times difference in my 1DX to my Z9. This may sound crazy but in some circumstances I'm finding my 1DX has less noise than my Z9.No, I would say it is OK to say so.Erik, so a slower shutter speed yields better better mid-tones? Or did I over simplify it? TonyHi,Hi Rich, My IQ is diminished because of noise and a lack of definition caused by too high ISO. Next show I'll be halfing my ISO and hopefully having greater latitude while editing my RAW'S. I have no control of lighting while shooting runway, perhaps shutter speed and aperture adjustments within reason will get me where I need to be. Thanks
Most of the noise depends only on exposure. When raising ISO, we would generally reduce exposure.
In the darkest part of the image, 'readout noise' will come into play. Readout noise can vary quite a lot between CMOS sensor generation, especially at base ISO.
Late generation CMOS sensors can achieve high DR at base ISO. 'Engineering DR' is just full well capacity divided by readout noise.
This compares 'old' Canon with new 'Canon' and modern Sony. Huge difference in DR at low ISO but all are very close at high ISO.
What happens is that the 'pixels' have a very clean signal on all CMOS sensors. But, reading out the pixels is quite noisy on old CMOS.
At high ISOs the gain before readout is increased, at some level readout noise drops to minimum.
But, keep in mind, read noise just affects the extreme darks. Mid tone noise is probably pretty close on modern sensors of similar size. Also, more exposure yields better mid tones.
Best regards
Erik
The explanation is that most of the noise is coming from photon statistics. Pixels that detect few photons will be noisier than pixels seeing more photons.
When we increase ISO, we generally reduce exposure. That means that there are fewer photons to detect, so photon statistics get worse, so noise goes up, or more correctly Signal Noise Ratio goes down.
The noise coming from photon arrival statistics is often called shot noise.
There is some noise coming from the pixels themselves and from reading out the pixels. That is sometimes called readout noise. That adds to shot noise. That addition is normally 'in quadrature', but we can ignore that in this simple discussion. The readout noise is sometime reduced when increasing ISO.
Some sensors have very low 'readout noise'. Some modern sensors have 'dual gain' where the sensor switches to 'high analogue gain'. Older sensors often reduce readout noise when increasing ISO. That essentially means 'clean pixel and noisy readout'.
Typically both full well capacity (FWC) and readout noise would be measured in electron charges (e-). Typical values for FWC would be around 40000 e-. Readout noise may be around 10e- on older sensors with external ADC-s while modern sensors with per column on sensor ACDs may have readout noise around 2-3 e- at base ISO.
SNR (Signal Noise Ratio) from shot noise is always the square root of the well capacity used.
Assume FWC a 50000 e-, and assume mid tones are 3 EV below saturation. Well usage will be 50000/8 that is 6250 e-. That will yield a signal noise ratio of sqrt(6250) -> 79.
In this case, signal would be 6250 and readout noise would be SQRT(6250) that is 79 e-.
Now, assume a readout noise is 10 e-. That would add to shot noise in quadrature, that is total noise would become sqrt(79*79 + 10*10) -> 79.7, leading to midtone SNR 6250/79.7 ->78.4. So readout would have very small effect on midtone noise.
Best regards
Erik
--
Erik Kaffehr
Website: http://echophoto.dnsalias.net
Magic tends to disappear in controlled experiments…
Gallery: http://echophoto.smugmug.com
Articles: http://echophoto.dnsalias.net/ekr/index.php/photoarticles
No, the 1DX was older generation.By "older" sensor, I'll assume it's pre Canon 1DX? I don't see a 3 to 5 times difference in my 1DX to my Z9. This may sound crazy but in some circumstances I'm finding my 1DX has less noise than my Z9.No, I would say it is OK to say so.Erik, so a slower shutter speed yields better better mid-tones? Or did I over simplify it? TonyHi,Hi Rich, My IQ is diminished because of noise and a lack of definition caused by too high ISO. Next show I'll be halfing my ISO and hopefully having greater latitude while editing my RAW'S. I have no control of lighting while shooting runway, perhaps shutter speed and aperture adjustments within reason will get me where I need to be. Thanks
Most of the noise depends only on exposure. When raising ISO, we would generally reduce exposure.
In the darkest part of the image, 'readout noise' will come into play. Readout noise can vary quite a lot between CMOS sensor generation, especially at base ISO.
Late generation CMOS sensors can achieve high DR at base ISO. 'Engineering DR' is just full well capacity divided by readout noise.
This compares 'old' Canon with new 'Canon' and modern Sony. Huge difference in DR at low ISO but all are very close at high ISO.
What happens is that the 'pixels' have a very clean signal on all CMOS sensors. But, reading out the pixels is quite noisy on old CMOS.
At high ISOs the gain before readout is increased, at some level readout noise drops to minimum.
But, keep in mind, read noise just affects the extreme darks. Mid tone noise is probably pretty close on modern sensors of similar size. Also, more exposure yields better mid tones.
Best regards
Erik
The explanation is that most of the noise is coming from photon statistics. Pixels that detect few photons will be noisier than pixels seeing more photons.
When we increase ISO, we generally reduce exposure. That means that there are fewer photons to detect, so photon statistics get worse, so noise goes up, or more correctly Signal Noise Ratio goes down.
The noise coming from photon arrival statistics is often called shot noise.
There is some noise coming from the pixels themselves and from reading out the pixels. That is sometimes called readout noise. That adds to shot noise. That addition is normally 'in quadrature', but we can ignore that in this simple discussion. The readout noise is sometime reduced when increasing ISO.
Some sensors have very low 'readout noise'. Some modern sensors have 'dual gain' where the sensor switches to 'high analogue gain'. Older sensors often reduce readout noise when increasing ISO. That essentially means 'clean pixel and noisy readout'.
Typically both full well capacity (FWC) and readout noise would be measured in electron charges (e-). Typical values for FWC would be around 40000 e-. Readout noise may be around 10e- on older sensors with external ADC-s while modern sensors with per column on sensor ACDs may have readout noise around 2-3 e- at base ISO.
SNR (Signal Noise Ratio) from shot noise is always the square root of the well capacity used.
Assume FWC a 50000 e-, and assume mid tones are 3 EV below saturation. Well usage will be 50000/8 that is 6250 e-. That will yield a signal noise ratio of sqrt(6250) -> 79.
In this case, signal would be 6250 and readout noise would be SQRT(6250) that is 79 e-.
Now, assume a readout noise is 10 e-. That would add to shot noise in quadrature, that is total noise would become sqrt(79*79 + 10*10) -> 79.7, leading to midtone SNR 6250/79.7 ->78.4. So readout would have very small effect on midtone noise.
Best regards
Erik
The more I learn, the more I realize how little I know
That would be true with 100% pixel views, but those are meaningless for comparing different pixel densities. The 1Dx has about 40% the pixel density of the Z9.
If "we" includes myself, I can be very brief!
I thought that the SD9 had the same DR at every ISO, and ISO is just metering/metadata, with the same gain/digitization?
Yes, if the DR is derived from noise floor and sensor saturation a la I.S.O. I was referring more to DR vs. ISO setting a la Bill Claff:
That is correct.
