A sensor and a noise

JimBim

Well-known member
Messages
221
Reaction score
35
Location
Kaunas, LT
New cameras have a lower noise. Why? Because of a sensor, or a software? Can one say how much the sensor is responsible and how much the software is responsible for fighting with the noise? Can one say, that by shooting in RAW with the older sensor and PP say in LR I'll get the same result as from the newer camera?

Audrius
 
I think the answer is more complex

Newer sensors deliver lower noise, more even black-levels and smaller grain per pixel.

The switch to cmos has seen rapid development of on-chip and near chip non destructive Nr pre RAW all cameras do this as cmos output is unusable prior to cleaning up at higher ISO's hence why ccd was preferred in early days.

As cameras ability to shoot hi ISO has improved so have converters ability to deal with residual noise during the PP stage this whether done in camera or PC is destructive by nature but very effective.

If you take an old RAW and tune your PP to it (different NR processes work better with different sensors) then you can get a marginal improvement in the result but not match current cameras.
 
Most of the noise in the images we see today come from shot noise (randomness of how the light strikes the sensor). Newer sensors can capture more of the light striking the sensor (more efficient) . Because more light is captured you have a better sampling of the light data kind of like using a larger group of people when doing a survey to equal out the random oddities that can affect a smaller group of a survey. Software cannot give you more light data but with better and newer software can reduce the appearance of the noise with the light you have gathered.
 
I think the answer is more complex

Newer sensors deliver lower noise, more even black-levels and smaller grain per pixel.

The switch to cmos has seen rapid development of on-chip and near chip non destructive Nr pre RAW all cameras do this as cmos output is unusable prior to cleaning up at higher ISO's hence why ccd was preferred in early days.

As cameras ability to shoot hi ISO has improved so have converters ability to deal with residual noise during the PP stage this whether done in camera or PC is destructive by nature but very effective.

If you take an old RAW and tune your PP to it (different NR processes work better with different sensors) then you can get a marginal improvement in the result but not match current cameras.
I would agree with this assessment. The better the output of the imager, the more accurate the guesses of the RAW converter become.

If you are holding out hope that a RAW converter will come along and bring an older camera up to modern specs, it won't happen.
 
Have you ever noticed that all the new cameras have blurry raws ? which equals less noise ? if you added raw nr to the k7 shot it would hold its own. what a con.

cheers don







--
Pentax K7, Panasonic fz150, Olympus XZ1, my main toys.
 
I don't think so

If you look at the blocks of red and black in the k7 Vs the d7100 the K7 is blotchier with more noise and less data.

With nearly half as many pixels to work with this is disastrous for a converter, Unlike you they can't 'recognize' the image and fill the gaps from memory partially ignoring noise obliterated data.

From your example above I'd expect the d7100 to be able to recover detail 3X-4X finer than the k7 can at any ISO.

Quite impressive when it has twice the pixel density and smaller photo sites.
 
awaldram said:
I don't think so

If you look at the blocks of red and black in the k7 Vs the d7100 the K7 is blotchier with more noise and less data.

With nearly half as many pixels to work with this is disastrous for a converter, Unlike you they can't 'recognize' the image and fill the gaps from memory partially ignoring noise obliterated data.

From your example above I'd expect the d7100 to be able to recover detail 3X-4X finer than the k7 can at any ISO.

Quite impressive when it has twice the pixel density and smaller photo sites.

--
My PPG
http://www.pentaxphotogallery.com/artists/andrewwaldram
My Photo Stream
Andrew Waldram
1x.com
1x.com • In Pursuit of the Sublime
d7100 left quite impressive LOL ? iso 100 200% crop

cheers don







--
Pentax K7, Panasonic fz150, Olympus XZ1, my main toys.
 
I don't think so

If you look at the blocks of red and black in the k7 Vs the d7100 the K7 is blotchier with more noise and less data.

With nearly half as many pixels to work with this is disastrous for a converter, Unlike you they can't 'recognize' the image and fill the gaps from memory partially ignoring noise obliterated data.

From your example above I'd expect the d7100 to be able to recover detail 3X-4X finer than the k7 can at any ISO.

Quite impressive when it has twice the pixel density and smaller photo sites.

--
My PPG
http://www.pentaxphotogallery.com/artists/andrewwaldram
My Photo Stream
http://www.flickr.com/photos/awaldram/
1x.com
http://1x.com/artist/awaldram/wall
d7100 left quite impressive LOL ? iso 100 200% crop

cheers don



--
Pentax K7, Panasonic fz150, Olympus XZ1, my main toys.
I am just waiting for the plastic dolls head to come into play, time for some popcorn



--
The Camera is only a tool, photography is deciding how to use it.
The hardest part about capturing wildlife is not the photographing portion; it’s getting them to sign a model release
 
d7100 left quite impressive LOL ? iso 100 200% crop

cheers don

Not sure what your trying to say , What I see left image very accurate exposure , very sharp hi resolution. if you expected the blacks to be black you maybe need to read up on how a camera works out exposure.

Right hand image WB way off ( no 18% grey here) purple tinge to image, bad smudging and awful CA.

Neither camera is showing appreciable noise so not sure what your trying to show?

Or do you think the detail the d7100 has captured in the dark (paper multch or screen noise) is in camera noise.?

--
My PPG
http://www.pentaxphotogallery.com/artists/andrewwaldram
My Photo Stream
http://www.flickr.com/photos/awaldram/
1x.com
http://1x.com/artist/awaldram/wall
 
Last edited:
Last edited:
d7100 left quite impressive LOL ? iso 100 200% crop

cheers don

Not sure what your trying to say , What I see left image very accurate exposure , very sharp hi resolution. if you expected the blacks to be black you maybe need to read up on how a camera works out exposure.

Right hand image WB way off ( no 18% grey here) purple tinge to image, bad smudging and awful CA.

Neither camera is showing appreciable noise so not sure what your trying to show?

Or do you think the detail the d7100 has captured in the dark (paper multch or screen noise) is in camera noise.?
everyone complains about the low iso noise the d7100 poduces in blue skies so I guess its the same in blacks, these images are from dpr you can down load them and see for yourself, I did and was interested in the resolution tests , if you go to the dpr review raw res test states the k7 can resolve 3450 , pretty good for a 14meg sensor.

cheers don



--
Pentax K7, Panasonic fz150, Olympus XZ1, my main toys.
 
Have you ever noticed that all the new cameras have blurry raws ? which equals less noise ? if you added raw nr to the k7 shot it would hold its own. what a con.

cheers don



--
Pentax K7, Panasonic fz150, Olympus XZ1, my main toys.
don, please select a region with more fine details!

for example:

ba11d38711a3494f875e875de0a8c7a0.jpg

cheers anton
 
Have you ever noticed that all the new cameras have blurry raws ? which equals less noise ? if you added raw nr to the k7 shot it would hold its own. what a con.

cheers don



--
Pentax K7, Panasonic fz150, Olympus XZ1, my main toys.
don, please select a region with more fine details!

for example:

ba11d38711a3494f875e875de0a8c7a0.jpg

cheers anton

interesting that the em5 looks better than the k511, I think that the lens plays a big part in the final image, and for a 4 year old sensor the k7 does a pretty good job at iso 3200.

cheers don

--
Pentax K7, Panasonic fz150, Olympus XZ1, my main toys.
 
interesting that the em5 looks better than the k511, I think that the lens plays a big part in the final image, and for a 4 year old sensor the k7 does a pretty good job at iso 3200.
Sorry I don;t see that at all

in noise order best to worst

1 k5iis

2 7100

3 Olympus

4 K7

in detail retention (potential least noise for image size)

1 7100

2 k5iis

3 k7

4 Olympus

And though we have no idea compression ratios file sizes broadly match my synopsis.

Far from the K7 doing well, Given these images I would not shoot either the Olympus or K7 if Hi-ISO was a requirement.

It would be a toss up k5iis and 7100 and would probably (if not invested) plumb for the Nikon based on potential.

Actually on close examination move the K7 to bottom for detail retention the chroma noise in the Olympus fooled me it is retaining more detail than the k7.
cheers don

--
Pentax K7, Panasonic fz150, Olympus XZ1, my main toys.
--
My PPG
http://www.pentaxphotogallery.com/artists/andrewwaldram
My Photo Stream
http://www.flickr.com/photos/awaldram/
1x.com
http://1x.com/artist/awaldram/wall
 
Last edited:
Have you ever noticed that all the new cameras have blurry raws ? which equals less noise ? if you added raw nr to the k7 shot it would hold its own. what a con.

cheers don



--
Pentax K7, Panasonic fz150, Olympus XZ1, my main toys.
don, please select a region with more fine details!

for example:

ba11d38711a3494f875e875de0a8c7a0.jpg

cheers anton
interesting that the em5 looks better than the k511, I think that the lens plays a big part in the final image, and for a 4 year old sensor the k7 does a pretty good job at iso 3200.

cheers don

--
Pentax K7, Panasonic fz150, Olympus XZ1, my main toys.
The K-5 IIs looks is better than the K-5 II. I do not know whether it is the lens, the aa-filter or some test-variation.

c9b025eddfe4448aa6b1efea42310deb.jpg

cheers Anton
 
Ian Stuart Forsyth said:
Donald B said:
awaldram said:
I don't think so

If you look at the blocks of red and black in the k7 Vs the d7100 the K7 is blotchier with more noise and less data.

With nearly half as many pixels to work with this is disastrous for a converter, Unlike you they can't 'recognize' the image and fill the gaps from memory partially ignoring noise obliterated data.

From your example above I'd expect the d7100 to be able to recover detail 3X-4X finer than the k7 can at any ISO.

Quite impressive when it has twice the pixel density and smaller photo sites.

--
My PPG
http://www.pentaxphotogallery.com/artists/andrewwaldram
My Photo Stream
Andrew Waldram
1x.com
1x.com • In Pursuit of the Sublime
d7100 left quite impressive LOL ? iso 100 200% crop

cheers don



--
Pentax K7, Panasonic fz150, Olympus XZ1, my main toys.
I am just waiting for the plastic dolls head to come into play, time for some popcorn

--
The Camera is only a tool, photography is deciding how to use it.
The hardest part about capturing wildlife is not the photographing portion; it’s getting them to sign a model release
feel free to post a 100% crop from one of your cameras.









100% crop


Pentax K7, Panasonic fz150, Olympus XZ1, my main toys.
 
Last edited:
interesting that the em5 looks better than the k511, I think that the lens plays a big part in the final image, and for a 4 year old sensor the k7 does a pretty good job at iso 3200.
Sorry I don;t see that at all

in noise order best to worst

1 k5iis

2 7100

3 Olympus

4 K7

in detail retention (potential least noise for image size)

1 7100

2 k5iis

3 k7

4 Olympus

And though we have no idea compression ratios file sizes broadly match my synopsis.

Far from the K7 doing well, Given these images I would not shoot either the Olympus or K7 if Hi-ISO was a requirement.

It would be a toss up k5iis and 7100 and would probably (if not invested) plumb for the Nikon based on potential.

Actually on close examination move the K7 to bottom for detail retention the chroma noise in the Olympus fooled me it is retaining more detail than the k7.
cheers don
 
New cameras have a lower noise. Why? Because of a sensor, or a software? Can one say how much the sensor is responsible and how much the software is responsible for fighting with the noise? Can one say, that by shooting in RAW with the older sensor and PP say in LR I'll get the same result as from the newer camera?

Audrius
There are multiple things at play with the sensor.

3 important ones are - what is the level of read noise which is superimposed on the data returned by the sensor - this has got lower.

How many electrons does the sensor record at maximum exposure. This has got higher.

When you take the first two together the signal to noise ratio of the sensors has improved.

There is a third factor - how many photons arriving at the sensor get converted into electrons. Some don't get through the various filters, some land on the gaps between Pixels, not every photon hitting the sensitive part of the chip results in an electron to count. If you keep everything else the same and double the number of recordable electrons you reduce the base ISO of the sensor - i.e. you need twice as much light to hit it to get maximum exposure, but you now have one stop more dynamic range, and a better signal to noise ratio. To keep the base ISO of the chip at a nice workable 100, the sensors need to convert more of the light which arrives.

So the sensors are better.

Noise reduction has also improved. If I fire up my *ist-D and get it to convert a PEF file it shot, the noise processing will be with a 2001/2 vintage low power embedded processor running 2003/4 vintage software (optimized for processing time and Power consumption). If I load that same image into a lightroom, noise reduction techniques have improved in 10 years, and they can put a lot more computation into the process.

Essentially your question is "Can I get everything with a processing upgrade" (this couldn't just be done in firmware because we need more computation, and taking minutes to process an image in the camera wouldn't be workable). To which the answer is you can get something, but not parity.
 
New cameras have a lower noise. Why? Because of a sensor, or a software? Can one say how much the sensor is responsible and how much the software is responsible for fighting with the noise? Can one say, that by shooting in RAW with the older sensor and PP say in LR I'll get the same result as from the newer camera?

Audrius
There are multiple things at play with the sensor.

3 important ones are - what is the level of read noise which is superimposed on the data returned by the sensor - this has got lower.

How many electrons does the sensor record at maximum exposure. This has got higher.

When you take the first two together the signal to noise ratio of the sensors has improved.

There is a third factor - how many photons arriving at the sensor get converted into electrons. Some don't get through the various filters, some land on the gaps between Pixels, not every photon hitting the sensitive part of the chip results in an electron to count. If you keep everything else the same and double the number of recordable electrons you reduce the base ISO of the sensor - i.e. you need twice as much light to hit it to get maximum exposure, but you now have one stop more dynamic range, and a better signal to noise ratio. To keep the base ISO of the chip at a nice workable 100, the sensors need to convert more of the light which arrives.

So the sensors are better.

Noise reduction has also improved. If I fire up my *ist-D and get it to convert a PEF file it shot, the noise processing will be with a 2001/2 vintage low power embedded processor running 2003/4 vintage software (optimized for processing time and Power consumption). If I load that same image into a lightroom, noise reduction techniques have improved in 10 years, and they can put a lot more computation into the process.

Essentially your question is "Can I get everything with a processing upgrade" (this couldn't just be done in firmware because we need more computation, and taking minutes to process an image in the camera wouldn't be workable). To which the answer is you can get something, but not parity.
Thank you James, your explanation is very clear.

Audrius
 
I don't think so

If you look at the blocks of red and black in the k7 Vs the d7100 the K7 is blotchier with more noise and less data.

With nearly half as many pixels to work with this is disastrous for a converter, Unlike you they can't 'recognize' the image and fill the gaps from memory partially ignoring noise obliterated data.

From your example above I'd expect the d7100 to be able to recover detail 3X-4X finer than the k7 can at any ISO.

Quite impressive when it has twice the pixel density and smaller photo sites.

--
My PPG
http://www.pentaxphotogallery.com/artists/andrewwaldram
My Photo Stream
http://www.flickr.com/photos/awaldram/
1x.com
http://1x.com/artist/awaldram/wall
d7100 left quite impressive LOL ? iso 100 200% crop

cheers don



--
Pentax K7, Panasonic fz150, Olympus XZ1, my main toys.
I am just waiting for the plastic dolls head to come into play, time for some popcorn

--
The Camera is only a tool, photography is deciding how to use it.
The hardest part about capturing wildlife is not the photographing portion; it’s getting them to sign a model release
feel free to post a 100% crop from one of your cameras.



100% crop

Pentax K7, Panasonic fz150, Olympus XZ1, my main toys.
1200.jpg


I bite here’s 100% so what now and what does this prove ?



--
The Camera is only a tool, photography is deciding how to use it.
The hardest part about capturing wildlife is not the photographing portion; it’s getting them to sign a model release
 

Keyboard shortcuts

Back
Top