Camera light sensitivity specifications

[J-Telectro 0]

Please forgive some of my crazy questions, but I am still new to this.

 

I'm trying understand how to compare the light sensitivity of various cameras based

on the manufacturers specifications.

 

1. What does the AGC value mean? Is this the amount of boost in gain for a

given light level? For instance does an AGC value of 50db mean that the

(voltage) gain from the sensor has been increased by a factor of 316?

Then for an AGC value of 40db the (voltage) gain has been increased

by only 100?

 

2. How does IRE relate to the light input in Lux? Is this a linear relationship?

For instance if camera A = 50 IRE for 1 Lux

and camera B = 30 IRE for 1 Lux

does this mean that camera A would be 3/5 Lux for 30 IRE?

How do you calculate the Lux value when you change from

one IRE value to another?

 

(This all assumes the "f" value for the lens remains the same)

[Stanislav 0]

Please forgive some of my crazy questions, but I am still new to this.

 

I'm trying understand how to compare the light sensitivity of various cameras based

on the manufacturers specifications.

It is almost impossible between specifications of different manufacturers.

1. What does the AGC value mean? Is this the amount of boost in gain for a

given light level? For instance does an AGC value of 50db mean that the

(voltage) gain from the sensor has been increased by a factor of 316?

Then for an AGC value of 40db the (voltage) gain has been increased

by only 100?

Not quite. 40dB is the maximal AGC value. It means that the signal from the videosensor can be amplified with the gain factor adjusted automatically in the range of 40dB to keep optimal image contrast.

2. How does IRE relate to the light input in Lux? Is this a linear relationship?

Yes but in case of the Gamma factor =1 and AGC is OFF.

For instance if camera A = 50 IRE for 1 Lux

and camera B = 30 IRE for 1 Lux

does this mean that camera A would be 3/5 Lux for 30 IRE?

In real cameras with Gamma=0.45 and AGC =ON this dependence is more complicated.

How do you calculate the Lux value when you change from

one IRE value to another?

It is not simple. Please see my article Illumination and camera sensitivity in CCTV. I hope it will be helpful.

[Fiona 0]

Hi Stanislav. I touched on light a little at college. As I recall, there was general talk of a paradigm shift in the way physics understood it (light). They talked about the obvious waves and wavelengths and then onto individual photons. It wasn't explained beyond this.

 

At the most basic level, if such a concept even exists, do we conceptualise the light wavelengths within the electromagnetic spectrum as a particle or a wave? And how does this relate to a photon?

[Soundy 1]

Of course, all the theory is irrelevant as long as manufacturers fudge their measurements and numbers...

[Stanislav 0]

Hi Stanislav. I touched on light a little at college. As I recall, there was general talk of a paradigm shift in the way physics understood it (light). They talked about the obvious waves and wavelengths and then onto individual photons. It wasn't explained beyond this.

 

At the most basic level, if such a concept even exists, do we conceptualise the light wavelengths within the electromagnetic spectrum as a particle or a wave? And how does this relate to a photon?

 

In my article I tried to explain only basic things of lighting and sensitivity which are useful in CCTV design. As a rule the light is considered as a superposition of wavelengths(spectrum) when we consider Spectral distribution and Spectral sensitivity. In light engineering and in CCTV engineering it is sufficient to consider light as a waves.

But the Light must be considered as a number of photons when we calculate noise of videosensors.

 

If you are really interested in this, I recommend excellent article

Nikolai Uvarov,"The secret of higher sensitivity CCTV cameras", "CCTV focus" Issue 23 - May/June 2003. It is really interesting reading if you are interested in the Physics.

 

See also David Elberbaum,"Unravel and undo the unreal CCTV camera specifications", "CCTV focus" Issue 37 - 2006

 

Of course, all the theory is irrelevant as long as manufacturers fudge their measurements and numbers...

 

Quite the contrary. Because of the wrong parameters in specification, who really want to investigate these questions must look into the theory and be able to measure camera sensitivity independently. It is not difficult. Here are the technique

Edited January 27, 2012 by Guest

[Fiona 0]

Stanislav, this report by David Elberbaum puts into question the principle specifications of any camera.

Elberbaum argues that many widely held and basic assumptions are wrong.

Where do we go from here?

Edited January 27, 2012 by Guest

[Stanislav 0]

Stanislav, this report by David Elberbaum puts into question the principal specifications of any camera.

Elberbaum argues that many widely held and basic assumptions are wrong.

Where do we go from here?

 

I agree with David Elberbaum, we can't use values of sensitivity from specifications. The best way is understanding what is the sensitivity and be able to measure sensitivity using open techniques.

 

Or you can rely on your practical experience only and avoid any calculations in this field. Many designers consider lighting and sensitivity very complicated field which can't be understood. In spite of this fact the designers do their work .

[Soundy 1]

Or you can rely on your practical experience only and avoid any calculations in this field. Many designers consider lighting and sensitivity very complicated field which can't be understood. In spite of this fact the designers do their work .

I wouldn't say it's too complicated to be understood, but there's more to a camera's low-light performance than just its sensitivity to photons.

 

Lens aperture directly affects how much light is collected.

 

So does shutter speed.

 

Electronic gain processing can have a major effect on a camera's actual performance under different lighting conditions.

 

So can electronic noise reduction and other such processing.

 

Even the quality (components, design, construction, QA) of the circuitry supporting the sensor can make a big difference in how a camera does with low light.

 

Two cameras may use the same sensor, the same lens, and one will still do better with low light because of better realtime image processing - boosting the gain, cleaning up the noise, etc.

 

Realistically, experience IS going to be your greatest ally most of the time.

[rory 0]

 

I agree with David Elberbaum, we can't use values of sensitivity from specifications. The best way is understanding what is the sensitivity and be able to measure sensitivity using open techniques.

the only way to know the sensitivity of a camera is to test it out, specs for cameras are all over the place, you get some that claim 100' IR when it only does 30', one that claims 80' Ir when it only does 40'. Bottom line is one cant rely 100% on specs to tell them how a camera will perform. One needs to actually get out in the field and test them.

[Stanislav 0]

Generally I agree with you, Soundy. But why do you think that the factors you mentioned can't be taken into account in calculation and modeling?

Please see my article above.

These factors are taken into account: Light spectrum and camera spectral sensitivity, Lens aperture, Shutter speed, AGC (Electronic gain processing ?), Gamma, Reading Noise and Photon Noise (in internal camera parametric model).

Even the quality (components, design, construction, QA) of the circuitry supporting the sensor can make a big difference in how a camera does with low light.

Real cameras have no such problems as a rule. In any case these factors are taken into account while practical measuring.

Two cameras may use the same sensor, the same lens, and one will still do better with low light because of better realtime image processing - boosting the gain, cleaning up the noise, etc.

I don't agree. The Sensitivity in 99% is determined by the Lens and the Sensor (with fixed exposure). Other processing can't increase sensitivity, it can improve some parameters (for example signal/noise ratio) at the expense of making worse other parameters (for example -resolution). I don't like cameras with complicated processing because their behaviour is unpredictable. I have to examine such cameras more carefully Many "realtime image processing" are advertising slogans only.

[Stanislav 0]

 

I agree with David Elberbaum, we can't use values of sensitivity from specifications. The best way is understanding what is the sensitivity and be able to measure sensitivity using open techniques.

the only way to know the sensitivity of a camera is to test it out, specs for cameras are all over the place, you get some that claim 100' IR when it only does 30', one that claims 80' Ir when it only does 40'. Bottom line is one cant rely 100% on specs to tell them how a camera will perform. One needs to actually get out in the field and test them.

 

You need not " get out in the field " to test them. Just make a simple stand in your room. Actually, you can make some simple tests even without my software. But understanding is needed in any case.

[rory 0]

I don't agree. The Sensitivity in 99% is determined by the Lens and the Sensor (with fixed exposure). Other processing can't increase sensitivity, it can improve some parameters (for example signal/noise ratio) at the expense of making worse other parameters (for example -resolution). I don't like cameras with complicated processing because their behaviour is unpredictable. I have to examine such cameras more carefully Many "realtime image processing" are advertising slogans only.

I agree with Soundy.

For example, when the specs for an $80 camera says it has a Sony Super HAD CCD and 100' IR, and a $200 camera has a Sony Super HAD CCD and 100' IR .. it matters little what the specs say and that they both claim to have the same CCD. Real world testing would prove the $200 camera to be much better in many ways - not just the IR either. There really is no way to calculate the light capabilities of cameras these days without actually testing the camera - if the specs were correct to the T then sure, but they are not. And then one also has to take into consideration how cameras are displayed differently on various DVRs.

[rory 0]

You need not " get out in the field " to test them. Just make a simple stand in your room. Actually, you can make some simple tests even without my software. But understanding is needed in any case.

Testing IR cameras indoors is next to useless (or any low light camera). You need a long wide area to test to really put the camera to task. I test in my car park, 70x60', and also test in a small area, and in the garage, and on a long narrow balcony - the difference between the car park and the other areas is like night and day with how the cameras perform, most suck in the car park, it really tells what camera is the better camera. To test cameras there is no other way but out in the field, thats one reason why the specs for cameras are so incorrect, most manufacturers themselves dont even go outside to test.

[rory 0]

You need not " get out in the field " to test them. Just make a simple stand in your room. Actually, you can make some simple tests even without my software. But understanding is needed in any case.

I should clarify, testing for real world apps should be done outside, unless ofcourse the app is only indoors

[Stanislav 0]

I agree with Soundy.

For example, when the specs for an $80 camera says it has a Sony Super HAD CCD and 100' IR, and a $200 camera has a Sony Super HAD CCD and 100' IR .. it matters little what the specs say and that they both claim to have the same CCD. Real world testing would prove the $200 camera to be much better in many ways - not just the IR either.

If these cameras have the same sensor and the same lens, same exposure time and these cameras have no problems, their sensitivity must be the same. I carefully tested tens of cameras of different price and I assert that the Image sensor and lens have much more influense on real sensitivity than camera price. Actually the price have no influence on the sensitivity .

But may be you used cheap imitation and your $80 camera has no really Sony Super HAD CCD? Or it has some problems?

 

Other question is that threshold parameter did you accept for estimation of sensitivity? It must be Signal/Noise ratio, NO IRE! If you use IRE, you fall into a trap.

IRE depends on maximal AGC gain, which can be different. I highly recommend my article to understand this, see Page 10 "Parameters, limiting image quality at defining sensitivity".

There really is no way to calculate the light capabilities of cameras these days without actually testing the camera - if the specs were correct to the T then sure, but they are not.

I don't deny real testing. Moreover I developed procedures and software for real testing. But we can't test all possible combinations of lens aperture, exposure time, light sources etc. We can test only one mode then calculate other situation using know physics laws.

 

And then one also has to take into consideration how cameras are displayed differently on various DVRs.

DVR can be examined too, but it is out of question of camera sensitivity.

[rory 0]

I agree with Soundy.

For example, when the specs for an $80 camera says it has a Sony Super HAD CCD and 100' IR, and a $200 camera has a Sony Super HAD CCD and 100' IR .. it matters little what the specs say and that they both claim to have the same CCD. Real world testing would prove the $200 camera to be much better in many ways - not just the IR either.

If these cameras have the same sensor and the same lens, same exposure time and these cameras have no problems, their sensitivity must be the same. I carefully tested tens of cameras of different price and I assert that the Image sensor and lens have much more influense on real sensitivity than camera price. Actually the price have no influence on the sensitivity .

But may be you used cheap imitation and your $80 camera has no really Sony Super HAD CCD? Or it has some problems?

 

Other question is that threshold parameter did you accept for estimation of sensitivity? It must be Signal/Noise ratio, NO IRE! If you use IRE, you fall into a trap.

IRE depends on maximal AGC gain, which can be different. I highly recommend my article to understand this, see Page 10 "Parameters, limiting image quality at defining sensitivity".

There really is no way to calculate the light capabilities of cameras these days without actually testing the camera - if the specs were correct to the T then sure, but they are not.

I don't deny real testing. Moreover I developed procedures and software for real testing. But we can't test all possible combinations of lens aperture, exposure time, light sources etc. We can test only one mode then calculate other situation using know physics laws.

 

And then one also has to take into consideration how cameras are displayed differently on various DVRs.

DVR can be examined too, but it is out of question of camera sensitivity.

 

Well if all cameras were the same then we only need to test one

[Stanislav 0]

You need not " get out in the field " to test them. Just make a simple stand in your room. Actually, you can make some simple tests even without my software. But understanding is needed in any case.

Testing IR cameras indoors is next to useless (or any low light camera). You need a long wide area to test to really put the camera to task. I test in my car park, 70x60', and also test in a small area, and in the garage, and on a long narrow balcony - the difference between the car park and the other areas is like night and day with how the cameras perform, most suck in the car park, it really tells what camera is the better camera. To test cameras there is no other way but out in the field, thats one reason why the specs for cameras are so incorrect, most manufacturers themselves dont even go outside to test.

 

Please try explain your reasons scientifically. You are really mistaken here, Rory. All professional tests are performed on stands, not in outdoor field. All camera parameters can be measured on laboratory stand.

[Stanislav 0]

Well if all cameras were the same then we only need to test one

 

I didn't write so. Please be accurate in the dispute. I assert that all cameras in my tests with the same sensors and the same lenses with the same shutter speed produce images with the same threshold signal/noise ratio (17dB in my test) at approx the same illumination.

 

Actually the type of image sensor is much more informative than sensitivity value in the spec.

Edited January 27, 2012 by Guest

[rory 0]

Please try explain your reasons scientifically. You are really mistaken here, Rory. All professional tests are performed on stands, not in outdoor field. All camera parameters can be measured on laboratory stand.

And that is why those tests are inaccurate for real world apps.

[Stanislav 0]

Please try explain your reasons scientifically. You are really mistaken here, Rory. All professional tests are performed on stands, not in outdoor field. All camera parameters can be measured on laboratory stand.

And that is why those tests are inaccurate for real world apps.

 

 

Right tests are accurate. But parameters in the specifications weren't obtained from real tests. Many manufacturers don't test their cameras! I have in my customers several camera manufacturers. They really don't know how to test cameras They don't need to test them. These parameters became more marketing than technic.

[rory 0]

 

Right tests are accurate. But parameters in the specifications weren't obtained from real tests. Many manufacturers don't test their cameras! I have in my customers several camera manufacturers. They really don't know how to test cameras They don't need to test them. These parameters became more marketing than technic.

 

Useful tests are done out in the field.

[Stanislav 0]

 

Right tests are accurate. But parameters in the specifications weren't obtained from real tests. Many manufacturers don't test their cameras! I have in my customers several camera manufacturers. They really don't know how to test cameras They don't need to test them. These parameters became more marketing than technic.

 

Useful tests are done out in the field.

 

OK, it is your opinion.

[rory 0]

Useful tests are done out in the field.

 

OK, it is your opinion.

Ofcourse, everything is just opinion.

I do my tests in the field though, as thats what people spending money on a product really care about.

Not what looks good on paper.

[Stanislav 0]

I do my tests in the field though, as thats what people spending money on a product really care about.

Not what looks good on paper.

 

I know, Rory

[Stanislav 0]

Today I have contacted with Nikolai Uvarov and he just gave us his great article in English.

Nikolai Uvarov,"The secret of higher sensitivity CCTV cameras", "CCTV focus" Issue 23 - May/June 2003. It is really interesting reading if you are interested in the physics.