What causes noise?

D

DanHD

Well-known member
What causes video noise, and why do lower end video cameras exhibit higher levels of noise than others.

Is it more prevalent when the camera is balanced for very warm or very blue light? (gained up color channels = noise)
Why is it more prevalent in low light, even with gain at Zero?



does the "gain" on the color channels caused by white balancing add noise, or is it insignificant?

what is the native color balance of the 170 CCD's, or are they equally sensitive
 
are you serious?
because lower end cameras have smaller price tags...

imagine a still camera. cranking the ISO (to accept lower light) increases the grain, as such do video cameras. but they introduce noise not grain. bigger the sensors, less light they need to expose a frame, thus lower the noise. but as sensors get bigger so does the price tag.
 
Visible noise is caused by low-level random electric fluctuations in the camera circuitry, caused by all sorts of things... leakage from other electronic circuits, RF, heat, cosmic rays, you name it. These fluctuations are present in all electronics. As a rule, higher-quality electronics have better noise shielding (thus lower noise); closer-packed components (read: smaller cameras) have more noise.

In general, this "low-level" random noise will be masked by a decent level signal (that is, by enough light). The darker the part of the image, the less signal strength (light) there is, and thus the more apparent the low-level noise is. In other words (with unrealistic numbers), the random noise may vary between 0 and .1 volt, while the dynamic range of the imager may be 0 to 5 volts. When there's a lot of light (say, generating 4 volts), the desired signal from the light makes the noise unnoticeable (a fluctuation around 2% of the overall brightness). When there's very little light (say, generating .5 volts), the fluctuation is much more apparent... it's about 20 percent of the overall brightness. With no light, the small fluctuation is all you see.

If the camera adds gain, what it does is amplify the output of each pixel... and that output contains BOTH the desired signal (light) and random low-level noise. So... the more gain applied, the higher the resulting level of the "low-level" noise.

Since the noise is random, it tends to cancel out more over a larger area. So, in general, a larger imager (chip size) at the same resolution will have lower noise. Cheap cameras have smaller imagers, and thus more noise.

In-camera or post- processing often involves raising levels of some aspect of the image intensity (to change color balance, etc). This of course adds noise.

"Black balancing" attempts to idfentify the noise inherent in the imager in the presence of no light, and correct for it... therefore reducing noise.

Hope this all makes sense!

-Barry
 
Last edited:
one thing that is very annoying is that everyone refer to blue as cold and yellow as warm, it's the opposite, and the 170 is daylight balanced.
 
are you sure the 170 is daylight balanced?

Does that mean it's less sensitive to blue by default?
 
DanHD said:
are you sure the 170 is daylight balanced?

Does that mean it's less sensitive to blue by default?
Click to expand...

Dan,

I am curious as to why you keep bringing this up. While I do not remember the name of the last thread you started but it certainly was all about the same thing. This question was answered not just once but many times. So I am really curious as to what the real question is.

Please advise,

Jan
 
Jan


because I'm recieving a ton of contradictory information

some say the camera is daylight balanced and not to shoot tungsten, and some say it's not


who do I believe
 
Believe your eyes. Set up a shot indoors under tungsten lighting, use the tungsten preset, and look at the footage.

Then set up outdoors, perform a manual balance under daylight, shoot the same thing, and look at the footage.

On an HPX170 you shouldn't see an appreciable difference in the noise level.
 
Okay, actually, give me a minute and I'll go put this to the test. The CamBelles, shot under tungsten and under HMI, same shot, on an HPX170, coming up...
 
Dan,

I said this in the previous thread. The camera performs well under both conditions. Barry's test should show you.

The problem is that you are listening to people that are relatining their experiences on different technologies and trying to relate them to the CCD. As I said in the previous thread, CMOS and CCDs are completely different. Do not assume that rules or situations for one will be like the other.

Best,

Jan
 
If the same test was conducted with a single chip CMOS sensor, would the tungsten image be slightly nosier? Seeing that there is less surface area to collect blue
 
As was mentioned in the other thread, the CMOS imager is different in its sensitivity to Tungsten, and thus may be noisier. This is where the other thread became confusing as many joined in with their experiences about the Red and the EX cameras being more senistive. I have not had a lot of experience with Tungsten and CMOS so I suggest going back to the other thread for the answer on this as it definitely was there.

All the best,

Jan
 
I don't have a CMOS camera to test with... hopefully someone out there with an EX1 or other CMOS camera, a chart, and HMI & tungsten lighting will perform a comparable test.
 
thanks for being patient with me guys


so the best way to avoid noise is to use more light than you'd think you'd need and underexpose a bit, in order to provide the chips with ample signal?
 
single chip cameras have less available surface area to collect blue and red light right? Compared to a three CCD system with equally sized chips
Click to expand...
If the chips are equally sized, then yes. A three-chip system can gather twice as much green light and four times as much red light and four times as much blue light than a single-chip system of equal size (assuming a Bayer pattern).
 
You must log in or register to reply here.
Back
Top