View Full Version : camera noise


Ian Planchon
January 17th, 2011, 03:46 PM
does this seem like an odd amount of noise from my camera? I set it up according the recommended BBC settings (PP settings that is), this is the first time I have noticed all the noise. the black background probably helped me notice it.

http://i243.photobucket.com/albums/ff209/515productions/noise.jpg

if anyone has a recommended setting for this type of shot, let me know.

thanks!

Alister Chapman
January 17th, 2011, 04:07 PM
Looks normal to me.

I'd light the subject a little more rather than trying to do more in the PP's.

Doug Jensen
January 17th, 2011, 04:52 PM
Sorry, but it doesn't look good to me. It's blah.

The subject is underexposed by at least 1-2 stops and the blacks are not deep enough. That's probably why you're seeing more noise than usual.

Also, the subject could really benefit from a backlight on the shoulder and hat, plus a kicker on one of the the cheeks.

Ian Planchon
January 17th, 2011, 06:54 PM
i should have stated this was NOT a final shot. we were testing some eye light setups, not really working on the whole picture just yet. maybe it was because it was under exposed at the time. I just noticed it and grabbed that frame as an example.
would I be better off turning the PP off and adding contrast in post?

Doug Jensen
January 17th, 2011, 10:01 PM
i should have stated this was NOT a final shot.

Why bother posting it then? Why not post something that you feel is finished if you're going to ask for comments. There is nothing in your original post to indicate that it is a half-baked work in progress.
Once again, I feel like I've been suckered into wasting time with nonsense.

Ian Planchon
January 17th, 2011, 10:03 PM
jeez. someone woke up on the wrong side of the bed. next time you feel you got suckered, just ignore the post and move on. normally you are a pretty nice guy, I must have caught you on a bad day.

the point of the post is I have seen under exposed shots come from this camera, and they have never been this noisy. but if this looks like typical under exposed noise, then I can rest at ease.

Alister Chapman
January 18th, 2011, 02:03 AM
Ian, that was a bit like asking for advice about your car when it isn't running right, but neglecting to tell anyone that there is no fuel in it.

We can only go on the information provided to us, if important parts of that are missing, incorrect assumptions get made. You made no mention of the fact that this was not how the shot was going to be lit or exposed, yet asked for advice on how to make the shot look better. The first advice from any self respecting cameraman would be to sort out the lighting first. PP's are not a replacement for a well lit, well composed shot, just a complimentary tool.

Vincent Oliver
January 18th, 2011, 02:49 AM
It's not your day today Ian, take it on the chin like a man and try again.

Bo Skelmose
January 18th, 2011, 03:42 AM
Hi Ian
I cannot see any noise in the JPEG picture, but you would probably have some noise anyway.
There is some on this forum that knows a lot more about this than me, but it seems as if the have chosen not to give you any answers, you can use.
There are 2 kinds of noise - one type from the compression and one from the Chip.
The easiets way to get around it is to have some well light objects. Then you would probably not notice it.
Compression noise could be reduced when recording on another media like KIA mini or the nanoflash.
Chip noise could be reduced by recording at -3dB. The gain level 0dB is chosen because the chip produce a certain mount of noise there often lying about -60dB and a certain light sensitivity - it is needed in the data to sell the camera. Changing to -3dB would raise the signal/noise difference and lower the light sensitivity.

I have a HPX 2100 in which I mounted a AVC-Intra 100 board - I thought that recording in 10 bit would help me with the noise but it did not.
On my EX350 and the EX3 I have reduced the noise in big, one colour parts, by recording on my nanoflash.

Please correct me if I am wrong on this - I do not like noise either.

Alister Chapman
January 18th, 2011, 07:46 AM
The gain level 0dB is chosen because the chip produce a certain mount of noise there often lying about -60dB and a certain light sensitivity - it is needed in the data to sell the camera.

That is incorrect information. The unity (0db) gain level is normally determined by sensor linearity and dynamic range. 0db is normally the point at which maximum dynamic range is achieved. Reducing the gain beyond this point while certainly reducing the noise also reduces the dynamic range and linearity of the camera. The EX series of cameras have a noise figure of about 54db, very few HD cameras achieve 60db, certainly none with less than 2/3" sensors that I know of.

To gain a true benefit from 10 bit recording you want a noise figure better than around approx 58db. Less than 58db and the imager noise levels are greater than the 10 bit sample size, so in effect 8 bits will record just as much real picture information as 10 bit. With a noise figure better than 58db 10 bit can bring an advantage, but you really want better than 60db for 10 bit to really start to become a significant improvement.

Bo Skelmose
January 18th, 2011, 11:12 AM
Great thanks for the information Alister. I am not very familiar with the small chip standards. So if you can live with the lower linearity of the camera you would have lower noise at -3 dB. In a full studio light situation, you would be able to control this lower linearity with the light and keep it unvisible.

Alister Chapman
January 18th, 2011, 02:48 PM
Absolutely Bo. In a studio environment use your lighting to control your lattitude. Then you can use -3db without any issues (other than the Cinegamma level drop). Use standard gamma 3 and turn the knee off for the most natural looking image.

If your still struggling with noise turn off the detail correction or leave it on at a reduced level with the crisping turned up to help keep detail off the noise.

-3db drop in lattitude is only half a stop, so it's not big. However if you are using Cinegammas 1,3 or 4 note that the peak white level also drops down by 3db, so you end up loosing a full stop.

Mitchell Lewis
January 19th, 2011, 09:08 AM
I shoot the majority of projects at -3db and Cini Gamma4. (didn't know I was loosing a full stop....that's a LOT). But I rarely get noise in my videos.

Mark Savage
January 19th, 2011, 01:25 PM
jeez. someone woke up on the wrong side of the bed. next time you feel you got suckered, just ignore the post and move on. normally you are a pretty nice guy, I must have caught you on a bad day.

the point of the post is I have seen under exposed shots come from this camera, and they have never been this noisy. but if this looks like typical under exposed noise, then I can rest at ease.

Ian, I don't think Doug was being unreasonable with his comment: 'Why post it?' Noise will be present when something is considerably underexposed. It's probably more respectful of people's time to post something that you believe is satisfactory (as in: you feel you've done your bit to make it good) before seeking an opinion. Then, the opinions will be a lot more valid and helpful.

To me, it's all about lighting it adequately or over-adequately so you have room to pull things down or flag before having to resort to electronic assistance.

I've never found the camera "noisy" after I've done the necessary lighting.

Still, we're always learning by continuing to do.

Mitchell Lewis
January 19th, 2011, 01:34 PM
Some other suggestions:

1) Use the built-in histogram. It really shows if you're under or overexposed.
2) Use the built-in Zebras. But note what they are set to as they can be misleading if you don't have them set to the warning level you're expecting.

Mark Savage
January 19th, 2011, 01:39 PM
I second Mitchell's histogram suggestion.

I found that once I had gained a pretty deep understanding of the histogram's parameters in relation to what I'm ultimately getting, things exposure-wise improved because it enabled me to fine tune more.

And, yeah, the Zebras are quite subjective (based on percentage choice) and you need to be rigorous in how you read and interpret them.