Jump to content

What causes video noise?


DS Williams

Recommended Posts

My point is though, 3 chip CCD cameras do well in tungsten balance. My friend and mentor owns an HPX2700 with AVC I board and just shot a national commercial on it. He chose to just tungsten balance in camera, rather than lose two stops filtering blue, because he didn't a difference in blue shadow and midtone noise. He was watching HD SDI out on a panasonic monitor. The RED one on the other hand has a very hard time with tungsten balance for some reason..

Link to comment
Share on other sites

  • Premium Member
I'll have to find the graph. A study was done. I'm going to try to dig it up. I do remember it showed CMOS sensors being more inherently daylight balanced.

That would be a function of the dyes used in masked single chip cameras, or of the dichroics in the optical blocks of three chip cameras. You hear mostly of single chip CMOS, though Genesis and F-35 are single chip vertical stripe CCD. You hear mostly of three chip CCD, though Sony's XDCam's are three chip CMOS.

 

The underlying silicon, be it CMOS or CCD, should get you from photons to electrons in very much the same way. It's what they do with the electrons thereafter that differs.

 

 

 

 

-- J.S.

Link to comment
Share on other sites

  • Premium Member
so is the RAW image from a bayer filtered single chip like the Red useable without interpolation?

 

No - as with any bayer image, including most DSLRs, and other non-bayer matrixed cameras such as F35 and Genesis.

 

is it true the Red isnt true 4k because of interpolation? Sorry for all the question

 

I am among those who find such claims spurious, to put it mildly. The objection is that terms like "2K" and "4K" have customarily been used to describe the resolution of things like film scans, which are invariably made using devices which sample every pixel for every channel. Bayer pattern sensors have, in the DSLR world, traditionally been measured in megapixels. What Red are doing is, at best, mixing terminology, and it's clear that they're doing so in order to imply a higher resolution for their camera than that of which it is actually capable. So no - take a 4000 pixel wide chip and debayer it, and you no longer have anywhere near 4000 pixels of real information. The usual response to this is that the company concerned is using proprietary algorithms to offset this problem, these algorithms being conveniently commercially sensitive so that they cannot be objectively assessed. In any case, this sort of thinking all amounts to a claim that you're getting something for nothing, and should be dismissed as such.

 

Call me a pedant, but it's really only a matter of defending terms so that they don't gradually lose their meaning; I don't want to see technical excellence diluted by someone's sales team. In the case of Red, particularly, the pitch was widely viewed as being unnecessarily hyperbolic and had therefore set itself up for exactly this sort of criticism.

 

P

Link to comment
Share on other sites

  • Premium Member
My point is though, 3 chip CCD cameras do well in tungsten balance. My friend and mentor owns an HPX2700 with AVC I board and just shot a national commercial on it. He chose to just tungsten balance in camera, rather than lose two stops filtering blue, because he didn't a difference in blue shadow and midtone noise. He was watching HD SDI out on a panasonic monitor. The RED one on the other hand has a very hard time with tungsten balance for some reason..

 

 

Hi,

 

Bear in mind the HPX2700 has a built in filter wheel, unlike the HVX 200.

Are you saying your friend white balanced using tungsten light with the daylight filter selected? If he did I very much he could not observe the difference, you must only look at the blue channel, it will be very very obvious.

 

The Red one is not an ENG camera & does not have a built in filter wheel, which is why the problem was more obvious.

 

Stephen

 

Edit, if you just TURN OFF the detail & filter white balance on your HVX 200, the pictures will be far less noisy, however you may then understand why I am not as big a fan of an HVX 200 as you.

Link to comment
Share on other sites

  • Premium Member
I'm not a fan of any particular camera i'm just fair.

 

 

So, when using the HPX2700 filter wheel, are you losing stops of light like using an 85 filter? are they the same thing?

 

Hi,

 

Your balancing the 3 color channels, I don't know the exact native balance of the HPX, so I won't speculate on the exact filter combination. An 85 filter is only 2/3 stop correction. Boosting gain in 1 channel is not a good alternative.

 

Stephen

Link to comment
Share on other sites

I was talking to Jan about it on the DVXUSER boards and she said the HPX2700 as well as all their three chip cameras to comparitively equal in tungsten light and daylight, whereas CMOS might have the potential to have more of an issue with tungsten. I'm still trying to find the article

Link to comment
Share on other sites

  • Premium Member
I was talking to Jan about it on the DVXUSER boards and she said the HPX2700 as well as all their three chip cameras to comparitively equal in tungsten light and daylight, whereas CMOS might have the potential to have more of an issue with tungsten. I'm still trying to find the article

 

Hi,

 

Funny thing is that the Viper in filmstream is also partial to daylight, I think you will find it's a silicon thing. I would also take what Jan says with a pinch of salt. The HPX2700 is comparatively equal because of a filter wheel to balance the sensor, the HVX 200 does not have one.

 

Stephen

Link to comment
Share on other sites

Hi,

 

Funny thing is that the Viper in filmstream is also partial to daylight, I think you will find it's a silicon thing. I would also take what Jan says with a pinch of salt. The HPX2700 is comparatively equal because of a filter wheel to balance the sensor, the HVX 200 does not have one.

 

Stephen

 

Ah this is very interesting - so to avoid extra noise caused in a channel by white-balancing, use traditional filtration?

 

So the Red where the chip is best under daylight, either shoot with daylight balanced light (like I believe David Mullen described when shooting Manure) or filter using an 80A filter, which will obviously lose you 2 stops. Correct?

 

Same with Sony's EX cameras, keep the white-balance high at daylight temperatures and filter for tungsten light - presumably you would be able to tweak the white balance figure to get exactly what you want without to much additional noise?

 

Then with ENG 3-chip CCD's the optimum is under tungsten light, and under daylight use the internal 85B on the filter wheel?

 

So presumably this gives a camera native to tungsten a big advantage, as you only lose 2/3s of stop from the 85B rather than 2 stops from 80A?

 

This is fascinating.

 

Cheers,

Andy

Link to comment
Share on other sites

  • Premium Member
so is the RAW image from a bayer filtered single chip like the Red useable without interpolation? is it true the Red isnt true 4k because of interpolation? Sorry for all the questions

 

Raw Bayer data has to be de-Bayered to work with the pixel-grid based equipment and software that is in universal use today.

 

Theoretically, we could build a Bayer masked display, say a monochrome LCD with the same color mask and site layout as the chip, and just let the viewer's eye/brain combination do the de-Bayering. Such a thing might be reasonable as a viewfinder on a Bayer camera. But while an end-to-end Bayer-data based system would work, the economic reality is that nobody's going to build this alternative universe. The voice of God never spoke from on high and said "Thou shalt use a rectilinear array of co-located tri-color data sets, and thou shalt call their name to be pixels." It's just that that way of organizing picture data has become a market-generated standard. It is perhaps unique in the history of standards in that everybody really does use it, and no committee had to document it first. ;-)

 

Red has 4K photosites or more across its chip. But 4K worth of one-color-per-site data isn't going to give you as much resolution as 4K worth of three-color-per-site data. How much less is the subject of dreary, contentious, and inconclusive debate.

 

 

 

-- J.S.

Link to comment
Share on other sites

Whatever Jan says it whatever Jan says. Of course I take it with a grain of salt, she's a sales rep. But a wise and intelligent one at that.

 

 

Anyways,

We are making the mistake of assuming here that all or most silicone sensors are daylight balanced, which is not true. The RED happens to have a noticeable preference for daylight. It looks like manure in tungsten light (no pun intended). But I 1rst AC'd on an hpx3000 commercial shoot where the DP and I, nore anyone else in the camera department, saw a major difference in noise between tungsten balance and daylight. We were watching the SDI out on a 27 inch panny broadcast monitor, so I assume it's very accurate. We were doing IN CAMERA balancing too. So he ended up using his Moles on the shoot. (We only had two HMIs)

I HAVE heard DP's complain alot about the image coming out of the RED camera when balanced for tungsten though, even right on the set, viewing the monitor. But David Mullen, who I respect and admire, says he doesnt think the tungsten noise on the RED is the end of the world. I believe he said somewhere that he's used tungsten balance on the camera before, and was still impressed by the results. It was on Reduser I believe, don't quote me on it word for word

So take all this with a grain of salt, in my opinion. I wish Panny and Sony would publish more often the native balance and MTF curves of their sensors though.

Link to comment
Share on other sites

  • Premium Member

Read Art Adam's columns here:

http://provideocoalition.com/index.php/aad...reen_screen/P2/

http://provideocoalition.com/index.php/aad..._on_the_f35/P1/

 

My experience was that at 320 ASA, the noise in the blue channel is fine at 3200K on the RED. There is some noise, but it would only stand out if you tried pushing the image to be more blue-ish or had accidentally underexposed more.

 

The pale 80D filter would be a good compromise (see Adam's test) if you were worried about it, but for normal dialogue scenes, it's OK to shoot in 3200K at 320 ASA with the RED without a blue filter if you aren't underexposing.

Link to comment
Share on other sites

  • Premium Member

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
×
×
  • Create New...