CMOS S/N deteriorating with time?
 

Those reading my posts here and on the nanoFlash forum may think I've become paranoid with this issue, but it's my impression that the particular, fine noise in my EX1 is increasing with time...

I have just compared some clips, recorded at the same lighting, scene and PP - tow years ago and today, and I can bet the noise level is higher in the current recordings!

My knowledge on CMOS technology is very limited, so if anyone out there have noticed the same, and know how to explain it - I'm all ears.

The S/N may decrease over time and usage, and it may be irreversible - but perhaps some calibration could bring it back to what it was when new? Interestingly, the Black Calibration is greyed out (inactive) in my EX1 menu...

You probably use your camera more than I do Piotr so I haven't noticed this problem. However, is there perhaps a filter on the front of the camera that you are using now that maybe you weren't using back then?

John

Thanks John, but no - it's not a filter...

As I said, I compared takes almost identical in every aspect.

can you upload two examplepictures in full-res?
i got paranoid too :D

Frankly? No, I will not post any pics. I've done my fair share of them in the past - the only outcome being people commenting how poor examples they were - JPEGs instead of PNGs, or that one cannot judge video using a still frame, blah blah... Finally, it all boiled down to the fact they were using 20" displays so they just couldn't see what I see on my 50-incher - but I've had enough :)

So, just an open subject for discussion: has anyone observed a noise increase with time on their CMOS-based cameras, and if so - can it be measured, or is it only subjective?

'Cause you know - once you become obsessed about something, it can become reality...

I don't recall seeing anything when I got the EX3 a year ago, but back in June when I was thinking about buying the Nano Flash, I started noticing mosquito noise in really busy shots, like trees and bushes that were rustling in the wind. I don't know if it was there before (I've got other things to do than go digging through old footage), but I notice it now.

No Charles - I don't mean the typical mosquito noise, if by that you understand the compression artifacts around contrasty edges / details. This is not a problem with the EX1, and even less so with the nanoFlash, when using higher bit rates.

Neither do I mean the typical low-light noise in the form of those largish "blobs", mainly present in shadows in most cameras of not-so-great sensitivity.

What I do mean is this typical for EX cameras, very fine & short, busy little "needles" that change their pattern from frame to frame, thus causing the shimmering - easily visible on larger displays. It tends to be present across the entire luma range, and is not restricted to just shadow and dark areas (though it's most apparent on those mid-bright, large areas of uniform color)...

I hope you know by now what I mean, in spite of my limited English :)

Hello Piotr,
I know that you are not joking with this. It has been a very long time since you started the thread at CD NF. But now you are observing the same thing directly from your camera and the logical question here is When was the last time you've serve your camera? It could give you a mind rest if nothing else...
Cheers

Good point, Luben!

Believe me, I keep my rig in the best condition imaginable. Nevertheless, I have purchased a 1-year extension of my Sony Prime Support just before my camera turned 2 years old, and I'm going to send it in for an overhaul before it expires (which is in the coming January) - but before I do, just wanted to draw fellow users' attention to the noise issue.

I did mention I'm becoming paranoid, didn't I :)

PS Just for the records: my thread on CD nF Luben mentions is about the nanoFlash actually augmenting this noise, especially in high bitrate, Long-GoP modes.

You could just ignore the troll posters and show others who are more interested in seeing the problem and hopefully offering their findings or solutions.

I haven't noticed anything going downhill with my footage and I use my camera a lot. Maybe the "Rose Tint" for your camera has worn off. It's funny I never noticed a dead pixel on my Nikon D90, now every shot I take I look for it. It has always been there, but I never looked for it when I first got the camera.

Hope you solve the problem

I haven't noticed any issues with my EX3, and I've had it since 2007. I have started to become very picky over what is happening to the picture though through compression. I have noticed a lot of compression related issues with some shots. For example a rough river in harsh sunlight with a lot of trees really does kick the stuffing out of the codec. Keeping my eye on the Ki Pro Mini.

hey, I just had the chance to compare my ex1 ( about 400 hours in use) with a brand new one. i made a lot of test-videos to compare noise levels: i could not see any difference, even in extreme settings or extreme colorgrading.
so: No, the chip is not deteriorationg over time.

The only type of sensor noise that fits that description is photon shot noise. The only sensor-based explanation for a change in photon shot noise is that the quantum efficiency reduced over the span of two years and conversion gain was increased to keep output levels the same.

While that might be theoretically possible, the likelihood is so remote that I'm certain the cause of what you're seeing lies elsewhere.

There are so many non-sensor (or external) factors that affect noise and its perceptibility that the most likely explanation is a gradual or marked change in one of those -- not the sensor itself. Here is a brief sampling of the possible factors:

* Sharpening

Sharpening increases noise visibility. Did you ever change your sharpening settings (in camera or in post) over the last two years?

* Saturation

Increased saturation can also increase noise visibility.

* White balance setting

Have you changed any of the methods you use in selecting white balance over the last few years? This can have a tremendous impact on noise visibility. For example, if you shoot in Tungsten light and allow a very reddish color cast, the noise wont be that bad. But if you use a typical white balance (that removes the color casts), it will often have more noise.

* Light

This isn't about absolute light levels (lux) -- obviously low light has more noise than ample light. But what I'm talking about here is how different types of light have different noise levels because of white balance. Magenta light has the last noise of all. Blue-defficient light such as Tungsten can be a lot noisier.

* Subject

Even your subject matter can greatly affect the perception of noise. For example, a highly detailed subject with lots of microcontrast does not look as noisy as a clear blue sky.

* Optical sharpness

Because of the preceeding factor, anything that reduces sharpness can increase the perception of noise. If the lens was bumped and no longer performs as well as it did brand new (e.g. decentration issue), then the lack of fine detail may increase the perception of noise.

Those are some of the factors that could be causing the issue you see. Hope that helps.

@ Thomas:

from a scientific point of view your test only reveals there is no noticable deterioration after 400 hours :-p
(When shooting a lot 400 hours can be reached within 10 weeks: comparing with 2000 hours would be more interesting.)

The "almost identical" part is a killer. There are so many variables on a camera like EX1. Are you using the identical profile with identical parameters? Can you reset the camera back to the factory defaults?

Thank you guys for the valuable input.

Unfortunately, what Daniel described as the "photon shot noise" looks like it is the case with my camera, as I have tried hard to eliminate all other, obvious factors (also mentioned by Daniel, and in the others' posts).

Of course, I cannot be 100% sure I didn't miss something, but I think that with some 99% certainty, I can see more sensor noise than before.

One thing that may have some relevance here is the Black Balance (or whatever it's called in the menu - in my case of 1.11 firmware, it's grayed-out anyway). I've heard there is no need to execute it manually, as the camera is supposed to do it automatically - who knows, perhaps my EX1 doesn't?

Luckily enough, I extended my Prime Support for an extra year, so will be trying to draw Sony's attention to it...

The other thing that hasn't been mentioned is how are you viewing your shots? Have you changed your monitor or have you created a new profile for it recently (a new monitor profile should be created at least once or twice a month).

I too have noticed some noise creeping in, but then I have also been messing about with the EX3 Detail settings. On a recent shoot I noticed some of the shots were soft - but I think this could have been my cameraman (woman actually) didn't hit the mark.

ps. what Detail settings are other users using?

you might want to check the back focus

Thanks Tom, it was actually the first thing that I did.

The only thing that was different is that I had detail turned OFF, I have now turned it back ON and set it to +4. still can't explain why the shots looked soft on that shoot but they look OK now.

Maybe I should sack the camerawoman - but then I must show some compasion towards her, after all she is Mrs Oliver

I was noticing some soft shots on my original EX1 and as others have suggested did the usual back focus etc. This made no difference. I checked the shots through the zoom range and there was definitely a discrepancy in the backfocus. No amount of back focussing would make the problem go away and my pictures were all just a little soft. However updating the firmware and the hard reset that this performs restored my backfocus to normality and my pictures were pin sharp again.

I have a very old EX1 (1200 hours) and I recently got a new EX1R, I can't see any difference in noise between the two.

Thanks for your advice Alister, I haven't upgraded the firmware on my EX3 (yet) I am in the middle of a production and I know from past experience - don't touch anything until the job is in the can.

At the moment the focus seems to be OK, I am just going to blame Mrs O for the time being, when I discover it is the camera they she will get a bunch of roses :-}

As far as I know the actual photodetectors in CMOS and CCD cameras are the same. I've neither read or seen anything to indicate that photodiodes gradually decay in efficiency.
The noise in most things electronic is a function of temperature, seasonal variations in ambient or even how long the camera was on could account for variations. I should also ask about any filters in front of the lens.

Maybe you are paying so much attention to noise that you see it in everything? How about just starting with the basics and try shooting on SxS, using the simple Vortex Media PP and see what you get.

Many of us are shooting with SxS cards, although I do not see what difference that should make. As good as Doug Jensen profile is, it is not the holy grail of profiles.

I think Bob Grant may have hit the nail on the head with his post.

Ha Ha Ha . I love it.

John

My point Vincent, is that the Nano records detail that does not appear on SxS due to the fact that certain settings in the PP are bypassed (I think) and noise is exposed on Nano recordings. Jensen's PP is not a very extreme change from the factory EX1 settings. I don't get very much noise at all shooting on SxS with the Jensen PP, so if the OP thinks his sensor is going bad and causing noise, he could go back to SxS and a general purpose PP to see if that looks noisy. If it does not, then there is something about his settings that is causing the noise. Either that or he is obsessed with noise (he has been talking about it for months) and sees it all the time now.

None of the PP settings are bypassed when recording the SDI output, Chad. DSP acts between the imagers and SDI out (or own compression to the SxS). Of course this is a simplification, but yo get the idea ...

Good to know!

Perhaps the difference is caused by difference in compression?
More compression = less info.
A good way to decrease data is a kind of 'blurring/smoothing' which makes almost similair pixels more silimair (this can be done with pixels next to each other, but also with the same pixel over time (from frame to frame). Noise is a distinct difference between pixels (both in location and over time) that 'should' be the 'same' and is caused by the physical properties of the 'lightcapturing-proces'. (Remember: celluloid has grain, which is also the result of that light-capturing-proces.)
= > bypassing compression reveals the noise that was already there but got 'compressed away'.

(If the PP got bypassed in the SDI-signal everything would look like factory-settings)

Well I'm just wondering if Nano footage looks more grainy as a rule, or under certain conditions, or is it a non issue. I don't want to spend 4 grand getting set up with Nano, only to have to put an unsharp mask filter on all my footage.

Put simply compression = low pass filter.
I think I'm also correct in saying that chroma subsampling is also compression and will also reduce noise, slightly. What the SDI signal does bypass is the mpeg-2 compression, it also has better chroma sampling.
I think the best example of the impact of compression on noise is found in the cheap AVCHD / H.264 palmcorders. Despite their very small sensor and slow lens in low light the image is pretty much noise free. It's also pretty much unwatchable due to the horrid macroblocking.

While I tend to generally agree, if you care to see the last couple of posts in the "noise" thread on the nanoFlash forum it seems that CD recognizes their Long-GoP structure is not necessarily best optimized yet, resulting in the higher datarates (like 140, and particularly 180 Mbps) producing frames of severely fluctuating quality (noise content) inside any given GoP. It can easily be seen when stepping though the GoP inside an NLE - the I, P, and B frames differ considerably; much more so than with 35 or 50 Mbps. This tends to augment the perception of shimmering noise in the actual video, due to the fast fluctuation.The "sweet spot" of L-GoP nanoFlash format, 100 Mbps, seems to be the best compromise in this regard, as well.

The consensus is that for the uncompromising quality, one needs to record in I-Frame only at 220 Mbps or above.

But I guess we're getting slightly off-topic in this thread :)

Piotr

Thanks Alister for chiming in.

So, back to the thread's topic:

- could you enlighten us about the Black Balance adjustment on the EX1? Is it done automatically indeed, or does it make sense to run it manually? Its very presence in the Maintenance Menu would suggest the latter - but then again, it's inactive in my EX1 (greyed-out)...

- while we're at the Maintenance Menu: when you run the Back Focus adjustment, do you repeat it for all 3 ND positions as was touted in the EX1 early days, or is a single execution enough?

Thanks,

Piotr

I believe the EX1 does a black balance every time you turn it on, but I cannot confirm this.