Testing my GH2's sensor

Started Feb 20, 2011 | Discussions
bg2b
Contributing MemberPosts: 897Gear list
Like?
Testing my GH2's sensor
Feb 20, 2011

For amusement, I've been doing some tests on my GH2's sensor; here are a few random observations.

"Real" ISOs are probably just 160, 320, and 640. 200, 400, and 800 show odd artifacts (gaps or near gaps) in the RAW histograms. 250, 500, and 1000 show them even more. Everything beyond ISO 1000 is completely artificial; they're obviously just digital scalings of lower ISOs. So if you're a RAW-only type, there's no point in going past ISO 640 (or maybe 800, which is probably the least artificial-looking of the x00's). All you're doing is just costing yourself highlight headroom without gaining any signal-to-noise ratio (SNR) in the shadows.

I derived the SNR curves using the procedures described in these references:

http://www.clarkvision.com/imagedetail/evaluation-1d2/ (seems to be offline at the moment)
http://theory.uchicago.edu/~ejm/pix/20d/tests/noise/index.html

Here are the characteristics of ISOs 160, 320, and 640:

ISO 160
Well capacity 12350 electrons
3.1 electrons/RAW unit (electrons = detected photons)
Model read noise 6.5 electrons

(The "model read noise" is from a numeric fit of the measured SNR curve to a simple "total noise = photon shot noise plus constant read noise" model)
Per-pixel "useful" dynamic range (DR for SNR 12dB) 8.5 stops
Per-pixel DXO-like dynamic range (DR for SNR 0dB) 10.8 stops

Full SNR curve (The colors indicate samples from different pixel colors; there are two greens, green1 and green2. The horizontal line is at 12dB, and the diagonal line is the SNR limit due solely to photon shot noise.):

ISO 320
Well capacity 6200 electrons
1.6 electrons/RAW unit
Model read noise 4.0 electrons
Per-pixel useful dynamic range 7.8 stops
Per-pixel DXO-like dynamic range 10.4 stops
Full SNR curve:

ISO 640
Well capacity 3250 electrons
0.8 electrons/unit
Model read noise 3.5 electrons
Per-pixel useful dynamic range 7.0 stops
Per-pixel DXO-like dynamic range 9.7 stops
Full SNR curve:

I didn't measure the true saturation ISOs, but according to DXO, ISO 160 corresponds to a saturation ISO of 167, and going by the well capacities for 320 and 640, I think they must also be labeled accurately.

If you want to compare to DXO's "print" graphs, you need to add 0.5 stop to the DR numbers.

The camera's read noise performance seems actually pretty good--not state-of-the-art at base ISO, but certainly no slouch. Where it's lacking somewhat is in full-well capacity. I'd guess the limit is actually the 12-bit converter. A 14-bit converter with the 12K electrons from ISO 160 and the same read noise as ISO 640 would give excellent DR at high ISO. Maybe for the GH3...

The unity gain ISO is approximately ISO 640, with just below one electron per RAW unit. It would be ISO 500, but that's not a real ISO. Probably 640 is the best one to use when you're light-limited and you're not worried about highlight clipping, except maybe you have to consider thermal noise differences among the different ISOs. I haven't tested thermal noise performance.

ulfie
Veteran MemberPosts: 3,224Gear list
Like?
Re: Testing my GH2's sensor
In reply to bg2b, Feb 20, 2011

Heck, most of us don't even know what the heck thermal noise performance is! But we do know when we like an image.

Reply   Reply with quote   Complain
cameron2
Senior MemberPosts: 1,124Gear list
Like?
Re: Testing my GH2's sensor
In reply to bg2b, Feb 21, 2011

So what do you set the ISO at when it's really dark and/or you need to freeze motion? i.e. if ISO 3200 is called for, do you still shoot at 640 and just fix it in PP?

 cameron2's gear list:cameron2's gear list
Sony RX100 Panasonic Lumix DMC-GH2 Panasonic Lumix DMC-GF3 Panasonic Lumix DMC-GX7 Panasonic Lumix G 14mm F2.5 ASPH +16 more
Reply   Reply with quote   Complain
kenw
Veteran MemberPosts: 4,486Gear list
Like?
Thanks!
In reply to bg2b, Feb 21, 2011

These kinds of measurements never really help my photography, but as an engineer I find them entertaining. Not surprisingly your data just about perfectly matches the DxO data.

The "fake" ISO observations are useful. Thanks for that, I was curious about that specific issue and hadn't taken the time to investigate myself.

The other thing looking at your data made me realize (should have noticed this before, for some reason your presentation just made it click in my head) is that the DxO practice of measuring DR from 0dB tends to overly penalize a camera for read noise. As an example, using the photographically useless 0dB reference gives the K5 (which has incredibly good read noise) a 3eV advantage over the GH2 while using the more appropriate 12dB point reduces the advantage down to 2eV.
--
Ken W

Rebel XT, XTi, Pany G1, LX3, FZ28, Fuji F30, and a lot of 35mm and 4x5 sitting in the closet...

 kenw's gear list:kenw's gear list
Sony RX100 Panasonic Lumix DMC-G1 Olympus OM-D E-M5 Panasonic Lumix DMC-GM1 Panasonic Lumix G Vario 7-14mm F4 ASPH +29 more
Reply   Reply with quote   Complain
jeffharris
Veteran MemberPosts: 4,800Gear list
Like?
Whatever floats your boat...
In reply to bg2b, Feb 21, 2011

bg2b wrote:

For amusement, I've been doing some tests on my GH2's sensor...

What difference does any of that make?
Wouldn't you rather just use your camera for it's intended purpose?

 jeffharris's gear list:jeffharris's gear list
Panasonic Lumix DMC-GX7 Panasonic Lumix G Vario 7-14mm F4 ASPH Voigtlander Nokton 25mm F0.95 Voigtlander Nokton 17.5mm F0.95 Aspherical Voigtlander Nokton 42.5mm F0.95 +25 more
Reply   Reply with quote   Complain
bg2b
Contributing MemberPosts: 897Gear list
Like?
Re: Testing my GH2's sensor
In reply to cameron2, Feb 21, 2011

Theoretically, yes, you underexpose and then push it during the RAW conversion. Of course practically, the camera may be easier to operate if you ignore all this. And if you're not in danger of blowing highlights, it doesn't really matter anyway; the camera can multiply just as well as the RAW converter can.

Reply   Reply with quote   Complain
bg2b
Contributing MemberPosts: 897Gear list
Like?
Re: Thanks!
In reply to kenw, Feb 21, 2011

kenw wrote:

Not surprisingly your data just about perfectly matches the DxO data.

FWIW, here's the print numbers compared to DXO's measurement of the GH2 and GH1. Sorry about the ISO labels, but the points are for ISOs 100, 160, 200, 320, 400, 640, and 800. (Note that for the GH1, they're the measured saturation ISOs, which differ a fair amount from the indicate ISOs.) Of course, I'm sure they're much more careful about things than I am, but I get the impression that they may have penalized the GH2 a little by using ISO 200, 400, and 800 instead of 320 and 640.

Reply   Reply with quote   Complain
thinkfat
Senior MemberPosts: 1,184Gear list
Like?
well capacity
In reply to bg2b, Feb 21, 2011

I'm just wondering, why would the well capacity change for different ISO? Is it not a physical parameter of the sensor? At anything but base ISO, it should not be the sensor that saturates but the AD converter.

-- hide signature --

Everybody loves gadgets, until they try to make them
http://www.flickr.com/photos/thinkfat
http://thinkfat.blogspot.com

 thinkfat's gear list:thinkfat's gear list
Panasonic Lumix DMC-GH1 Canon EOS 5D Mark II Panasonic Lumix G Vario 14-45mm F3.5-5.6 ASPH OIS Panasonic Lumix G 20mm F1.7 ASPH Panasonic Lumix G Vario 45-200mm F4-5.6 OIS +10 more
Reply   Reply with quote   Complain
Kikl
Contributing MemberPosts: 986Gear list
Like?
Re: Testing my GH2's sensor
In reply to bg2b, Feb 21, 2011

bg2b wrote:

For amusement, I've been doing some tests on my GH2's sensor; here are a few random observations.

Well thanks for posting, but what does "Stops down from saturation" mean? You measured the signal to noise ratio at different exposures of the sensor? I am only guessing....

Reply   Reply with quote   Complain
bg2b
Contributing MemberPosts: 897Gear list
Like?
Re: well capacity
In reply to thinkfat, Feb 21, 2011

It's not a physical number of electrons that the well can hold; it just means that numbers above that will saturate the analog-to-digital converter. If they could add another bit to the A-to-D converter while keeping the gain and read noise the same, they'd add another stop to the dynamic range.

Reply   Reply with quote   Complain
bg2b
Contributing MemberPosts: 897Gear list
Like?
Re: Testing my GH2's sensor
In reply to Kikl, Feb 21, 2011

You measured the signal to noise ratio at different exposures of the sensor? I am only guessing....

Yes, that's basically right.

Take the strongest light/longest exposure that leaves the sensor at just below saturation. Whatever that exposure is defines the true saturation ISO (that's the first thing DXO measures, and the one that leads to endless confusion and accusations of manufacturers cheating, etc.) But if you're working only in RAW, it's what you care about. That level is where your sensor provides the best signal-to-noise ratio, and is the basis for the whole expose-to-the-right business.

After you know the strongest signal, start reducing the exposure. The signal-to-noise ratio will start decreasing. Initially it will decrease by about 3 dB per stop; that's due just to photon shot noise. Once the exposure starts to get low enough, the SNR will start to drop faster because other noise sources (like the analog-to-digital converter noise) will start to make a significant contribution. (If there were no other noise sources, the SNR plot would follow the dashed diagonal line on the graph all the way down.)

That's how the graphs are made; start with the largest non-saturating exposure and then reduce the exposure and plot the SNR at different points.

Reply   Reply with quote   Complain
2014imaging
Forum MemberPosts: 85
Like?
Re: Testing my GH2's sensor BECAUSE? I have entirely too much time on my hands..
In reply to bg2b, Feb 21, 2011

...and, I don't actually consider myself a photographer...more of an engineer.

Reply   Reply with quote   Complain
Kikl
Contributing MemberPosts: 986Gear list
Like?
Re: Testing my GH2's sensor
In reply to bg2b, Feb 21, 2011

bg2b wrote:

You measured the signal to noise ratio at different exposures of the sensor? I am only guessing....

Yes, that's basically right.

Take the strongest light/longest exposure that leaves the sensor at just below saturation. Whatever that exposure is defines the true saturation ISO (that's the first thing DXO measures, and the one that leads to endless confusion and accusations of manufacturers cheating, etc.) But if you're working only in RAW, it's what you care about. That level is where your sensor provides the best signal-to-noise ratio, and is the basis for the whole expose-to-the-right business.

After you know the strongest signal, start reducing the exposure. The signal-to-noise ratio will start decreasing. Initially it will decrease by about 3 dB per stop; that's due just to photon shot noise. Once the exposure starts to get low enough, the SNR will start to drop faster because other noise sources (like the analog-to-digital converter noise) will start to make a significant contribution. (If there were no other noise sources, the SNR plot would follow the dashed diagonal line on the graph all the way down.)

That's how the graphs are made; start with the largest non-saturating exposure and then reduce the exposure and plot the SNR at different points.

Thanks for the detailed explanation, that makes a whole lot of sense. So your measurements also show effectively the dynamic range (maximum SNR) of the sensor/electronics at different ISOs; 6dB being equal to 1 stop, I gather.

Now people continuously talk about dynamic range, but how much dynamic range does the sensor really need? That is how much dynamic range is perceptible to the human eye in a single scene/picture? Can this dynamic range be reproduced in printing or on a screen?

I wish there were more information about this readily available, because that's what really counts. What good is a dynamic range of 15 stops if this is imperceptible or not reproduceable?

Reply   Reply with quote   Complain
kenw
Veteran MemberPosts: 4,486Gear list
Like?
Re: Testing my GH2's sensor
In reply to Kikl, Feb 21, 2011

Kikl wrote:

Now people continuously talk about dynamic range, but how much dynamic range does the sensor really need?

There is eventually a limit set by the optics since all lenses have a "veiling flare" that will fill in the shadows. Figuring out exactly what this limit is can be a bit dicey, but reading through papers that have measured it and try to take those measurements into a more useful "photographic" domain it would seem that using the DxO DR measurement (based on 0dB) sensors with even 16eV of dynamic range would still be useful with prime lenses.

That is how much dynamic range is perceptible to the human eye in a single scene/picture?

Absolutely enormous amounts of dynamic range because of the way we see and scan a scene. Typically the human eye can handle about 17 stops of dynamic range in a single scene (and that would be roughly equivalent to 19 stops in DxO's measurements).

Can this dynamic range be reproduced in printing or on a screen?

No, but that doesn't matter. Typically a scene with high dynamic range will either be compressed into the print dynamic range through a combination of tone curve, burning/dodging, or local tone mapping. Only local tone mapping is a recent digital invention, tone curves and burn/dodge have of course been used in the darkroom for a century. It is always a bit dicey comparing different dynamic range measurements, but print films can have a dynamic range of over 16 stops and darkroom practitioners have been able to get all that DR onto a print in the past.

Personally, for me, I rarely found my self very "handicapped" by the "limited" DR of the G1. I just didn't take photos that required wide DR. Since I spent a lot of my film days shooting slide film (which has a strong tone curve and thus much lower DR than print film) I guess I'm just familiar with the "limitation".

-- hide signature --

Ken W

Rebel XT, XTi, Pany G1, LX3, FZ28, Fuji F30, and a lot of 35mm and 4x5 sitting in the closet...

 kenw's gear list:kenw's gear list
Sony RX100 Panasonic Lumix DMC-G1 Olympus OM-D E-M5 Panasonic Lumix DMC-GM1 Panasonic Lumix G Vario 7-14mm F4 ASPH +29 more
Reply   Reply with quote   Complain
mpgxsvcd
Veteran MemberPosts: 7,939Gear list
Like?
Re: Testing my GH2's sensor
In reply to cameron2, Feb 21, 2011

How do you fix motion blur from too slow of a shutter speed in post processing?

cameron2 wrote:

So what do you set the ISO at when it's really dark and/or you need to freeze motion? i.e. if ISO 3200 is called for, do you still shoot at 640 and just fix it in PP?

 mpgxsvcd's gear list:mpgxsvcd's gear list
Panasonic Lumix DMC-LX100 Panasonic Lumix DMC-GF1 Panasonic Lumix DMC-GH3 Panasonic Lumix DMC-GH4 Panasonic Leica Summilux DG 25mm F1.4 +2 more
Reply   Reply with quote   Complain
bg2b
Contributing MemberPosts: 897Gear list
Like?
Re: Testing my GH2's sensor
In reply to mpgxsvcd, Feb 21, 2011

You'd shoot at the same aperture and shutter speed, so there's no difference in depth-of-field or motion blur. The camera will say you're underexposing. If you're not using manual mode, you'll have to dial in negative exposure compensation to tell the camera to do that. Image review will also be less useful, since the camera will be showing you the underexposed shots. That's why I said that in practice it may be easier to just let the camera use the higher ISO. Certainly if you're not risking highlight clipping--then there's no downside to letting the camera do the multiplication.

But if you're backed into a corner and think you need f/2, 1/100, ISO 3200, and as much dynamic range as you can get, and if you have time to set everything up, then theoretically you're better off shooting at f/2, 1/100, ISO 800, and then pushing it two stops in the RAW converter. You'll get two more stops of DR for highlight recovery, while your shadows will be no noiser than if you had used ISO 3200 in-camera .

Reply   Reply with quote   Complain
exdeejjjaaaa
Veteran MemberPosts: 6,841Gear list
Like?
Re: Testing my GH2's sensor
In reply to bg2b, Feb 21, 2011

bg2b wrote:

Where it's lacking somewhat is in full-well capacity. I'd guess the limit is actually the 12-bit converter. A 14-bit converter with the 12K electrons from ISO 160 and the same read noise as ISO 640 would give excellent DR at high ISO. Maybe for the GH3...

http://www.luminous-landscape.com/forum/index.php?topic=51510.0

posts by Emil Martinec...

14bit is enough to count every single electron w/o noise in GH2 wells, minus at least 2 bits for noise... no need to 14 bits then in the output.

-- hide signature --

 exdeejjjaaaa's gear list:exdeejjjaaaa's gear list
Olympus E-M1 Sony Alpha 7 Panasonic Leica DG Macro-Elmarit 45mm F2.8 ASPH OIS Panasonic Leica Summilux DG 25mm F1.4 Panasonic Lumix G 20mm F1.7 ASPH +23 more
Reply   Reply with quote   Complain
453C
Senior MemberPosts: 4,622Gear list
Like?
Re: Testing my GH2's sensor
In reply to kenw, Feb 21, 2011

bg2b and kenw , thank you for the research and technical discussion.

 453C's gear list:453C's gear list
Olympus PEN E-PL1 Olympus E-M1 Olympus M.Zuiko Digital ED 9-18mm 1:4.0-5.6 Panasonic Lumix G 20mm F1.7 ASPH Olympus M.Zuiko Digital ED 14-42mm 1:3.5-5.6 +5 more
Reply   Reply with quote   Complain
bg2b
Contributing MemberPosts: 897Gear list
Like?
Re: Testing my GH2's sensor
In reply to exdeejjjaaaa, Feb 21, 2011

To put it another way, I think they're currently "wasting" a couple of bits for ISO 640 at the bottom, since the read noise seems to be something like 3-4 electrons and the gain is only about 1.

If the A-to-D converter at ISO 640 had a 4x higher gain and the same read noise, then certainly 12 bits would be fine for counting up to the ISO 160 well capacity.

Reply   Reply with quote   Complain
SterlingBjorndahl
Senior MemberPosts: 1,846Gear list
Like?
Much appreciated, thanks.
In reply to bg2b, Feb 21, 2011

bg2b wrote:

For amusement, I've been doing some tests on my GH2's sensor; here are a few random observations.

Thanks very much for this. I appreciate learning how things work "under the hood".

Sterling
--
Lens Grit

 SterlingBjorndahl's gear list:SterlingBjorndahl's gear list
Panasonic Lumix DMC-FZ50 Panasonic Lumix DMC-GH2 Olympus PEN E-PL1 Panasonic Lumix DMC-L1 Olympus E-1 +7 more
Reply   Reply with quote   Complain
Keyboard shortcuts:
FForum MMy threads