G10-G11-s90-GF1_73 noise test

[col]

My logical side is puzzled by your statements however.  While it is true the amount of available light of the sensor is limited by the aperture of the lens, the size of the sensor should have some correlation to the ability to gather light.  Each sensor site is covered primarily by non light gathering area ... I believe only 10% of a CMOS sensor site is actually light sensitive, whereas about 30% of a CCD sensor site is light sensitive.

If we increase the sensor size it increases the area of that light sensitive portion, and thus increases the ability of the sensor site to record light.  The increase in sensor site size doesn't increase noise in a proportionate amount, leaving me with the conclusion that a well designed sensor that is 5 times larger has a light gathering advantage vs. noise against a sensor that is substantially smaller. I assume there comes a point of diminishing return, but I believe the difference between G11 and the micro 4/3rds sensors is significant enough to help with improved noise.

You are absolutely correct. Larger sensors or, more precisely, sensors with lower pixel density, generally have a higher 'fill factor", meaning the light sensitive area is a higher proportion of the total area. This is a useful second order advantage of larger sensors. Another benefit of larger sensors is the superior dynamic range on account of the larger pixel full well capacity, though this advantage is only available when there is "light to burn" and the lowest ISO can be used.

However, the main reason why a 4/3 sensor equipped camera such as the GF1 outperforms a G11 for image noise in low light, is because the GF1 has a much larger lens (greater aperture area) to gather more light, consistent with all the previous discussion.

Let's see whether I'm talking BS. The G11 weighs 355gms including lens, and the camera (with lens) is 112x76x48mm. The lens of the GF1 is HUGE by comparison. The nearest matching GF1 lens in terms of zoom range is the Panasonic/Leica 14-50mm, and at 434gms, the lens alone weighs more than the entire G11! Dimensionally the lens is huge by comparison as well, being 74mm in diameter and 93mm long. Yep, everything I have said adds up in the real world ....

but I believe the difference between G11 and the micro 4/3rds sensors is significant enough to help with improved noise.

Certainly the difference between the G11 and 4/3 lenses is significant enough to help with improved noise.

[col]

May I ask you to compare the noise characteristics of the Leica X1 and the Canon G11 and then explain from where any difference arises?

Most certainly.

The X1 will whip the pants off a G11 for image noise in low light.

As per Wayne's last posting, a small part of the X1 low noise superiority can be attributed to the higher pixel fill factor of the larger sensor. However, most of the substantially lower noise of the X1 will be due to the larger aperture area of the X1 lens. Let's see if that prediction is true.

The X1 sports a prime lens of L=24mm (actual), and F2.8. Therefore the effective aperture diameter is D = L/f = 24/2.8 = 8.6mm
The "35mm equivalent" focal length is 36mm

The G11 sports a zoom lens of L=6.1-30.5mm (actual), L=28-140mm (equivalent) and F2.8-4.5 Therefore the effective aperture diamter ranges from 6.1/2.8 = 2.2mm at the wide end, to 30.5/4.5 = 6.8mm at full telephoto.
 At the same equivalent focal length (ie, FOV) as the X1, 36mm, which is very close to the wide end of the G11, we could roughly estimate the effective G11 aperture as about 2.5mm at 36mm effective focal length.

OK. At the same FOV, the X1 has an aperture diameter of 8.6mm, while the poor old G11 has only 2.5mm. In terms of aperture areas, the X1 is therefore 58 mm^2, while the G11 is only 4.9 mm^2  Therfore, the aperture area of the X1 is a whopping 12 times larger than the G11. Wow!!! I should damned well hope the X1 would give a lower noise image, with 12 times as much light being gathered by the lens and dumped on the sensor.

Once again, it is seen that a large aperture lens is required to collect a large amount of light, but there is one last question to be answered re the X1. The lens on the X1 is NOT particularly big or heavy, so how does it manage to have a much larger aperture diameter than the similar sized lens of the G11? Simple. The G11 is a fixed-focal-length lens whereas the G11 sports a x5 zoom range. If you can live without zoom, (I can't), then prime lenses really are MUCH smaller than zoom lenses, and this is the only "trick" to the X1. Relative to an SLR I guess Leica also save a bit of bulk and weight by making the lens non-removable, but this makes the fixed-focal-length lens even more limiting.

In the end, it get's down to this. Large sensors can produce wonderful results if you can throw enough light at them. To do this, you need a lens with a large aperture diameter, which means either a physically large zoom lens, or else use a prime and put up with no zoom. No exceptions, unfortunately.

Therefore the holy grail of a G11 style compact camera with useful zoom range, and a large APS sensor to supposedly give low noise as per an SLR, is simply not going to happen.

[paratom]

Lets compare the gf1+20/1.7 to the g11

1) One thing not to forget is DOF. While sometimes shallow DOF is nice for isolating the subject  - very often a certain amount of DOF is wanted to have everything sharp which should be sharp. I am talking about a group portrait, a mother with daughter, a portrait from an certain angle, macro, interieurs, etc etc.
To get similar DOF with a 4/3 and a small sensor one has to stop down the 4/3 lens at least 1-2 steps (I assume probably more).
From thispoint of few one could argue that the smaller the sensor the further can I open the f-stop and the lower can I keep the ISO.
2) The 20/1.7 as far as I know doesnt have IS while the g11 has it
3) the g11 has a zoom, the 20 is a prime
conclusion from 1) and 2) the low light advantage of a gf1 over a g11 is mainly valid for moving subjects and shallow DOF images. Otherwise one could use the g11 at least at 2 steps lower ISO because of the larger DOF and the IS

For me the idea of a 20mm prime was to restricting. I found that out with the dp2 allready which also has a 40mm FOV.
If I had to live with one focal length it would need to be 35mm like the x1 for example, but still I really like the flexibility of the zoom in the g11.
Sometimes 28mm wide angle is just really a nice to have, and sometimes its also nice to be able to zoom to 100 or 140 mm.

If I had to use this camera as my 1 and only I would prefer a 4/3 camera (probably the GH1) and use it sometimes with the prime and sometimes with a zoom.
But as an additional small camera I much prefer the g11-style of camera, even if the cost was the same as a 4/3 camera. Another nice thing is 1/2000 flash -sync.

[TheSuede]

Wayne - sorry for hijacking your thread, but I do think the discussion is relevant to the theory, just as well as your comparison was beautifully coherent to the practical side of it.

Col:
Then I have nothing to object to about the your conclusions, except for the way you word it. Let me try to explain why...:
The physics of it is correct, but it is a very confusing way to try to explain it (as photographers are mostly F/no "thinking"). In the middle ranges of focal lengths, size of the optics doesn't really increase that much when you increase sensor size and want to keep the available field of view fairly constant - IF you keep aperture area constant. But to get an increased light amount to hit the sensor (which was why you wanted a larger sensor) you have to increase aperture area - which you do by keeping F/no constant! Much of the "normal" confusion lies here.

Small sensor, small focal length at F/4 > X amount of light per second.
2x sensor, 2x focal length, still at F/4 > 4X amount of light per second. Same FoV, Half DoF, half noise.
- And the logical explanation for the above: You keep field of view the same, but the aperture area increases with 4x. This system will be bigger, and heavier. As compared to:

Small sensor, small focal length at F/4 > X amount of light per second.
2x sensor, 2x focal length, at F/8 > Still X amount of light per second. Same FoV, same DoF, same noise - as the aperture area/FoV is kept constant.

In this case the lens doesn't HAVE to get that much larger, but it will most oftenly get a bit longer physically. My 50F/1.4 isn't that much smaller than my 100F/2.8. In fact, it's heavier, and has a larger outside diameter.
So I get what you're saying, and it's ok in my book. You can't just pop a 24x16mm sensor in a compact camera and expect it to "work" as in "gathering more light". But on the other hand I'm all for a slightly smaller zoom-range, a slightly larger sensor and keeping size increase to a minimum... And you can make small compact lenses wickedly sharp ("sharp" used in the sentence where "good" maybe had been more in line with what I was thinking, but more commonly misunderstood) these days - !
Most people do no direct connection between aperture area and the F/no, and that's the confusion.

The 4/3 system is still hampered by it's ancestral heritage - the systems could made even smaller. Their longer, large aperture zooms are in fact 36x24mm optimized zooms with a slightly smaller diameter - with a wide-converter built in to the rear end of the lenses to adapt the lens to the 4/3 size sensor... (WC is the opposite to TC's, and yes, they do exist. Very abundantly in the astro-photography community in fact...). They could be at least 40% smaller if they were developed from "scratch", or just downscaled originals - which I have NO idea why they didn't do from the start. It probably (as usual) has something to do with developing- and production run costs...

But I'm very much a "prime" person, I rarely find uses for superzooms or even anything above the normal 3x zoom lenses. Larger sensors in compacts with smaller zoom ratios would fit me perfectly - but I'm aware that those products may be hard to sell in quantities enough to be lucrative to the maker.

[JeffKohn]

(Amount of light striking sensor) = (Field of view) x (Aperture Area)

Note that the size of the sensor does not come into it, exploding the myth that the sensor size is directly responsible for collecting more light.

The problem with your argument is that you're assuming FOV is held constant for a lens. If the lens projects a large enough image circle to cover both sensors, then the FOV will not be constant, and the larger sensor will collect more light. Your equation also leaves out one rather important variable: exposure time.  If two sensors used different lenses that had the same aperture area for the same FOV, that would imply that the larger sensor was using a longer lens at a smaller f/stop. In that case, the larger sensor would be using a longer exposure time, so it would still collect more light.

While nothing I have said to date is incorrect, it turns out that a larger sensor permits the use of larger lenses

The larger sensor doesn't just 'permit' a larger lens, it requires a larger lens. It's not like you could stick a 4/3 sensor behind the G11 lens, the image circle is far too small. Sensor size and aperture area are linked.  So what you say about aperture area is technically correct, but amounts to semantic quibbling that is essentially meaningless regarding the relationship between sensor sizes and image quality.

It's true that using the G11's lens with a larger sensor wouldn't make much sense, but I don't think that's what Wayne was suggesting.

[col]

In previous postings, I wrote:

... the sensor size does not, fundamentally, have anything to do with how much light is collected by the sensor, despite popular myth ....
However, the sensor size is an indirect player in how much light can be gathered onto the sensor with physically realizable lenses ....

While nothing I have said to date is incorrect, it turns out that a larger sensor permits the use of larger (aperture) lenses, which therefore gather more light onto the sensor and decrease image noise, which is a very good reason why SLR and professional camera use larger sensors. As yet I have not told the whole story ...

It's time I fessed up and told the whole story. The commonly given advice, is that if you care about image quality in general, and low image noise under poor lighting conditions in particular, then buy a camera with a large sensor, such as 4/3, APS, or even full-frame. That advice is correct.

However, it is also correct that the amount of light falling on the sensor is set by the physical size of the lens (aperture area), not by the sensor size, so how can these two apparently contradictory facts be reconciled?

A hypothetical example is the easiest way to explain. Consider a camera such as the Canon G11, with a smallish sensor, and a physically small lens. A previous posting showed just how small the absolute aperture actually is on a G11, from 2.2 to 6.7mm. Therefore we immediately know that the low light performance of this camera won't be good, because the absolute aperture of the compact lens is too small to collect much light, irrespective of the size of the sensor. Without altering the zoom range of the camera, the only way to improve matters and get more light on the sensor is to increase the aperture diameter of the lens.  However, there are two ways to do this.

Method (1) is to fit a 'faster" lens, with lower f-number, and therefore larger aperture diameter. The G11 has an F2.8-4.5 lens, and it would certainly be possible in principle for Canon to fit this camera instead with an F1.4-2.2 lens. Yep, that would double the aperture diameter, and gather 4 times as much light. However, if we want even more light, and therefore an even larger aperture, we have a problem, because it becomes very difficult and expensive in practice to build lenses with an f-number less than around F1.4. What can be done? How can we increase the absolute aperture still further, without the need for impractically small f-numbers?

The solution, method (2), is to scale up the absolute aperture and the focal length in unison, leaving the f-number unchanged. Whoa, bingo, that's a great idea, because we can now increase the all-important aperture diameter almost without limit, without the need for impractically small f-numbers, but there is a pesky problem, because the increased focal length results in a reduced field of view. Fortunately the FOV can be maintained by scaling up the sensor, indeed, for this reason the sensor MUST be scaled up whether we want to or not!  

Arguably then, the greatest advantage of a large sensor is that it allows the use of large aperture lenses without the need for impractically small f-numbers, which is why high-quality cameras, with large lenses to gather lots of light, will (must) always use a large sensor.

It is worth understanding that the FUNDAMENTAL requirement for gathering lots of light is a large aperture lens, and as a CONSEQUENCE of this, a large area sensor must be employed, if impractically small f-numbers are to be avoided. The converse is not true.

To summarize: (low image noise) needs (lots of light on sensor) needs (large lens) needs (large sensor, to avoid impratically small f-numer).

I realize that for many this will be a new and foreign way of looking at things, but in reality it is simple, factually correct, and very empowering. I also know that myths and old ways of looking at things die hard ....

[col]

The problem with your argument is that you're assuming FOV is held constant for a lens. If the lens projects a large enough image circle to cover both sensors, then the FOV will not be constant, and the larger sensor will collect more light.

There is no problem with my argument. I think you objection was covered in posting#13. If not, get back to me.

Your equation also leaves out one rather important variable: exposure time.  If two sensors used different lenses that had the same aperture area for the same FOV, that would imply that the larger sensor was using a longer lens at a smaller f/stop. In that case, the larger sensor would be using a longer exposure time, so it would still collect more light.

I have deliberately left exposure time out of the equation. It goes without saying that more photons can be collected with a longer exposure time. Any meaningful lowlight/image noise comparison between two cameras must obviously be made with the same exposure time, otherwise the cameara with the longer exposure has a grossly unfair advantage.

The larger sensor doesn't just 'permit' a larger lens, it requires a larger lens.

I'm not sure what you mean by "requires". It is certainly possible to use a small aperture lens with a large sensor. If you are saying that a larger sensor requires a larger lens to obtain any significant benefit, then I agree, that is precisely what I have been saying all along ....

It's not like you could stick a 4/3 sensor behind the G11 lens, the image circle is far too small.

The image circle is not the issue, again, see posting #13.

Sensor size and aperture area are linked.  So what you say about aperture area is technically correct, but amounts to semantic quibbling that is essentially meaningless regarding the relationship between sensor sizes and image quality.

You would have to be more precise before I could comment. However, my later posting #26 may make my postition clearer, possibly we might even end up agreeing

Col

[col]

Hi Suede,

I doubt you and I will ever disagree on a matter of theory or physics, or if we do it would quickly be resolved - that's the beauty of science. Any disagreement will likely be over matters of preference, such as do you prefer red or blue.

The physics of it is correct, but it is a very confusing way to try to explain it (as photographers are mostly F/no "thinking").

I don't believe the way I have explained things is inherently confusing, especially since I hopefully wrapped things up a bit better in posting#26. However, I agree entirely that most photographers are "f-number" thinking, and I believe that is a large part of the reason why so many find my way of explaining things foreign and maybe even confusing. It's not as if anything I have said is difficult to understand, and I honestly believe that someone learning photography for the first time would would find all I have said extremely easy to understand, and would regard the old way of looking at things as confusing and misleading, which in my opinion it is.

I could write a good deal about about both the usefulness, and drawbacks, of talking about everything in terms of f-numbers. Back in the good-old-days of film photography, most amateur photographic enthusiasts used the same 35mm film format. That made life very easy, because when one person talked about a 50mm F1.8 lens, everyone knew what that meant in terms of its absolute aperture, depth of field and FOV. The advent of digital photography, with everyone using cameras with different sensor sizes, has completely changed the game, but unfortunately the photographic community has not (for the most part) changed their way of looking at things. Confusion abounds, and much of that confusion is because people have clung excessively to thinking about everything in terms of f-numbers, which served them so well in the past.

 

But I'm very much a "prime" person, I rarely find uses for superzooms or even anything above the normal 3x zoom lenses. Larger sensors in compacts with smaller zoom ratios would fit me perfectly - but I'm aware that those products may be hard to sell in quantities enough to be lucrative to the maker.

We all have different needs and preferences. I agree that it makes sense to put larger sensors in compact cameras if the lens size can be kept compact, while still maintaining a decent aperture area, by either using primes, or a limited zoom range.

I think the S90 is a brilliant camera, not so much for what it is, but for what it shows could be built. I don't need my camera to be quite as small and light as the S90, so for me, a "scaled up" S90, where the sensor and lens were both scaled up such that the total camera size was more like a G11, would be a very interesting proposition. Maybe for you, too.

Cheers, Col

[paratom]

Very theoretical discussion. Personally I have been surprized how sensors have been improved when looking at images from the g11.

There is a risk that discussing such things theoretically one does oversee small things and reality is totally different.

For example the S90 faster lens f2.0 which should lead to better images in low light since one could use either lower ISO or a faster shutter time to avoid shake. But then it seems the g11 with a little more mass and a larger grip is steadier to hold so you would hold the g11 as steady at 1/15 where you would need 1/30 for the S90.

[thierrylegros396]

Very theoretical discussion. Personally I have been surprized how sensors have been improved when looking at images from the g11.

There is a risk that discussing such things theoretically one does oversee small things and reality is totally different.

For example the S90 faster lens f2.0 which should lead to better images in low light since one could use either lower ISO or a faster shutter time to avoid shake. But then it seems the g11 with a little more mass and a larger grip is steadier to hold so you would hold the g11 as steady at 1/15 where you would need 1/30 for the S90.

And more, for the S90, the antishake seems to be not as good as the one in the G10 !

And the lens is not as sharp, especially for edges !

And because there is a lot of barrel distorsion, the firmware corrects for, but you loose a lot of pixels.

So comparing G11 and S90 is comparing oranges and apples.

Have a Nice Day.

Thierry

[Wayne Fox]

And more, for the S90, the antishake seems to be not as good as the one in the G10 !

And the lens is not as sharp, especially for edges !

And because there is a lot of barrel distorsion, the firmware corrects for, but you loose a lot of pixels.

So comparing G11 and S90 is comparing oranges and apples.

Have a Nice Day.

Thierry

Of course, the firmware doesn't correct the distortion if you shoot RAW, and on many subjects the distortion isn't significant enough to need correction.

I also believe if comparing actual prints from the two cameras most of the time to see any differences would require some serious pixel peeping.

[paratom]

Of course, the firmware doesn't correct the distortion if you shoot RAW, and on many subjects the distortion isn't significant enough to need correction.

I also believe if comparing actual prints from the two cameras most of the time to see any differences would require some serious pixel peeping.

I woud expect so too.
IQ has been not the main factor for me to choose between S90 and g11 but more user interface.

[markhout]

FWIW, I have been shooting with the S90 for the last two months, coming from the G10. The portability of the S90 is great, but IQ wise I must say that the DR seems to be even more limited than with the G10, and also the sharpness/clarity of the S90 is less impressive. I confirm "muddy fields and foliage" that has been reported elsewhere.

This is all from judging prints up to 11x14 - no scientific reasoning here.

I will continue to use the S90, but I will be more careful in making sure that the subject matter fits the camera, and not the other way around.

Mark

[BJL]

A 100 pixel square was cropped from each image and the resulting grid is a 1 to 1 pixel ratio on the screen.

So taking a smaller portion of the G10 image than the G11 and S90 images (2/3 as much of the total image area) and enlarging it by a greater degree. How many times does it have to be said: one could make such a comparison between two different crops from the same file and the noise and other imperfections would look worse in the more enlarged display of a smaller portion of the image.

Could someone please do a test in which the cropped images compared cover the same part of the image, and so use the same fraction of the sensor area?

It would also be interesting to do "lens wide open" comparisons of various camera and lens combinations like G11, S90 and GF-1+14-45 so adjusting ISO speed to get equal shutter speed despite differences in lens speed. With compact camera options, "total camera speed" interests me far more than either "sensor speed" or "lens speed" in isolation. (I do not suggest comparing to the GF-1+20/1.7 as that would be a very one-sided contest against the 1/1.7" format cameras!)


That said, the noise gap looks more than the 50% increase in "used image area" can explain, so Sony and Canon have probably made some worthwhile progress with noise reduction.

[thierrylegros396]

Hi BJL,


I use S90 and G10 and I've made a lot of comparisons.

What is evident, the G10 is by far better up to ISO200.

Above ISO800, S90 seems better than the G10.

But it's difficult to say ! Why ? Because the G10 High ISO raw files are color noise processed (very strongly for ISO1600).

Yes S90 High ISO colors are stronger, but sometimes you have "color cast" especially in yellow areas.

The G10 lens is very good with low distorsion and it's easy to remove chromatic aberrations.

With S90, from 28mm to more than 35mm equivalent, distorsion is high and in some cases you cannot completely suppress S90 chromatic aberrations.


So I will use both. S90 for everyday, and G10 when I carry more !

Thierry

[Wayne Fox]

I confirm "muddy fields and foliage" that has been reported elsewhere.

Mark

How are you processing those files?  I'm not really seeing this using LR beta 3 to process the files.

I think it goes without saying the s90 is not an idea landscape camera.  It simply fills the need of making sure "The best camera is the one that's with you" is at least always better than my iPhone, and it's small enough to fill the bill of the family snapshot camera as well.

[markhout]

How are you processing those files?  I'm not really seeing this using LR beta 3 to process the files.

I think it goes without saying the s90 is not an idea landscape camera.  It simply fills the need of making sure "The best camera is the one that's with you" is at least always better than my iPhone, and it's small enough to fill the bill of the family snapshot camera as well.

Oh absolutely. It's just that I have not been able to replicate the same level of separation with the S90 like I have with the G10 in for example this image:

.

Admittedly I have significantly processed this image in LR2, but was able to pull off enough detail to make it into a nice 11x14 print.

Using LR3 on the S90 as well - comparison-wise with the G10, grass/plains/foliage/trees etc just seem a bit more muddied.

Mark

[Wayne Fox]

So taking a smaller portion of the G10 image than the G11 and S90 images (2/3 as much of the total image area) and enlarging it by a greater degree. How many times does it have to be said: one could make such a comparison between two different crops from the same file and the noise and other imperfections would look worse in the more enlarged display of a smaller portion of the image.

Could someone please do a test in which the cropped images compared cover the same part of the image, and so use the same fraction of the sensor area?

It would also be interesting to do "lens wide open" comparisons of various camera and lens combinations like G11, S90 and GF-1+14-45 so adjusting ISO speed to get equal shutter speed despite differences in lens speed. With compact camera options, "total camera speed" interests me far more than either "sensor speed" or "lens speed" in isolation. (I do not suggest comparing to the GF-1+20/1.7 as that would be a very one-sided contest against the 1/1.7" format cameras!)


That said, the noise gap looks more than the 50% increase in "used image area" can explain, so Sony and Canon have probably made some worthwhile progress with noise reduction.

I understand your point, and thought about doing this test as well, but it seemed sort of irrelevant.  I don't think the G11 noise performance is so outstanding it merits upgrading over the G10, and this would most likely just reinforce that.  So the feature set is what we have left, such as the articulating screen

To me these are just point and shoot cameras anyway ... I don't expect that much out of them.  I'm never going to opt for an s90 or a G10/11 if I'm headed somewhere I believe a likely photo opportunity will present itself anyway.

I guess my bottom line is the s90 fits in my pocket, is easy to carry around.  If want something "better" I won't bother with my G10, I'll grab the GF-1.

[hassiman]

I have both the G10 and the G11 and I have found that the IQ of the G11 is almost identical to that of the G10 at ISO 80/100.  At higher ISO 200/800 the IQ of the G11 is far and away better  I have found the appearance of the G11 images smoother and quite pleasing.  I think any comparison of the G10 VS the S90 would hardly be fair as the quality of lens on the G10/G11 is probably better than that on the S90.

The bottom line is that if I really want a quality image file and quick focusing/composition I will have to cough up the money for a M9 to use with my M4's lenses.

All-in-all these little G series cameras... especially the G11 produce images of a quality level that only 5 years ago would be impossible to capture with cameras ten times the price.

[BJL]

I guess my bottom line is the s90 fits in my pocket, is easy to carry around.  If want something "better" I won't bother with my G10, I'll grab the GF-1.

Fair enough: my categories are "go absolutely everywhere" (like a S90 or something even smaller) followed by "go almost everywhere, in a jacket or back-pack pocket", where the GF-1 or E-P2 are very tempting. The G10, G11 and LX3 fall between chairs for me.