High ISO & Noise issue with Canon G10

[heater]

Please excuse my ignorance (I am not a pro. photographer).  What happens if I take shots at less than the max. MP rating (14 MP), for example at 5-7 MP?  Will the noise at high ISO be the same as shooting at the 14 MP?  

I love the user-friendliness of the G10 (which led me to consider it over other cameras) but am somewhat concerned with the reports of high noise.  I suppose I could also ask how critical this noise issue is for a casual photographer who, among other things, will be taking photos of grandchildren indoors as well as outdoors?  

Thanks for help.  

[jdemott]

Shooting using one of the reduced resolution settings on the G10 is not a solution for high ISO noise.  However, you are thinking in the right direction.  The best solution for high ISO noise is to shoot with the full 14.7 MP in RAW mode and then process the file using your preferred noise reduction software (I like Noiseware Pro).  The unfortunate by-product of any noise reduction software is to eliminate detail, resulting in what is effectively a lower resolution image similar to a file with fewer megapixels, just as you suggest.  

Noise in low light files taken with the G10 is similar to what is seen in all other compact cameras.  The only real solution is to use a DSLR, but that would sacrifice the convenience you are looking for.   The G10 is certainly capable of taking typical family photos indoors and outdoors with very pleasing results.  If you apply a little noise reduction, you will still be able to produce files at ISO 400 or even 800 that are quite satisfactory for e-mailing, posting on the Web, or printing at sizes up to 8 by 10 or so.  The noise is likely to be a problem only if you want to make large prints with lots of detail.  Viewing high ISO files side by side with DSLR files, you would notice the difference, but for family photos I think most people will be very happy with the G10.  And for files taken in brighter light, the G10 output is excellent.

[heater]

Thanks for the reply.  I appreciate it.  One question:

Noise in low light files taken with the G10 is similar to what is seen in all other compact cameras

Wouldn't this statement apply only to cameras with equal MP's?  For example, wouldn't the noise issue be less significant with either the Lumix FZ28 or Olympus  SP-565 UZ (10 MP cameras)?  These are other cameras I have considered.  My gut tells me there might not be any significant difference   However, I defer to the experts.  Thanks.

[DarkPenguin]

Thanks for the reply.  I appreciate it.  One question:



Wouldn't this statement apply only to cameras with equal MP's?  For example, wouldn't the noise issue be less significant with either the Lumix FZ28 or Olympus  SP-565 UZ (10 MP cameras)?  These are other cameras I have considered.  My gut tells me there might not be any significant difference   However, I defer to the experts.  Thanks.

If they have the same sensor size it should be the same at the print level.  At a pixel level it will probably be worse.

[walter.sk]

Please excuse my ignorance (I am not a pro. photographer).  What happens if I take shots at less than the max. MP rating (14 MP), for example at 5-7 MP?  Will the noise at high ISO be the same as shooting at the 14 MP?  

I love the user-friendliness of the G10 (which led me to consider it over other cameras) but am somewhat concerned with the reports of high noise.  I suppose I could also ask how critical this noise issue is for a casual photographer who, among other things, will be taking photos of grandchildren indoors as well as outdoors?  

Thanks for help.  

I use the G9 on the days when I don't take the 1DII and a bunch of lenses with me.  I was fed up with the high ISO noise, even at ISO 200.  I shoot RAW, and had tried both Noise Ninja and NIK DFine 2  for noise reduction but did not really like the results with the G9.  I downloaded DXO Optics Pro for use with the 1DII and just for the heck of it, included the G9 module.

DXO reads all the settings and for the G9 is able not only to correct for the noise, but also optical lens distortion and color fringing.  The "default" settings are tailored to the focal length, focal distance, aperture and ISO settings and more, and does an amazing job.  I convert my CR2 files to output as DNG files from DXO, and do the rest of what I do in Photoshop.  DXO's noise reduction does a better job on automatic than I can do manually with Noise Ninja or DFine, and it maintains more detail.

[jdemott]

Wouldn't this statement apply only to cameras with equal MP's? For example, wouldn't the noise issue be less significant with either the Lumix FZ28 or Olympus SP-565 UZ (10 MP cameras)? These are other cameras I have considered. My gut tells me there might not be any significant difference However, I defer to the experts. Thanks.

Other things being equal (which they seldom are), noise will be a function of the pixel density on the sensor.  That is, sensors that have smaller photosites for individual pixels will have greater noise...hence, a DSLR that puts 12 MP on a "full frame" sensor will have much less noise than a typical compact camera that puts the same 12 MP into a much smaller sensor.  The two cameras that you mention have pixel densities of 36 MP/cm2.  The Canon G10 has more megapixels but also a larger sensor resulting in a slightly "better" density of 34MP/cm2.  The Panasonic LX3, which is said to do slightly better than the G10 in high ISO noise, has 10 MP on a larger sensor than the two you mention resulting in a density of 24MP/cm2.  By contrast, the Nikon D3, which is generally thought to be the current low noise champion among DSLRs, has a pixel density of 1.4MP/cm2.  That is why I say that all current compact cameras are somewhat similar in noise performance when compared with DSLRs--there is an order of magnitude difference between the two types of cameras.

If you want the convenience of a compact camera, then you have to accept that high ISO performance is going to be less than optimal.  Among compact cameras, the G10s noise performance is certainly adequate.  If it is important to have the best noise performance in a compact camera (perhaps sacrificing other features or qualities), then you probably should look at the LX3 or the Fuji F200EXR.

[ejmartin]

Other things being equal (which they seldom are), noise will be a function of the pixel density on the sensor.  That is, sensors that have smaller photosites for individual pixels will have greater noise...hence, a DSLR that puts 12 MP on a "full frame" sensor will have much less noise than a typical compact camera that puts the same 12 MP into a much smaller sensor.  The two cameras that you mention have pixel densities of 36 MP/cm2.  The Canon G10 has more megapixels but also a larger sensor resulting in a slightly "better" density of 34MP/cm2.  The Panasonic LX3, which is said to do slightly better than the G10 in high ISO noise, has 10 MP on a larger sensor than the two you mention resulting in a density of 24MP/cm2.  By contrast, the Nikon D3, which is generally thought to be the current low noise champion among DSLRs, has a pixel density of 1.4MP/cm2.  That is why I say that all current compact cameras are somewhat similar in noise performance when compared with DSLRs--there is an order of magnitude difference between the two types of cameras.

This is quite incorrect, parroting a popular internet myth promoted by DPReview among others.  There are two properties under discussion -- pixel size and sensor size.  It is important not to conflate the two, or blame the shortcomings of one property on the other.  Far and away the greatest contributor to image quality is sensor size -- capture more photons and a better image results.  It is largely immaterial how many photosites a sensor of a given size has; at least, whether larger or smaller photosites are better on a fixed size sensor depends on the application (smaller photosites are better at low ISO, larger photosites win at high ISO, as a general rule, in current models).  

It is of course true that when two sensors of the same pixel count but different size are compared, one is varying both the pixel size and the sensor size.  So when one observes that the larger sensor produces better images, one cannot conclude it is because the photosites are larger; one has to isolate one variable at a time.  When one fixes the sensor size and varies the pixel size (and therefore the pixel count), one finds (eg compare the D3 and the D3x at DxO using the "print" tab to compare properties at a common output size), there is very little to choose.  It is harder to fix the pixel size and vary the sensor size in the same generation of technology, since typically one would not care for the low resolution of the smaller sensor that would result.  But it doesn't so much matter; compare any CMOS FF DSLR to any CMOS APS-C DSLR, again using DxO on the "print" tab, and the FF sensor, with over twice the area, will win handily on most measures.  

Compare DSLRs to digicams and the benefits of larger sensor area become obvious (again, one can't fix the pixel size because no one would want a 1MP digicam).  The typical digicam does poorly because its sensor is about 16-20x smaller than a full frame DSLR; so one shouldn't be surprised that the digicam images at ISO 100 look like the DSLR images at ISO 1600-2000   -- both capture about the same number of photons when properly exposed for the corresponding ISO's.

[headroom]

The G 10 has Pixel Binning 14.7MPix  gives 3 MPix with 3200 ASA its fantastic it is in the Szene Menue
Here my Picture with above Settings + Noise Ninja Habib Koite in Zuerich


http://www.wernergraf.ch

[Ray]

The G 10 has Pixel Binning 14.7MPix  gives 3 MPix with 3200 ASA its fantastic it is in the Szene Menue
Here my Picture with above Settings + Noise Ninja Habib Koite in Zuerich

Not at all bad, for the size. Large image and print size is the crucial test, however. If large print or image size is irrelevant, then the G10 is proibably ideal.

[BJL]

Far and away the greatest contributor to image quality is sensor size -- capture more photons and a better image results.

Emil, I agree with most of what you say, in particular the part about gathering photons faster from the subject being the main ingredient in lower noise at a given shutter speed. But for comparisons of noise levels when the need for adequate shutter speed forces the use of higher than minimum ISO speed, what matters most for photons captured and thus noise levels is not sensor size but effective aperture size. You are making the usual assumption of equal ISO speed and thus equal aperture ratio needed for equal shutter speed, which means that the larger sensor and the corresponding longer focal length used give a larger effective aperture size, and thus faster light gathering from same subject. For example, increasing linear format size by 1.4x ("one stop") and thus doubling sensor area also involves increasing focal length and effective aperture diameter by 1.4x and so doubling effective aperture area, which doubles the rate at which photons are gathered from the same subject.

But if one can reduce aperture ratio in the original format by one stop, one again increases effective aperture diameter by 1.4x, doubling light gathering speed, and one can get the same shutter speed with half the ISO speed, giving a similar improvement in noise levels. (We are talking about high ISO situations, so halving ISO speed is possible.)


Of course larger sesors do indirectly allow greater light gathering speed by allowing larger effective aperture sizes, since a given aperture ratio correspond to a larger aperture size with the longer focal length used to get the same FOV with a larger format. Even slow SLR lenses like f/3.5-5.6 zooms are faster (at gathering light) than the fastest current compact digicam lenses, due to larger effective aperture diameters.

But give credit where credit is due for increased speed: to the "big glass", not the "big silicon" alone.

[ejmartin]

Emil, I agree with most of what you say, in particular the part about gathering photons faster from the subject being the main ingredient in lower noise at a given shutter speed. But for comparisons of noise levels when the need for adequate shutter speed forces the use of higher than minimum ISO speed, what matters most for photons captured and thus noise levels is not sensor size but effective aperture size. You are making the usual assumption of equal ISO speed and thus equal aperture ratio needed for equal shutter speed, which means that the larger sensor and the corresponding longer focal length used give a larger effective aperture size, and thus faster light gathering from same subject. For example, increasing linear format size by 1.4x ("one stop") and thus doubling sensor area also involves increasing focal length and effective aperture diameter by 1.4x and so doubling effective aperture area, which doubles the rate at which photons are gathered from the same subject.

But if one can reduce aperture ratio in the original format by one stop, one again increases effective aperture diameter by 1.4x, doubling light gathering speed, and one can get the same shutter speed with half the ISO speed, giving a similar improvement in noise levels. (We are talking about high ISO situations, so halving ISO speed is possible.)


Of course larger sesors do indirectly allow greater light gathering speed by allowing larger effective aperture sizes, since a given aperture ratio correspond to a larger aperture size with the longer focal length used to get the same FOV with a larger format. Even slow SLR lenses like f/3.5-5.6 zooms are faster (at gathering light) than the fastest current compact digicam lenses, due to larger effective aperture diameters.

But give credit where credit is due for increased speed: to the "big glass", not the "big silicon" alone.

I would have said that it's part of the package.  A P&S lens on a DSLR sensor is going to vignette like crazy; a DSLR lens on a P&S wastes  most of the image circle; if that image circle were to be concentrated instead onto the P&S sensor, it would saturate way before it could collect enough photons to match the DSLR.  So, usually one makes the statement that larger sensors are better, and it is understood that the whole system should be scaled as a result.  Scalings of a factor of two or less (APS) can certainly use the same glass, but don't make use of the extra photons, since the extra photons coming through the aperture are not concentrated onto the sensor, but fall on the part of the image circle the sensor doesn't cover.

[Jeremy Payne]

The G 10 has Pixel Binning 14.7MPix  gives 3 MPix with 3200 ASA its fantastic it is in the Szene Menue
Here my Picture with above Settings + Noise Ninja Habib Koite in Zuerich


http://www.wernergraf.ch

Could you possibly post the same image without the NR?  I'd love to see them side by side.

[headroom]

Could you possibly post the same image without the NR?  I'd love to see them side by side.

...here