Canon G10 - You've Got to be Kidding!

[alanmcf]

I very much like the idea of a Canon G10 for all the reasons Michael mentions -- size being number one. Michael says that for a 13"x19" print you cannot tell the difference between the G10 (14.7meg-pix) and a high end medium format back.

I print with a 24" Z3100. I would love any comments on how well the G10 prints will hold up in the 17"x22", 20"x24" and 24"x30" range.

Thanks, Alan

P.S. I kinda think when such a camera is on a tripod at low ASA in a low wind setting shooting landscapes, that it can be as good as any other 14.7 meg pix camera. So my question could also be, how big does a 14.7 image go while maintaining high quality.

[Ray]

P.S. I kinda think when such a camera is on a tripod at low ASA in a low wind setting shooting landscapes, that it can be as good as any other 14.7 meg pix camera. So my question could also be, how big does a 14.7 image go while maintaining high quality.

 

Except for dynamic range. If the landscape you are shooting has a high range of brightness, from fluffy white clouds to deep shadows in the foliage, the G10 would prove to be very inadequate. Of course, even a DSLR can be inadequate is such situations, which is why it's sometimes necessary to bracket exposures for blending.

I get the impression that DR in general tends to be roughly proportional to sensor size (excluding patented technological advancements which can give one camera a DR edge over another with the same size sensor).

In circumstances where you would get by with a single shot from an MFDB, you might have to bracket a couple of shots with a DSLR to get another stop or so of DR. In circumstances where you might just get by with a single exposure from a DSLR, you would probably need to bracket a few G10 shots to provide another 3 or 4 stops of DR after blending.

In circumstances where you need an additional 3 stops of DR from the DSLR, you would need 6 or 7 additional stops from the G10. That's a lot of bracketing and would take quite some time with the G10 with its slow frame rate. Not ideal by any means.

[michael]

Don't expect magic.

The comparison is close when everything goes right, but when it doesn't (which is more often than not) the edge will always go to the larger format.

How big you can go depends on so many factors that it's hard to be definitive. My suggestion is that for under $500, why not give it a try? Or, at almost zero cost, simply download a couple of the sample files that I provided and print them yourself.

Michael

[ejmartin]

People might be interested in some technical analysis:

http://www.naturescapes.net/phpBB3/viewtop...21&start=22


My interpretation of the G10 results is that, with a sensor 20x smaller than a 1D3, you are capturing 1/20th as many photons at raw saturation as the DSLR sensor for a given sensor efficiency, which is what happens also when you increase the DSLR ISO by a factor 20. That's why the S/N plot of the 1D3 at ISO 1600 is comparable to the G10 plot at ISO 80 (give or take; the 1D3 has a 1.3 crop factor, the G10's ISO is understated relative to the 1D3 by the better part of a stop, the G10 has ~50% more pixels, and in the end it's a wash).

If you want to see how the G10 performs, take a 1D3 or FF DSLR, raise the ISO to about 1600, and stop down by a factor 4.5 (3.5 for 1.3 crop) from where you usually shoot.

[Ray]

My interpretation of the G10 results is that, with a sensor 20x smaller than a 1D3, you are capturing 1/20th as many photons at raw saturation as the DSLR sensor for a given sensor efficiency...

Slight exaggeration here, Emil   .

Just for the record, the G10 is described as a 1/1.7" sensor, or 7.176mm x 5.319mm. By my calculations that's only 14x smaller than the 1D3 and 22x smaller than full frame 35mm. If you crop the 3:2 aspect ratio of the 1D3 to the G10's 4:3 aspect ratio, then the G10 is only 12.3x smaller than the 1D3.

Small point though. You probably meant 20x smaller than the 1Ds3.

[ejmartin]

Slight exaggeration here, Emil   .

Just for the record, the G10 is described as a 1/1.7" sensor, or 7.176mm x 5.319mm. By my calculations that's only 14x smaller than the 1D3 and 22x smaller than full frame 35mm. If you crop the 3:2 aspect ratio of the 1D3 to the G10's 4:3 aspect ratio, then the G10 is only 12.3x smaller than the 1D3.

Small point though. You probably meant 20x smaller than the 1Ds3.

Yes, I meant 20x smaller than FF.  The parenthetic remark makes that point by noting the difference is made up by the relative normalization of ISO relative to raw data between DSLR and digicam.  The upshot is that raw data from the G10 is of the same quality as the 1D3 at ISO ~1600, or the 1Ds3 at ISO ~2500.

[Ray]

Yes, I meant 20x smaller than FF.  The parenthetic remark makes that point by noting the difference is made up by the relative normalization of ISO relative to raw data between DSLR and digicam.  The upshot is that raw data from the G10 is of the same quality as the 1D3 at ISO ~1600, or the 1Ds3 at ISO ~2500.

Again, I don't want to appear to be quibbling, but ISO 80 to ISO 2500 represents 5 stops difference, by my calculation. (80, 160, 320, 640, 1280, 2560). However, 20x the photon count represents 4.5 stops (4.5 squared=20.25).

All else being equal, we would expect the G10 at F3.5 and ISO 80 to equal the image quality of the 1Ds3 at F16 and ISO 1600. (Assuming equalised FoV from the same shooting position with appropriately adjusted focal lengths).

Can we hope that Michael will find the time to do this comparison   . I own neither a G10 nor a 1Ds3 so am unable to do the comparison. I think there's a possibility the G10 would produce better image quality in these circumstances.

[ejmartin]

Again, I don't want to appear to be quibbling, but ISO 80 to ISO 2500 represents 5 stops difference, by my calculation. (80, 160, 320, 640, 1280, 2560). However, 20x the photon count represents 4.5 stops (4.5 squared=20.25).

All else being equal, we would expect the G10 at F3.5 and ISO 80 to equal the image quality of the 1Ds3 at F16 and ISO 1600. (Assuming equalised FoV from the same shooting position with appropriately adjusted focal lengths).

Can we hope that Michael will find the time to do this comparison   . I own neither a G10 nor a 1Ds3 so am unable to do the comparison. I think there's a possibility the G10 would produce better image quality in these circumstances.

What is being overlooked is the difference in normalization of ISO relative to raw data between digicams and  DSLR's.   The typical DSLR meters middle grey at about 3.5 stops below raw saturation.  This is nearly a stop below the ISO standard for jpeg output, and I suspect it is done to accomodate the hard cutoff at saturation of digital sensors; by allowing an extra stop of headroom, one can roll off highlights in a way closer to film response via a tone curve.  On the other hand, the typical digicam meters middle grey about 2.5 stops below raw saturation.  This is much closer to the ISO standard.  I suspect the reason is that with much less DR to play with, one doesn't want to leave any DR on the table, and so the metering plays closer to the edge of saturation (which is why digicams are much more prone to blowing highlights).

Regardless, just look at the S/N plots I linked to.   You will see that the 1D3 at the pixel level at ISO 1600 is close to the G10 at ISO 80 at the pixel level:

G10
http://theory.uchicago.edu/~ejm/pix/20d/te...ras/G10_SNR.jpg

1D3
http://theory.uchicago.edu/~ejm/pix/20d/te...se/noise1d3.gif

1Ds3
http://theory.uchicago.edu/~ejm/pix/20d/te...e/noise1ds3.gif

To arrive at S/N ratio at comparable spatial scales, one must compensate by the square root of the MP count (sqrt[14.6]=3.8 for G10, sqrt[10]=3.2 for 1D3, sqrt[21]=4.6 for the 1Ds3); the upshot is, subtract about .25 from the vertical axis of the 1D3 curve, add about .25 to the 1Ds3 curve (which have logarithmic scales).

[Ray]

What is being overlooked is the difference in normalization of ISO relative to raw data between digicams and  DSLR's.   The typical DSLR meters middle grey at about 3.5 stops below raw saturation.  This is nearly a stop below the ISO standard for jpeg output, and I suspect it is done to accomodate the hard cutoff at saturation of digital sensors; by allowing an extra stop of headroom, one can roll off highlights in a way closer to film response via a tone curve.  On the other hand, the typical digicam meters middle grey about 2.5 stops below raw saturation.  This is much closer to the ISO standard.  I suspect the reason is that with much less DR to play with, one doesn't want to leave any DR on the table, and so the metering plays closer to the edge of saturation (which is why digicams are much more prone to blowing highlights).

Emil,
It sounds as though you are saying that ISO 80 on the G10 is closer to ISO 100 in practice. The 1Ds3 doesn't have any incremental settings between ISO 1600 and ISO 3200, so any real world photographic comparisons between the G10 and 1Ds3 should aim for a full ETTR with the G10 at the G10's setting of ISO 80, compared with an underexposure of about 1/2 a stop with the 1Ds3 at ISO 1600. Does this sound right?

The critical thing, in my view, is to have equal exposures with each camera, equal DoF and equal FoV, when making a comparison of image quality.

[ejmartin]

Emil,
It sounds as though you are saying that ISO 80 on the G10 is closer to ISO 100 in practice.

Closer to ISO 125 or 160.

The 1Ds3 doesn't have any incremental settings between ISO 1600 and ISO 3200, so any real world photographic comparisons between the G10 and 1Ds3 should aim for a full ETTR with the G10 at the G10's setting of ISO 80, compared with an underexposure of about 1/2 a stop with the 1Ds3 at ISO 1600. Does this sound right?

Yes that's about right, perhaps even a tad more underexposure on the 1Ds3; though such an underexposure will still give more highlight headroom to the DSLR since one still has the overall DR of ISO 1600 (for typical situations it will go unused however due to the underexposure).  Shadow noise will be close.

Anyway, this is all fine tuning.  The main point is that the G10 is in the ballpark of ISO 1600 on a  FF DSLR (or around 800 on 1.5 crop), stopped down by a factor of around 4 (or 3 for 1.5 crop) to approximate the DOF.  That gives a rough sense of the image quality one should expect.