G Series Resolution

.

In a previous post, I showed that my G12 has less noise at high ISO than does my G9. DxOMark shows this comparison more scientifically and in more detail. Some credit for this improvement in noise should go to the reduction in pixel density and, I assume, some credit goes to noise reduction software (in-camera in this case). I'd also like to think that some credit would go to true improvements in sensor technology. Whatever the real case, the first explanation is usually that an increase in megapixels means an increase in noise.

The ultimate in noise performance would be, I suppose, a single but very large pixel; however, that design would not be very useful to photographers. We can feel intuitively that there must be an advantage to having many pixels. That advantage is resolution.

Dpreview includes resolution testing as part of its intensive camera reviews and reports. I've summarized the Dpreview resolution test results for the G series cameras below.

In the above comparisons, “Resolution” is the horizontal “Absolute Resolution”in lines per picture height as measured and reported by Dpreview. Dpreview includes a vertical resolution in their test reports but as that number is similar to the horizontal resolution, I've omitted it in favor of brevity. Remember that G series sensors have been essentially (but not exactly) the same physical size. The G12 has not been tested by Dpreview at this date.

Interestingly and as expected, G series resolution has generally increased with pixel count but there also appears to be an effect of time, or more likely, improved technology. The 10MP G11 has better resolution that does the 10MP G7 and even better resolution that my 12MP G9. I expect that the resolution of the G12 is likely to be about the same as the G11.
..

G12: RS60-E3

.



I was pleased to see that the G12 could be triggered remotely and especially pleased to discover that I already had the remote! I'd bought the RS60-E3 some years ago to use with a film camera (anyone remember those?). Anyway, it still works and works well. The trick (if there is one) is to treat the RS60-E3 the same as you would the on-camera shutter button: press the button on the RS60-E3 lightly (just as you should do for the on-camera shutter button), think "Focus", hesitate (or watch the G12 screen for the green rectangle confirmation) and then finish off pressing the button.

If you only fully press the RS60-E3 button then the result is the same as if you had fully pressed the G12 on-camera shutter button: the G12 tries to focus but gives up and fires after a brief internal even if focus has not been fully achieved. The result is images that seem randomly focused or unfocused.
.

G12 and G9 in Noise Modes

.

This attractive scene was especially arranged for purposes of testing G9 and G12 at high ISO. The G9 and G12 were tripod mounted and set for 2 second shutter delay to minimize camera shake. Cameras were set in Av mode at f4 and zoomed to produce roughly the same composition. The picture above was taken at ISO 80 with the G12 at 0.6 second exposure. The picture below is at ISO 3200 with the G12 at 1/60 second. All images in this post are from the in-camera JPEGs.

At first glance, the G12 ISO 3200 shot is nearly indistinguishable from the ISO 80 shot – especially for these reduced and compressed images. (Click on the images for an enlarged but not full size view.) The image below is from the G9 at ISO 1600 and 1/40 second.

As noted by DxOMark, the actual and indicated ISO are different for the G9 and G12. For purposes of testing, I decided to first use the G9 at ISO 800 and 1600 and then adjust the ISO of the G12 to produce the same exposure as the G9. The corresponding G12 ISOs were 1000 and 2000 but this turned out to have a relatively minor effect.

Yielding to the temptation to pixel peep, the image below is 100% crops at ISO 800 for the G9 and 1000 for the G12, click for a larger image.

Even though we know not to pixel peek, doing so shows that higher ISO produces noisy images. The noise is even worse at ISO 1600 for the G9 and ISO 2000 for the G12 as shown below.

From the 100% crops of high ISO images, one would think that prints would be terrible but I printed a full frame 8x10 of each shot and they were all surprisingly better than expected.  I made a game out of guessing the camera and ISO for each print and included ISO 800 and 1600 for the G12. Although I sometimes confused G12 ISO 800 vs 1000 and G12 ISO 1600 vs 2000, I usually got the ISO order approximately correct and could always sort out the G9 prints from the G12 prints with careful examination. What I'm trying to say is that at first glance all the prints looked OK but there are differences. I suspect that a casual observer would not readily detect the difference between a G9 ISO 800, G9 ISO 1600, G12 ISO 1000 and G12 ISO 2000 print but I'd like to think that the ISO 80 and ISO 3200 prints would be noticed by most people.

So, what to make of all this? It is obvious to me that the G12 produces better high ISO images than does the G9; in fact, I'd say that the G12 is about one “stop” better than the G9. That is, if ISO 400 is OK on the G9 then ISO 800 is OK on the G12, etc. I'll try to stay at ISO 800 or less on the G12 but for casual snapshots even ISO 1600 is probably OK.

With the release of Adobe Camera Raw for the G12, I hope to make further reductions in noise by shooting in RAW – more later.
.

Compare G Series in DxOMark

.
DxO Labs has a website, DxOMark, that can be used to make various comparisons of cameras, lenses and sensors.  To compare the various G series sensors, go to the menu on the left hand side, click "Camera Sensor" and then "Compare Sensors".  Up to three sensors can be compared. 

I first chose to compare the G12 and G11 sensors.  These are essentially the same so I then compared the G12, G10 and G9.  (I've linked my comparisons but the link doesn't seem to work for everyone.) The G12 has an overall score of 47, the G10 scores 37 and the G9 35 but it is more useful and interesting to examine the charts that compare the ISO sensitivity, signal to noise ratio, dynamic range, tonal range and color sensitivity.  The charts are accessed by clicking on the rectangles just above the pictures of the cameras. 

On the ISO Sensitivity chart, notice that the measured ISO of the G12 and G10 are relativity close to Canon's rating but the measured ISO of the G9 is noticeably higher than Canon's rating.  For example, when the G9 is set for ISO 800, the real ISO is 1068.  Taking another example, when making noise comparisons between the G9 and G12, if the G9 is set at ISO 1600, it is really working at 2281; the G12 set at ISO 1600 is really working at ISO 1378.  To account for these differences, the remainder of the DxO charts are based on measured ISO.

To my way of viewing the charts, the G12 sensor turns out better in every way than the G9 and G10.  The charts do not show image resolution because resolution would be a function of sensor and lens combined. 
.

G12 vs G9 Images: Sky, Grass, Telephoto

.



This is a sort of "for what it's worth" comparison -- and it may not be worth much but was easy to do.   The important aspect of these comparisons is they are probably representative of photos, settings and actions that a photographer might do with either the G9 or G12.  I made these shots while shooting the lake scene in the previous post.

In the photo above, I simply took a shot of the sky with the G12 in Av mode at f4.5, ISO 80 and zoomed all the way to telephoto.  The G12 selected a shutter speed of 1/640 second.  I did the same for the G9 but backed off the telephoto to somewhat duplicate the G12 focal length.  (Turns out I missed it.  The G12 shot is at 30.5mm, the G9 at 36.8mm.)  The G9 also selected 1/640 second.  I then cropped a 400 pixel x 400 pixel sample from the center of each of the in-camera JPEGs and pasted them side by side.  In addition to the G9 shot being slightly darker, it seems to me to have a "texture".  OK, so we're pixel peeping here but, still, I prefer the G12 sample.

Next, I repeated the experiment by photographing the grass.  The G9 was zoomed to its widest focal length of 7.4mm and set for f4 in Av mode; exposure was 1/320 second.  Although the G12 will go even wider (6.1mm), I tried to guess at 7.4mm (missed that one also -- actual setting was 8.1mm).  When set for f4, the G12 selected an exposure of 1/250 second.  The picture below shows the 100% crops side by side.  I thought the G9 might be a bit soft but it seems fine to me.  I'd accept either shot.

Next, I wanted to uprez the G12 telephoto to the same image size produced by the G9.  Recall that the G9 is has 12MP and 44.4mm telephoto as compared to the 10MP and 30.5mm telephoto of the G12.  I uprezzed the G12 telephoto shot from the previous post until the tree in the center of the G12 photo was the same size on my computer screen at 100% pixel view as was the G9 telephoto shot of the tree.  Then I cropped those 100% views and pasted them side by side as shown below.

The G12 does not come out so well in this comparison and I did not expect it to.  After all, "uprezzing" is a process of creating pixels where none existed previously.  If I'd thought to do so, I'd have used the digital zoom of the G12; however, the process I used is about the same (I think).  This side-by-side comparison shows that true optical zoom is better than digital zoom.  For scenes needing focal lengths longer than that of the G12, I'll be using the G9.

(I hope the differences show up as described after the posting process.  The JPEGs were saved at the highest setting in Photoshop.  You should be able to click on an image to see it full size.)
.

G12: Comparison to G9 Landscape and Zoom Range

.

Just as I often go to the Rural Life Museum to test photo gear, I also shoot this scene routinely. For the above shot, I put the G12 in Av mode at f4.5, ISO 80, small Flexizone focus rectangle and widest zoom setting (28mme). The matching shutter speed was 1/1000 second. The matching G9 shot is below (except that somehow the f-stop was set to f4 on the G9). These are the untouched JPEGs straight from the cameras except for downsizing.

There isn't a lot of overall difference between the G12 and the G9 images except that the G12 image includes slightly more of the overall scene because of the wide angle lens (28mme vs 35mme). What about zooming in?

The picture above is from the G12 at maximum zoom and f4.5, 1/400 second.  The 140mme telephoto of the G12 produces a bit wider view than does the 210mme of the G9 (f4.8, 1/250) shown below.  The G9 choose to slightly increase exposure as compared to the G12.  The G9 exposure is probably a result of the tighter zoom.

Not so apparent here, but it seemed to me that the G12 produced a slightly better image than the G9 when the focal lengths were forced to be the same; however, this may well be self-justification on my part.  I do think that the shadow areas in G12 images showed slightly better detail.

I suspect that RAW images (which I also obtained) from the G9 and G12 might well be processed  to have even fewer differences than these in-camera JPEGs.   I'll give those RAW images a try after Adobe releases the necessary update to Camera Raw.

.

G12: Screen blanking demo

. 
Others are beginning to discover and report on the screen blanking problem of the G12.  Here's a video by Tim Harmsen demonstrating the G12 problem while showing that the G11 does not have the same problem.  My thanks to Tim -- wish I'd thought to make a video!
.