G7 Review on dpreview

[jgjulio]

The G7 review has been posted on dpreview.
http://www.dpreview.com/reviews/canong7/

What do the G7 users think about their conclusions?

[radiohead]

I thought it was pretty fair - and dissected the pros and cons as I see them.

[puttick]

Yes, very thorough and mirrors my experience. But why did it take him so long - he says he's had the camera for over a month.

One thing all reviews seem to have missed - if you decrease the resolution to 6.1MP you can get almost the same IQ at the same print sizes. Why is this of any interest? Because by "combining" the pixels, the visible noise level in the output is made lower and it makes ISO 400 or even 800 cleaner.

[RafaPolit]

I wrote this in the G7 thread:

mattjs,

thanks for the info, I have been waiting eagerly for the DPreview review.

I have to admit that while reading I was somewhat dissapointed, several downsides, other very nice features, but the comparisson with the A640 showed almost no differences... I don't want to spoil everyone's "fun" but by the time you reach to the conclusion you will probably be (as I was) not very convinced with the camera.

Just "turn the page" to the final examples of a Japanese trade market, you will be blown away by this cameras real-life picture capabilites!!!! I was realy, realy impressed by the colors, the detail, the low-noise, and, of course, the fantastic pictures. I believe that if anyone doubted this cameras possibilities, this pictures will put your doubts to rest.

I often agreed with the author of the review, that had the G7 been made into a 6MP cammera (like the new Nikon D40) the IQ would have been much better.

Now, Puttick, the process of taking a smaller resolution image has been discussed in the past for other cameras, and if I remember correctly, most people agreed that, since the camera used all the sensors pixels to achieve the image and then reduced it to the desired output, there was no real advantaje to what you would achieve in post processing by simply reducing the image with a bicubic or so algorithm.

Can you please make a great service to us and post a controlled environment pictures at same settings (preferably at mid-hig ISO) with different resolutions selected to see if what I just described is not the case of the G7? That would be very interesting. It would save a lot of space in the SD cards as well!

Rafa Polit Jr.

[Toogy]

Yes, very thorough and mirrors my experience. But why did it take him so long - he says he's had the camera for over a month.

One thing all reviews seem to have missed - if you decrease the resolution to 6.1MP you can get almost the same IQ at the same print sizes. Why is this of any interest? Because by "combining" the pixels, the visible noise level in the output is made lower and it makes ISO 400 or even 800 cleaner.

How do I decrease the resolution? Which setting equals 6.1 MP? I'd be more than happy with 6.1 Megapixels if the noise was lower.

Thanks

[puttick]

How do I decrease the resolution? Which setting equals 6.1 MP? I'd be more than happy with 6.1 Megapixels if the noise was lower.

Thanks

Function menu - image size - M1 (2816x2112)

Actually 5.95MP, sorry for the slight miscalculation. And no, I haven't compared it yet with resampling in photoshop.

However I suspect there may still be a difference compared to downsampling in PP, because the Digic III processor is doing in-camera noise processing before we see the finished jpeg.

For a RAW image (which obviously we can't get at here) the argument about using the lower res image selected in-camera vs downsampling in PP would be exactly correct. I wonder if the reason Canon hasn't given us RAW on the G7 (or the other 10MP chip cameras) is that the output would be very obviously noisy...?

I will try and post some examples, but its not easy as you can't use 100% crops for comparison (the image scale will not be the same). What I did was shoot the same scene twice with the different image sizes (10 and 6MP - at the same high ISO for both), then with the lower res image at 100% view, zoom the other to approximately the same linear size on screen for comparison on-screen. For comparing printed output, I chose the dpi neccessary to make both images the same output size on paper, and printed a small section side by side. The lower res image showed less noise and when printed there was little if any difference in visible resolution of detail. I'll try the same with downsampling in PP next.

Also the Michael Reichmann review on Luminous Landscape suggests the higher ISO images respond well to noise reduction software such as NeatImage (I have Noiseware)

Nigel

[puttick]

Downsampling in Photoshop

OK, here's a quick comparison of an underexposed part of a darkish image shot at ISO200, that does show some graininess, chroma noise, and some noise reduction artefacts. Image 1 is 10MP, 2 is 6MP, 3 is 3MP, and all have been resized by downsmapling using bicubic smoother in PSCS2 to produce 50cm wide prints, then viewed at print size on screen, so all appear the same size - you can see the zoom ratio. This is a screenshot (part of a 1024x768 screen), actual pixels.

To me, they all look identical in terms of noise, so the downsampling doesn't appear to be having any effect in this experiment.

If i get the chance tomorrow I'll try to repeat this with in-camera image size.

[puttick]

Downsampling applied in-camera

OK, I've retrieved the images I shot before. Both are at ISO 400, within a few seconds of each other. The one on the left is at 10MP (large, superfine), and on the right 5.95MP (in camera - medium 1, superfine). No other manipulation.

I've set the lower resolution image to 72dpi (screen resolution, so you are seeing it at 100%) and calculated the necessary dpi to make the 10MP image the same size on-screen (in this case, 93 dpi).

Now this is more interesting. The 10MP image to me clearly shows more noise and more noise reduction artefacts. Look in the shadow areas for noise (graininess), and particularly the green grass for artefacts (dotty effect). The 6MP image is cleaner.

Note that the resolution of fine detail is virtually indistinguishable between the two (see the branches).

I must point out that this is pixel peeping: the actual print size, viewed at this level, would be 40" x 30". At a more normal size, or viewed from a greeater distance, there might well be no detectable difference.

Interesting, though, all the same. I do believe 6MP in-camera shows less noise at the same ISO, at least in this image, with my camera, in my hands. I intend to set this as a custom function and test it some more.

Nigel

[radiohead]

Very interesting Nigel - I see 6MP as more than adequate for what this camera will be used for so will set mine to M1, Superfine, for the week and see what occurs.

[jgjulio]

Nigel, nice comparison of the two resolutions. Yours is the first post where I can actually see the difference between the noise at the two resolutions.
Thanks

[GPete]

Great comparison, I to believe that Canon left raw out of the G7 because of the noise primarily due to the fact that I have a G6 and for the most part have to use noise reduction when tooling my raw shots in photoshop.

[RafaPolit]

puttick,

Thanks for the comparisons... but what I was hopping was a comparison between downsampling in camera and downsampling in photoshop. For that, what I think would be a good starting point will be taking a full 10MP resolution image, then a 5.95MP image at the same settings.
Latter on, use the photoshop Image Size in the 10MP image to match the pixel count of the 5.95MP image, that would be exactly: 2816 x 2112 pixeles, using bicubic downsampling, no other change made.

Then make a 100% crop of both at the same location. They should be exactly the same size without any DPI calculations. If you use the DPI approach, the printed results would be telling, but onscreen results are not that useful, since you have to introduce pixel interpolation for onscreen results and having different zoom values applied to each image.

If you could do this one more test I will highly appreciate it. Thanks again for the time taken to show useful comaprisons.

Rafa Polit Jr.

[puttick]

puttick,

Thanks for the comparisons... but what I was hopping was a comparison between downsampling in camera and downsampling in photoshop. For that, what I think would be a good starting point will be taking a full 10MP resolution image, then a 5.95MP image at the same settings.
Latter on, use the photoshop Image Size in the 10MP image to match the pixel count of the 5.95MP image, that would be exactly: 2816 x 2112 pixeles, using bicubic downsampling, no other change made.

Then make a 100% crop of both at the same location. They should be exactly the same size without any DPI calculations. If you use the DPI approach, the printed results would be telling, but onscreen results are not that useful, since you have to introduce pixel interpolation for onscreen results and having different zoom values applied to each image.

If you could do this one more test I will highly appreciate it. Thanks again for the time taken to show useful comaprisons.

Rafa Polit Jr.

I have done that, however with two different images. I have not had time to reshoot, nor will I for a while.

The first image shows the effect of Photoshop resampling from 10MP to 6 MP and 3MP. You are not correct - using 100% crops of these would not be the same size on screen, you require to make each lower MP image correspondingly larger to be able to compare the results.

Also with the in-camera downsampling - by setting the image size for the 6MP image using 72 dpi (screen resolution) and displaying at "print size" I have given you a 100% crop. A 100% crop of the 10MP image would have a bigger scale, which is why I chose to scale that one down (@ 93dpi), rather than the 6MP image because if I chose 100% for the 10MP image I woudl have had to use say 120% for the 6MP image, introducing visible pixellation. You have to introduce a scaling factor to get the images to the same visible size - note this was not resampled, just rescaled.

If I get the time I will go back to the 10MP image (form the second comparison) and downsample it in photoshop. This is the sort of thing the commercial reviewers should be doing!

Nigel

[puttick]

OK let's see if this answers the question posed by Rafapolit - note that I do not agree with him about the 100% crop issue, this gives different size images.

So here we have the G7 image shot at ISO400, all superfine quality. From left to right, we have (1) the original 10MP image, (2) the 10MP image downsampled to 2816x2112 i.e. 5.95MP, and (3) the in-camera 5.95MP image shot on M1.

(1) is displayed on screen at 93 dpit, (2) and (3) at screen resolution ie 72 dpi.

My interpretation of the appearances is as follows:
(1) shows grainy noise, particuarly in the shadow areas, and noise reduction artefacts (dotty effect) in the smooth grass areas.
(2) shows the same, essentially unchanged despite the photoshop downsampling. Maybe just a little improvement?
(3) shows a reduced level of grainy noise, and fewer noise reduction artefacts.

For me, the in-camera downsampling beats the photoshop downsampling.

I suggest this is due to the strong in-camera noise reduction, which has been applied to (1) at 10MP - it has converted "native" noise to "reduced noise plus artefacts", and so this still shows in (2) - in other words you can't remove the artefacts introduced. Whereas (3) had lower noise to start with due to pixel binning, before the in-camera noise reduction was applied and so has less noise and fewer artefacts.

The key here is the in-camera processing that has taken place. We don't have access to an un-processed image of course.

Nigel

[RafaPolit]

Nigel,

Thanks, this is exactly what I was looking for. Sorry if I wasnt clear before, what I meant was: A 10MP image resampled to the size of the 5.95MP would yield almost identical images in terms of distribution and location to the original 5.95MP image, and a 100% crop of each will not need aditional zooms to match scales, as is the case with your shoots (2) and (3). I was not refering to your first example, but to your second.

You were right about the noise, although (2) is a bit more sharp, it has more noise, while (3) is softer with less noise. The thing is that a smart sharpen (CS2) or an unsharp mask (CS or CS2) would make number (3) a better image than (2), while you will need some noise reduction and later a sharpening applied to (2) to better the result.

Thanks again for the time and info.

Rafa.

[puttick]

I'm glad we have come to the same conclusion - I have found it a worthwhile exercise. It's important to remember that the actual print size would be huge - in this case we are looking at a 40x30 inch print! I don't think it woud need any sharpening, as in a more normal size print (say 16x12 inch) it would be considerably smaller.

Nigel

[JaGWiRE]

Man, all this stuff is making me look harder at that Panasonic LMX-2, although it has some noise problems too it seems. Damn, why can't P&S cameras have less noise problems?

[puttick]

The G7 is definitely less noisy than the Panasonic, according to the images on reviews such as dpreview. Noise is a characteristic of the sensor, and is worse with smaller sensors and more megapixels, so more of a factor in P&S cameras, compared to dSLRs. Some of us remember film - most digicams, even small P&S, are less grainy than film - ever try ISO1600 film!!!?

The images I posted above at ISO400 are very clean indeed until you start looking at them at 100% (i.e. 30x40 inch print!). At smaller print sizes, but still quite big by most standards such as 16x12", the noise will hardly be visible.

If anyone worries about using the 6MP setting rather than the 10MP setting, again you porbably won't notice any real difference even at quite large print sizes. On my wall I have a B&W image printed 12x18" which most people assume is from medium format film. In fact it's from a 4MP Ixus (cropped to 3MP or so)! A little graininess is visible (noise) but the overall IQ is good enough to enlarge to a huge extent, so I would expect to be quite happy with 6MP from a G7 which has a better lens in the first place.

Equally, a friend has the Panasonic LMX2 and has some great large photos from a Himalayan trip, OK a little noise is visible, but when viewing the photos you don't really notice it.

You'll enjoy a G7. It's not a dSLR, though, and you will have to learn to resspect its limitations and work around them.

Nigel

[JaGWiRE]

The G7 is definitely less noisy than the Panasonic, according to the images on reviews such as dpreview. Noise is a characteristic of the sensor, and is worse with smaller sensors and more megapixels, so more of a factor in P&S cameras, compared to dSLRs. Some of us remember film - most digicams, even small P&S, are less grainy than film - ever try ISO1600 film!!!?

The images I posted above at ISO400 are very clean indeed until you start looking at them at 100% (i.e. 30x40 inch print!). At smaller print sizes, but still quite big by most standards such as 16x12", the noise will hardly be visible.

If anyone worries about using the 6MP setting rather than the 10MP setting, again you porbably won't notice any real difference even at quite large print sizes. On my wall I have a B&W image printed 12x18" which most people assume is from medium format film. In fact it's from a 4MP Ixus (cropped to 3MP or so)! A little graininess is visible (noise) but the overall IQ is good enough to enlarge to a huge extent, so I would expect to be quite happy with 6MP from a G7 which has a better lens in the first place.

Equally, a friend has the Panasonic LMX2 and has some great large photos from a Himalayan trip, OK a little noise is visible, but when viewing the photos you don't really notice it.

You'll enjoy a G7. It's not a dSLR, though, and you will have to learn to resspect its limitations and work around them.

Nigel
I wonder how the LMX2 at 7.5 megapixels (4:3 I believe) does noise wise against the G7 at 10 megapixels (4:3 too?)
I kind of like the wide aspect ratio feature in a way to be honest, haha. I might go with the Panasonic, noise on both cameras isn't great, but it's not like one is that much further ahead the other. The Panasonic does lack the zoom, but it is a bit smaller. I think RAW is something I won't be able to live without though. It would be nice to be able to use a hotshoe flash, but things don't always work out the way you would like them to I guess.

Can you shoot with these cameras indoors without the flash? With my SLR I always loved raw because even if I shoot indoors with no flash, I can still adjust the temperature (remove tungsten color casts), and other features, and get a good looking shot.

[puttick]

They're all compromises, of course, and you have to choose what fits best for you. Both are nice cameras, and both will no doubt be superseded by faster, better, quieter etc in future.

BTW you don't need RAW to shoot under tungsten lighting, but I'm sure you knew that already! The real issue shooting indoors with a P&S is lighting, whcih is usually low, so you need high ISO (noise) or image stabilisation (so you can use lower ISO, and still avoid shake, but can't stop a moving object). More compromises I'm afraid. For me, I would like to be able to use a decent flash when I need to, so the G7 wins.

As you have realised, the LMX2 is only a 10MP camera when used in 16:9 mode. Nor does it really have a 28mm lens, as only the horizontal angle of view is equiv to 28mm, not the vertical. This limits its usability as a "wide angle". At 4:3 it is a 7.5 MP camera (3/4 of the sensor is used) and the field of view is equivalent to a 37mm lens (compared to 35mm on the G7). The noise in 4:3 / 7.5MP mode will be EXACTLY the same as the noise at 16:9 / 10MP...because the only difference is that the ends of the sensor are not used, there is no combining of pixels to get the lower MP count. Not looking so good compared to a G7 now, perhaps?

:)

Nigel

[radiohead]

The D-Lux 3 verdict from AP (this follows a 4 page review btw):

"Leica equipment has traditionally been distinguished from the company's many competitors first and foremost by quality. As you might expect, such high standards of design and engineering come at a price. There have always been cameras from other makers which boast more functions for the same or lower price. but that's almost beside the point. The Leica brand represents luxury as well as durability, and when in the past you saw a camera that bore the Leica name and the instantly recognisable red cirle, you knew you were looking at something a bit special.

The D-Lux 3 certainly looks and handles like a qaulity piece of kit, and excepting the need to dip into menus rather too frequently I have few criticisms with regards to the handling. The included leather case also adds a touch of genuine class to the package. Even though 16:9 images can be awkward to print, I really like the format and look foward to see whether Leica stays with it for future models.

Unfortunately, although my experience of actually handling and taking photos with the camera has been largely enjoyable, I have concerns about the image quality, especially with regards to digital noise at high ISO's and the unpleasant affects of in-camera NR. It is a shame that Leica has chosen to pack more pixels onto the sensor of the D-Lux 3 than it appears to be able to cope with. The fact is that at ISO settings of 400 above, digital noise is simply destructive to the final image. Curvilinear distortion at the wideangle setting of the lens will also concern some users, although on balance I think I would rather have a slightly compromised WA facility than none at all. Importantly, it should be remembered that all the problems noted above will be to some degree less obvious in prints smaller than A4, but I suspect that a photographer willing to pay almost £500 for a Leica camera will want, and expect, more from their pictures"

For:

* Luxury build quaility
* Refined handling
* Good bundled software
* Reliable metering

Against

* Image noise
* Lens distortion at wide/tele settings
* Slightly fussy menus
* LCD visibility in bright sunlight

Specification - 27/30
Build - 18/20
Handling - 16/20
Performance - 22/30

AP Rating = 83%

-------------------------

I'll repeat this when I get the G7 review issue this coming weekend. Hope that's helpful anyway.

[JaGWiRE]

Heh, doesn't seem that bad.

Is there any hope firmware update wise? Does anyone think the G7 might get raw, and the panasonic might get better noise reduction? I remember Bill saying the G7 could get raw..

I'm also contemplating, how much better are these cameras, then the previous generation (Canon G6.) Is size the biggest thing happening here with the longer lens on the G7?

[mattjs]

Those are two big points. But also DIGIC III (versus DIGIC II in G6). LCD no longer flips out & twists but the G7 LCD is bigger (G7=2.5" -- G6 =2") with incresed resolution (G7=207,000 pixels -- G6=118,000 pixels). Image Stabilization (tradeoff that the G7 lens isn't quite as fast as previous G cams but still...IS is nice). I also think the ISO dial is a great thing, but I've seen others pan it as uneccessary.

Simon at dpreview did ask Canon directly about the possibility of adding Raw to the G7 via firmware. The Canon answer was no. They could always change thier mind on that but take it for what it's worth. If the camera sells well, they probly won't have much incentive to add Raw. Here's the post at dpreview -- http://forums.dpreview.com/forums/read.asp?forum=1010&message=20303766



.

[JaGWiRE]

Those are two big points. But also DIGIC III (versus DIGIC II in G6). LCD no longer flips out & twists but the G7 LCD is bigger (G7=2.5" -- G6 =2") with incresed resolution (G7=207,000 pixels -- G6=118,000 pixels). Image Stabilization (tradeoff that the G7 lens isn't quite as fast as previous G cams but still...IS is nice). I also think the ISO dial is a great thing, but I've seen others pan it as uneccessary.

Simon at dpreview did ask Canon directly about the possibility of adding Raw to the G7 via firmware. The Canon answer was no. They could always change thier mind on that but take it for what it's worth. If the camera sells well, they probly won't have much incentive to add Raw. Here's the post at dpreview -- http://forums.dpreview.com/forums/read.asp?forum=1010&message=20303766
.
Damn, I'm stuck in a pickel now. I want a P&S, and the G7 would be that p&S, but it has n oraw so it's not. The Panasonic / Leica looks iffy too, but it has raw, and I will not give up raw in my workflow, even if it's a P&S.

[RafaPolit]

Simon at dpreview did ask Canon directly about the possibility of adding Raw to the G7 via firmware. The Canon answer was no. They could always change thier mind on that but take it for what it's worth. If the camera sells well, they probly won't have much incentive to add Raw. Here's the post at dpreview -- http://forums.dpreview.com/forums/read.asp?forum=1010&message=20303766

Mattjs, I actualy read that thread at dpreview when it was posted. But I still have some hope, since the Pro1 got much improved autofocus via firmware. Probably, as you say, was due to low selling numbers, but they just might add it for the fun of it!;) :D

Rafa.

[jkdlee]

heres another review if you're interseted

http://www.dcresource.com/reviews/canon/powershot_g7-review/

[puttick]

Like I said, compromises, compromises...

;)
Nigel

[mattjs]

You're right Rafa. I mean you never know. Maybe they read all the talk about RAW and come to the conclusion that (even if it's selling well) they could crank up production and sell a boatload more by adding it.

Even though they said no, that was only maybe a week or two after they announced the thing. They could definitely soften on that over time. Would make a nice marketing point as an upgrade for a G8 though too so who knows. Or do you think they'll just abandon RAW altogether in all the Point & Shoots? Reserve it for DSLR only going forward?

[JaGWiRE]

You're right Rafa. I mean you never know. Maybe they read all the talk about RAW and come to the conclusion that (even if it's selling well) they could crank up production and sell a boatload more by adding it.

Even though they said no, that was only maybe a week or two after they announced the thing. They could definitely soften on that over time. Would make a nice marketing point as an upgrade for a G8 though too so who knows. Or do you think they'll just abandon RAW altogether in all the Point & Shoots? Reserve it for DSLR only going forward?

I'm not about to buy this moment, but if the G7 doesn't get raw, I'm not buying it, simple as that, and I'm sure I'm not alone in this decision.

[GNMink]

The G7 is an awsome camera, just take it for what it is....an advanced P&S

Here are some macro shots with the G7, its the best macro P&S I've used

http://photography-on-the.net/forum/showthread.php?t=239952

[BNF]

Those are two big points. But also DIGIC III (versus DIGIC II in G6). .

The G6 (and G5) had DIGIC. (I)

[mattjs]

Good catch, I forgot the G6 had DIGIC I. That should be a pretty decent jump in terms of speed then from DIGIC I to DIGIC III.

[puttick]

Good catch, I forgot the G6 had DIGIC I. That should be a pretty decent jump in terms of speed then from DIGIC I to DIGIC III.
No comparison. The G7 leaves them in the dust.

Nigel