LensRentals.com

G1X Imatest Results

Posted by

Many of you know there’s nothing I enjoy more than doing meaningless tests just out of curiosity. I tend to not publish most of them because I figure nobody but me is really interested, but I’ll throw this one out there. We got our first batch of Canon G1X cameras in. I’ll be blunt: I don’t particularly like them. I want the option of a viewfinder (the door of most hotel rooms have better viewfinders than the thing they put in this camera) and I’m not happy with a standard range lens that’s f/5.6 for most of its range.

But I have to admit the images I’ve seen with the G1X seem to have really high resolution. So I thought I’d do a comparison and run some G1X cameras through Imatest to see just what the MTF50 turned out to look like. Since I consider the G1X to be mostly a G12 (point-and-shoot) replacement I decided to do some G12 cameras too, just to see the difference. And since I had some 7D cameras with 17-55mm f/2.8 IS lenses handy I thought we’d just see how the G1X stacked up against them.

The usual “don’t read this stuff and go insane” cautions apply: this is a test of lens and sensor resolution done at a distance of about 14 feet. The results might be different at 4 feet or 400 feet. Focus is done manually and bracketed so it’s not a test of a camera’s autofocus ability. It’s tested using a controlled lighting test target with a hardware mounted camera and Imatest software, so photos taken in the dark of night or hand-held for 12 second exposures will give different results.

But the results are interesting just the same. For those of you not familiar with our usual charts, this shows the MTF 50 for 5 copies of each combination. The horizontal axis shows peak resolution in line pairs / image height. The vertical axis is weighted average resolution (center = 1, middle third = 0.75, corners and edges = 0.5) in line pairs / image height. We test with unsharpened RAW images so our numbers will be lower (and more valid) than tests using jpgs.

Also, unlike our normal quality control tests where we shoot the lenses wide open, these are all shot at f/5.6: at the distance chosen the maximum aperture of the G1X lens was f/5.6 so we stopped the others down to f/5.6 to make them equivalent. (Might be worth noting that even the G12 could have shoot at f/4 at this distance.)

 

Results for G12, G1X, and 7D with 17-55 f/2.8 lens

The results aren’t particularly surprising from one standpoint: both the G1X and 7D with 17-55 f/2.8 IS lens resolve way better than the G12 (which is a really good point and shoot, but still a point and shoot). I was a bit surprised that the G1X had slightly better peak resolution than the 7D with 17-55 f/2.8 (both shot at f/5.6); the 7D has more pixels and I would have thought the 17-55 was a better lens. Rumor has it the G1X has a very weak AA filter which may give it an advantage in peak resolution. The G1X didn’t hold up as well in average MTF across the entire image – in other words it wasn’t as sharp in the edges and corners. I can’t say why, exactly, but looking at the 17-55 f/2.8 IS lens and comparing it to the much smaller zoom built into the G1X I would just assume the lens has a lot to do with it.

What does it all mean? Not much. The G1X resolves well, but we all expected that. I’m not sure I expected it to resolve quite this well, and have to say it’s pretty impressive. If you want something as easy to carry around as a G12 that gives an SLR quality image, you won’t miss wide aperture lenses, and don’t mind using the LCD to compose and shoot it’s a good camera for you.

 

Roger Cicala

Lensrentals

March, 2012

9 Responses to “G1X Imatest Results”

Steven said:

I wonder if a slightly more budget DSLR would hold up as well. Eg the Canon T3i with the 18-135 or 15-85.

A said:

Thankyou for that Roger; I’d say those results are very interesting!

Your results confirm what I’ve seen on the sample images at DPReview too:
http://bit.ly/A9UU3X
(comparing JPEG files from the Nikon D4/NikonD3S/Canon5D2/CanonG1X)

Poor viewfinder aside, is the G1X reasonably responsive as a camera?

About the only reason I’m not instantly considering a G1X as a backup camera is that the Nokia 808 looks interesting… I guess in some ways they’re not that different – both are larger sensors than average compared to the market averages.

Speaking of cameraphones; I took some simple sample shots into my local independent camera store a couple of years back. Prints were A4/Letter size; one from my Motorola ZN5 cameraphone and one from my old 400D/XTi with the 24-105IS/f4L, both shot in auto (to level the playing field as the phone doesn’t have many camera features). All the staff preferred the ZN5 shots, as there was more detail in the grass in the foreground! Of course my test favours the cameraphone in many ways, as it’s brightly lit and a relatively small print size, but I thought it was still interesting nevertheless.
http://www.dpreview.com/galleries/7552041385/photos

Steve Fines said:

Even if it weren’t for the great rentals and service it is stuff like this that makes me tell friends about your site.

Really interesting stuff – always enjoy your thoughts and findings on the gear.

Kevin Purcell said:

Interesting data. You are pretty much the only place that publishes data on sample variance to we can see when results are significant. Thanks for that.

One issue with using the same f/5.6 aperture on all the cameras is the G12 with the much smaller 2.0 micron pixels will be starting to see the effects of diffraction at that aperture. The G1X has 4.2 micron pixels.

Scaling for pixel size the equivalent G12 aperture to f/5.6 on the G1X (for diffraction effects only) would be f/2.7 (which you can only do at 28mm eq zoom setting).

The G12 will probably give (slightly) better results at f/4.

The other side to this coin is that much above f/5.6 (e.g. f/8) you won’t need an AA filter as diffraction with do the anti-aliasing for “free”.

The real problem Canon have is putting a compact lens into the “G11/12″ form factor. They more than doubled the size of the sensor (“two stops”) but the lens’s front aperture only increased by 140% (“one stop”) so “too small” apertures are a given with that zoom range.

So all the enthusiasts are complaining about the small lens aperture (“The X10 can do it” … the X10 has a much smaller sensor).

Enthusiasts probably would have preferred this sensor platform with either a shorter zoom (24mm to 70mm eq or 28mm to 80mm and s as fast as you can make it and still have it fold up … f/2 to f/3.5 perhaps) or a fast prime lens (28mm or 35mm or 40mm or 50mm). A Poor Man’s X100 if they priced it at $500 or $600 and it might sell well.

These solutions probably didn’t work for marketing specifying the camera as OrdinaryFolks ™ know that it’s all about zoom range and megapixels. They’re probably already wondering why this doesn’t have 10x or 20x zoom like their travel zoom** It doesn’t seem to meet their needs either. Why buy a G1X when you can buy a “flexible” mirrorless.

This I think is the root of Canon’s problem: they don’t really know who this camera is for but they knew they wanted a compact with a large sensor. It’s priced oddly too high as well. I still think the “short zoom”

One also wonders if a future “Nikon P1X” (imagine a P7100 with Nikon 1 CX sensor) might be a better combination for this idea. The slightly smaller sensor making for a smaller lens and perhaps a 28-100mm eq range.

Still the underlying systems seems good as the measurements show. Perhaps in the next revision but I can’t see a way out other than reducing the zoom range.

** Yes, I know you know. I know too. But they don’t know.

Carl said:

Roger, I’m impressed that you would test such a new camera and post the results! As everyone seems to chime in…”another reason I love this site.”

Kevin, very good points about the difference in pixel size, and diffraction…and about the relative increase in size of the sensor compared to the lens’ maximum aperture.

Haha @ “so photos taken in the dark of night or hand-held for 12 second exposures will give different results.”

It’s interesting that the increased sensor size of the G1x over the G12, perhaps mandated that the lens is effectively an f/5.6 rather than an f/4, at the tested distance (or equivalent effective focal length)?

Carl said:

I personally like my Nikon P7000 just fine; it was well worth the $250 I paid in November 2011. So far I haven’t had the failure issues that a few other users have had. In my opinion, a camera like this isn’t built to take the abuse of constant use…of attempting to get 1000 usable shots in the first few days of using the camera (and therefore shooting 2000 or more). That’s sort of like buying a new car, taking it to the desert, propping up the drive wheels, and letting it run a tad past redline for a week straight.

I prefer the wider 7.1x zoom range the P7000 has, over the Canon G12…and also I prefer the smaller size and lighter weight. The G12 feels like it is lined with lead by comparison, and seems an inch bigger in every dimension (I’m sure it’s a lot less than an inch, but it did feel like it). Noise-wise in RAW mode…well this little Nikon sort of sucks, even at ISO 100…but post editing easily overcomes all of it with little or no apparent loss of resolution. High ISO is obviously another matter.

I also personally still prefer Canon as a company, but they are becoming a bit too much like Apple. Introducing products of ever-decreasing price/performance ratio…because they know there is market demand, much of it based on their name. Unlike Apple though, at least they aren’t hording $100 billion in cash! If they did, I bet the general public wouldn’t happily ignore it, the way they are with Apple.

alek said:

“Many of you know there’s nothing I enjoy more than doing meaningless tests just out of curiosity. I tend to not publish most of them because I figure nobody but me is really interested, but I’ll throw this one out there. ”

Ditto what others said to please keep “throwing stuff out” like this for us Roger!!!

As a 7D/17-55 owner, I too was a bit surprised that combo didn’t fare better. Don’t believe diffraction is playing too big of role at F/5.6 (but you could run a test at F/2.8 and F/4.0 to see – hint, hint! ;-) so yea, perhaps that AA filter is blurring us here. Comparing to a “hot-rod” 7D (with AA filter removed) would be interesting.

Along those lines, please considering doing a Nikon 800 versus 800E shootout … i.e. with and without AA filter.
alek

P.S. Suggestion: Don’t know if easy to do with your graphing software, but consider adding labels those axis. I know you talk about it in the text, but makes it easier at a glance.

Jerry Russell said:

I think it’s amazing how well the G12 did, considering the size of the sensor. 550 line pairs / image height on a 5.7 mm sensor is almost 100 line pairs per millimeter! Why can’t we do that in a DSLR?

Leave a Reply