Tuesday, December 20, 2011

Fuji X10 vs Olympus XZ-1 vs Panasonic LX5 vs Canon G12 – How close are they at 100 ISO?

DPReview has some shots on their comparator and I thought I would perform one of my analyses of the image quality to see how these cameras fare against each other in JPEG. Why JPEG? The comparator does not seem to have RAW files for the X10 … that will no doubt come later when DPReview releases their long-awaited review. But for now, a lot of people are saying that this camera was meant to shoot in JPEG so let’s take a trip around to see why.

Caveat: Olympus do not have a world class JPEG engine. They are here for completeness, but in fact I will have to do a comparison in RAW at some point to see how they really fare against one another. Panasonic might also suffer, but Canon has a world-class jpeg engine as does Fuji (for their recent CMOS EXR cameras and the X100.)

Obviously, these images and the comparator are copyrighted by DPReview. I claim fair use for the purposes of education and illustration.

Starting with ISO 100 …

Crop 1 – The Coin


The X10 easily beats the others in fine detail retention … the yellow fur on Pluto’s nose makes that abundantly clear. But we also see moiré in the form of bands of color in the coin. There is an interaction here from either (a) a very weak anti-aliasing filter, or (b) a strange effect in the demosaic algorithm, which of course must deal with side by side pixels in a pattern at a similar angle. Either way, the detail is magnificent, but as with other moiré-prone cameras like the Nikon D70s, there is a price.

Order of goodness: Tied X10 (for the dog) / XZ-1 (for the coin), G12, LX5

Crop 2 – The Face and the Needle-Point


This is my favourite crop, as it demonstrates 3-dimensionality in the wool and simultaneously demonstrates the ability to resolves very close lines. And since the X10 is shot here in its 12mp mode, we should see these lines quite clearly. Especially given the sharpness shown in crop 1.

But no … we don’t. How strange that. The lines in the face are replaced by more moiré, and this is starting to look like a pattern (pun intended.)

The wool is basically identical across the board since all do a nice 3D rendering at base ISO.

Order of goodness: XZ-1, G12, LX5, X10 (the X10 surprised me by losing outright.)

Crop 3 – The Watch Face


This is probably the crop that caused DPReview to put their full review on hold. The watch face is another loser for the X10. Soft and slightly blurred.

This is a difficult one to call, since the angle to the watch completely changes the visibility of the notch in the hour marks and the clarity of the writing.

Order of goodness: Tied G12 / XZ-1 / LX5, X10

Crop 4 – The Globe


This one tends to be unequivocal. Either the text is clear or it is not. And how small can you go before you cannot read the text? The X10 suffers a bit from softness. And it is hard to tell here if this is classic Fuji corner softness or the softness of the demosaic process. Doesn’t matter though, for this test.

Order of goodness: Tie G12 / XZ-1, LX5, X10

Crop 5 – Paper Clips


Here’s one that tends to show tonal gradations as they pertain to 3-dimensionality. It also shows how well edges are treated. And WOW! We have a very clear ordering for the first time.

The XZ-1’s local contrast is magnificent. Easy winner. The G12 is next with the same clarity but less contrast. The LX5 has unfortunate jagged edges. And the Fuji is just soft again. Really soft this time.

Order of goodness: XZ-1, G12, Tie LX5 (jaggies) / X10 (softness)

At this point, I give up. As did DPReview. The corners are all soft on the X10, and the middle line detail on the pictures (the heads for example) show moiré and other demosaic artifacts everywhere. Just a mediocre performance.

Result Table – ties included, so numbers don’t quite add up

  X10 XZ-1 G12 LX5
Firsts 1 5 3 1
Seconds     1  
Thirds 1   1 3
Fourths 3     1

Obviously, a large sensor and a fast, short range lens should make things easy. The XZ-1 makes it a clean sweep with a magnificent performance here. For a base ISO shooter, the XZ-1 stomps this copy of the X10. The G12 is a good alternative if you want the longer reach. Possibly the best compromise out there. The LX5 is decent, and certainly better than this copy of  the X10.

So … for JPEG shooters at base ISO, I see no contest here. A few owners of the X10 have shown magnificent images, but the majority have frankly fit in pretty well with this review. A lot of images look oversharpened, which would confirm the initial softness that must be compensated somehow.