Wednesday, February 1, 2012

Fuji X10 versus Canon S100 – High ISO Low Light Comparison

This is an interesting comparison. The most modern small sensor from Canon against a slightly larger modern small sensor from Fuji. The X10 should sweep all categories since it is of the same generation and a bit bigger. But the results will surprise you just a wee bit.

The sensors in fact score dead equal on the overall sensor scores. Now, the Fuji actually has the larger sensor, but the S100 is clearly holding its own here. My guess is that a lot of weight was put on values from the S100’s 80 ISO setting.

What we see when we examine the SNR graph is separation of the curves that is caused by differing “actual” ISO values. But when you look at each data point, you see that the SNR varies by less than 1/3 stops all along, and that is almost within the margin for error. In other words, the S100 is matching the SNR of the X10 at every data point, and with a smaller sensor.

You will also note in the above graph that the ability to shoot at 80 ISO (just over 1/3 stops down) actually gives the S100 a lower SNR at base ISO. Small difference, but again this is on a smaller sensor.

The larger sensor should have better dynamic range. But again, the difference is minimal and the S100 again bests the X10 at base ISO.

Tonal range shows the same trend so I won’t bother posting that graph.

The X10 asserts some of the expected dominance in color sensitivity. Approximately 0.9 bits better all along the ISO range, but again, at base ISO the S100 bests the X10.

Basically, all these graphs are tied together in one way or another by the noise the sensor produces at each ISO.

Which means that the S100 is performing well above its sensor size should allow. Nice job by Canon and why is the X10 not easily outperforming the S100? Perhaps Fuji should have consider back-lighting the EXR-CMOS sensor in the X10? It certainly helps the F550 easily match all sensors at the next larger level in most measurements. DR is one exception, where the S100 spanks the F550 easily.

So. let’s take a peek then at the 1600 ISO and 3200 ISO images at Obviously, they hold the copyright on the original RAW images, but I claim fair use for educational purposes. (Whether or not you agree that this article is useful is a secondary consideration :-)

I produced two sets of crops from their low light images at the same ISO. In fact, the S100 should be compared at one stop of lower ISO to account for sensor sizes, but since the X10 has a faster lens, let’s compare them equally. Remember that, in party situations, people usually shoot wide open, so these are actually pretty close in exposure with each camera at its best. I.e. this is pretty close to an apples to apples comparison because I am penalizing the S100 by one stop of ISO.

Click through to see the larger version.

There is a false sense of detail in the X10 because of the grain that is fairly hard to totally eliminate. I’ve noted this before. As good as the X10 is, the S100 is not all that far behind. But otherwise there is not much to choose.

The quality of the RAW conversion has something to do with this I’m sure, but that is no one’s fault but Fuji’s. They must ensure a quality ACR conversion to be taken seriously in the enthusiast market. Silkypix is not a useful alternative to many of us.

The surprising crop in this test is the 3200 ISO crop. I’ve seen comments on the Fuji forum suggesting that the S100 could not render the back of the book readable. Well, maybe in JPEG. But in RAW, the book is as readable as the X10’s rendering. And smoother too.

The S100 shoots very well in RAW at 3200 ISO. I’m really impressed. The F550 and X10 do have the DR400 trick up their sleeves, but otherwise the cupboard appears a little bare. Fuji need to continue to refine their EXR CMOS sensors, especially with ORBs threatening to hurt the newer and more expensive models.

Now, to finish off. Many people post mainly on the web or make small prints. So let’s have a look at these two cameras at 3200 ISO from RAW.



Each is further tweakable, but you can see that the details each captures are essentially the same.

I am constantly amazed that so many people talk about the image quality of the X10 like it is some sort of revelation. It’s excellent, no doubt (barring ORBs.) But the S100 is pushing the same buttons with a smaller and cheaper package.

Fuji might want to consider a return of the shorter zoom model with a fast(ish) lens to match the S100. The F770 looks amazing to me for concerts, especially if they have not ruined the sensor that makes the F550/F600 so good. But the lens is slow and there are inevitable compromises with so much range. Of course, if the X10 gets cheaper and the ORB fix works, then it will compete well with the Canon.