Sunday, June 3, 2012

Fuji Sensors … how does the X10 sensor compare with that in the F550EXR, HS30EXR or F770EXR?

The answer is … they are closer than expected.

The 2/3” sensor in the X10 is 89% larger in area than the 1/2” sensor in the F550EXR and the HS and newer long zooms from Fuji. The X10’s sensor is front side illuminated while the 1/2” sensor is back side illuminated,which improves the efficiency of smaller sensors. Whether the X10’s sensor should have been BSI technology is academic now, since it isn’t. But I personally feel that they should have simply made a larger version of the design they were already using, as it is excellent. Anyway, the difference we should be seeing on DXO Mark’s numbers should track around 5/6 stops …

Well, the difference in dynamic range at base ISO between the X10 and the F550 is about 7/10 stops, not 9/10 stops. But still not bad. The S100fs, the previous 2/3” bridge cam from Fuji using an SCCD implementation) is also slightly higher than the F550EXR, but by less than half a stop.

x10_f550exr_s100fs_DR

By the time you get to 800 ISO, there is a 1 stop difference, so the sensors are in their weight classes at this point.

So what about noise? The SNR graphs illustrate …

x10_f550exr_s100fs_SNR

That’s a lot closer, which should be a surprise.

At base ISO, the F550 is at 31.5dB and the X10 at 32.8dB. That 1.2dB, or about 1/3 of a stop difference. Not much to speak of, and easily overcome with the difference in exposures and skills.

At 1600 ISO, where noise performance really starts to matter, the F550EXR is actually closer to the X10. It is at 20.7dB against the X10’s 21.5dB. Less than 1/3 stops separate them. How can this be? In my opinion, this is a direct result of the BSI / FSI difference. And shame on Fuji for that.

And did you notice that 3200 ISO is missing from the X120 graph? This is a kinky thing about the new X series Fujis … the sensors are given analog amplification only to 800 or 1600 ISO. Then it is all digital internally. This works really well for the X100 and it seems to work well for the X10. It is, in fact, a time honored technique that is usually well hidden. I believe that the concept is called “unity gain”, where there is no more advantage to raising ISO and instead you should just underexpose to get your shutter speeds and digitally amplify.

However, Fuji hangs it out there for all to see, apparently.

x10_f550exr_s100fs_ISO

The ISO for 3200 is actually identical to the ISO for 1600. Set the dial and nothing happens inside the camera until you shoot. Then a curve is applied to pull up the output. I presume that more NR is applied too, since under exposing always amplifies noise. But I’m sure that there is not that much difference between analog and digital amplification these days … the algorithms are no doubt quite sophisticated.

The bottom line, of course, is that the X10, for whatever reason, is not performing in its class. Or perhaps the 1/2” sensors are performing way about their class. Considering that the S100fs is about the same as the F550, I’d say that the much newer generation 2/3” sensor is performing below its class.

Tonal Range and Color Depth are measured in bits. DXO’s own pages focus mainly on color depth for portraiture, where people shoot a lot at base ISO. They have this to say about the scores:

A color sensitivity of 22bits is excellent, and differences below 1 bit are barely noticeable.

x10_f550exr_s100fs_tonal_range

x10_f550exr_s100fs_color_sensitivity

None of these cameras gets close to 22 bits of color sensitivity. The X10 is, however, about 1.2 bits ahead, and that might be noticeable on a portrait. But since values under 1 are barely noticeable, a value close to 1 is not going to shatter the earth.

And for the tonal range, there is essentially no difference.


conclusion

Fuji is making some good cameras. Dynamic range is excellent at the sensor, even on the tiny sensors. And these numbers do not take into account the split sensor modes. These modes reduce noise and improve dynamic range, rather dramatically at DR400.

What that basically means is that even the tiny 1/2” sensor will easily stomp the S100fs for most shooting scenarios in bright light. The X10 also.

Should you automatically choose an X10 over an F770EXR, todays representatives of these two classes?

No. You should know what you want to shoot before you make that decision. If you want to chase kids around indoors, then the faster lens will make more difference than the paltry improvement in noise. But if you want to shoot birds and wildlife, then the reach of the HS or F series will clobber the X series every single time.

Once you know what you want to shoot, you can then decide how you want to shoot. I recommend that you seriously consider getting Lightroom 4 and shooting RAW on these cameras. Enthusiasts understand that the capture is half the problem. The rest is in what you do with it. But if you are only interested in having fun, then all of these cameras will allow you to have fun in JPEG mode. Remember to tweak the JPEG engines to match your taste in tone and color. The HS series and X series allow tweaking of more parameters, so factor that in.