Comparing Image Quality in the RAW

I know that pixel peeping is not representative of the perceived image quality you’ll see either at web-sized images or at print sizes up to 16″ x 20″ or even at 24″ x 30″. But still, just for the sake of comparing image quality of the Fuji X-Trans III 20 Mpix sensor with the MFT 20 Mpix sensor along with a few others, I went to the DPReview website and looked at their RAW file comparator. Here are the surprising results.

The first comparison is for high ISO between the Fujifilm X-Pro 2 (with the APS-C sized 24 Mpix X-Trans iii sensor) and the Olympus PEN-F (with the MFT sized 20 Mpix sensor) at 3200, 6400 and 12800 ISO. These are 100% crops of the RAW files as shown in the DPReview comparator, without noise reduction appleid.

X-Pro_Pen-F_3200 X-Pro_Pen-F_6400 X-Pro_Pen-F_12800

As you can see, the X-Pro 2 is quite usable at 6400 ISO and even 12800 ISO could do in a pinch (remember these are 100% crops). Conversely, the PEN-F is not very usable beyond 3200 ISO (keep in mind these have not had any noise reduction applied, these files would actually be usable at higher ISO after some processing). In fact, the MFT sensor has only a bit less noise at 3200 ISO as the X-Trans sensor has at 12800 ISO. I would say the Fuji sensor has about a 1.5 stop advantage over the MFT sensor and given that the FUji’s noise is more luminance than chroma, you might get a 2 stop advantage after noise reduction is applied.

Just for fun, I then compared the Fuji to the Sony A6300.

X-Pro_A6300_3200 X-Pro_A6300_6400 X-Pro_A6300_12800

As you can see, the Sony 24 Mpix APS-C CMOS sensor has about the same amount of noise at ISO 6400 as the Fuji has at ISO 12800, but, with much nastier chroma noise causing a lot of blotchy colour artifacts. Again, the Fuji has at least a 1 stop advantage, maybe more after noise reduction.

Next, I also compared the Fuji with the 36 Mpix FF Nikon D800 both at 6400 and 12800 ISO.

X-Pro_D800_6400 X-Pro_D800_12800

To my eye, the Fuji at 6400 is about mid-way between the Nikon at 6400 and 12800 ISO, so the full-frame Nikon only has about a half stop advantage over the Fuji. Again though, the Fuji noise appears to be more of the luminance kind, which is easier to clean-up than the chroma noise present in the Nikon file.

I then made another comparison with the PEN-F, but this time looking for the level of perceived detail, both at 200 ISO.


As you can see, the Fuji appears significantly sharper than the 20 Mpix MFT sensor.

The following one is with the 24 Mpix FF Nikon D610. To my eye, the Fuji has a lot more perceived detail even though the two sensors have the same resolution.


This next one is with the legendary 36 Mpix FF Nikon D810.


To my eye the Nikon has a slight advantage on perceived detail.

I repeated the test at 12800 ISO just for fun, the D800 loses most of its advantage in perceived detail and has much worse, blotchy chroma noise.


Last but not least, I compared the Fuji against the Sony Alpha 7R II, widely acknowledged as having the best FF sensor on the market with its back-illuminated 42 Mpix FF sensor.


Clearly the Sony is superior to the Fuji but at three times the cost and it has some weird colour blotches on the wall to the right of the frame.

I think these show that the Fuji X-Trans III sensor is really quite good and capable of holding it’s own, being only a little behind the best full-frame hi-res sensors in terms of detail and noise, but better than what you get from other 24 Mpix APS-C or FF sensors.

Clearly the MFT sensor is well behind the Fuji, but the real question is if one forgets pixel peeping and looks at real world situations (web images and prints up to 16″ x 20″), is the MFT good enough? All else being equal, the Fuji system clearly produces better image quality than the MFT system, but if the MFT is good enough in real-world scenarios, it also offers other advantages in terms of size, speed, in-body image stabilisation and other high-tech features like focus bracketing and hi-res mode.

The investigation continues…