Fuji’s X-Trans sensor sucks? (Part 1 – Moiré and Details)

“Fuji’s sensor sucks!”

“I wish Fuji made a bayer sensor!”

“Lightroom can’t process X-trans files”

“Worms, paint effects, X-Trans sucks!”

Above are some complaints towards the Fuji X-Trans sensor (paraphrased) I’ve read over time.  I’ve decided to try and explore them a little and maybe see if things have improved or if these ideas are actually true.

XtransSucksPart1Title

(Note: I know not everyone who has a problem with Fuji is a troll. I’m just having a bit of fun).

 

 

I will be looking around the web a little to see what I can come up with and I will be comparing some RAW files from other manufacturers to see how they compare (using DPreview’s studio scene comparison chart).

 

NOTE: Even though this post will compare the Xpro2 to many Full Frame cameras.  I’m not trying to debate Full Frame vs Crop anymore.  I’ve covered that in other posts.  I am merely trying to test if the X-Trans sensor is a good or a bad thing.

 

THIS IS NOT A FULL FRAME vs CROP POST

(it’s more like X-Trans vs Bayer)

 

What’s X-Trans?

I highly recommend looking at this first if you don’t know about bayer filters etc.

Understanding Digital Camera Sensors

Basically, a sensor is a bunch of light sensitive cavities (photosites) that collect color particles (photons).  In order to see colour you need to put filters of red/green/blue over them so they only capture those photons. The data can then be used to create a colour image (in a process called “Demosaicing”)

The main way of arranging these filters is called a “Bayer Array” and it looks like this:

Bayer_matrix.svg

Bayer Array/Matrix

 

X-Trans was developed by Fuji as an alternative array.  It looks something like this:

XTrans_matrix

X-Trans Array/Matrix

 

So why this new array?

In short:  Moiré. The problem with Bayer is it’s too even.  Old analogue film didn’t get moiré because the crystals in film and photo paper aren’t even in size and placement.  X-Trans is an attempt to simulate the randomness of film and reduce moiré.

What’s Moiré? You ask.  Moiré is when the sensor can’t handle the fine details and it gets confused.  The more MP you have the less chance you’ll get moiré because there’s more data to work with.  However it can still occur even with 50+MP

 

Moiré Comparison

“X-Trans sucks, I don’t even get any beautiful moiré!”*

*Is what a troll might say

This part of DPReview’s test chart (see HERE) is designed to show up any moiré.  Basically if it appears to bend around and back onto itself near the centre, that’s moiré showing up.

Moire test. Look in the middle.

Moire test. Look in the middle.

You can see the X-Trans just doesn’t show any moiré (or it’s invisible to me).

This just doesn’t affect test charts.  Anything with fine details can show it up.  Like this:

Moire test. Look at the colour artifacts on the wall.

Moire test. Look at the colour artifacts on the wall.

 

As you can see, the fine line details in the wall introduce colour artifacts. This is due to moiré.

In this comparison all of the Bayer filter cameras show some colour moiré.  The 5Dsr and A7RII being the worst.  APS-C cameras seem to fair better, but it’s still there slightly.  7DII doing the best, but also lacking some details (probably has an AA filter).   Fuji in this test seems to lack some details compared to the A6300, let’s see if that’s truly the case in our next test.

 

DETAILS COMPARISON

“X-trans reduces details because it sucks!”*

*that troll is back!

Well, I don’t exactly trust DPreview to process the X-Trans files right.  X-Trans sensors require a different approach to Bayer sensors (for one, they can take a LOT more pushing of the “Details” slider in Lightroom.  Something you’d never do with a bayer sensor.

So, I’ve decided to download a few RAW files and test it myself (to make sure DPreview didn’t cook them)

 

Testing parameters: 

RAW files imported into LR.

NR and Sharpening turned off

Profile set to Adobe Standard

As much as possible I tried to make this a blind test.  So as I was reviewing, I didn’t know which one was which until I checked later.

 

100% crops:

Open in new tab to see full size.

Detailss

 

I’ve decided to break them into 3 camps.  Lower/Medium/High MP cameras.  Remember these are 100% crops, you can certainly get away with 20-24MP for many things, but this is a details test:

Lower MP cameras:

  1. 7DII – Little soft. Some moiré and noise affects fine details.
  2. A6300 – Sharpness is good. Heavy Moire affects fine details.
  3. X-Pro 2 – Very little moiré. Fine details slightly worse than 2. Overall sharpness better than 2.
  4. D500 – Sharpest of the smaller MP sensors. Heaviest moiré.  Appears sharpening has been added at RAW level as there are slight sharpening artefacts.

Rank:

1st: A6300 and Xpro2 equally (A6300 more details, Xpro2 less moire)

2nd: 7D mark II

3rdD500

 

Medium MP cameras:

  1. K-1 (with Pixel Shifting) – One of the best overall. Sharp, no moire.
  2. D810 – Less sharp than 1, shows moire.
  3. K-1 (standard) – Softer than 1 or 2. Less moire than D810

Rank

1stK-1 (with pixel shifting)

2nd: K-1  (less sharp than the D810, but softness is easier to deal with than moire).

3rd: D810 – Still good, just the K-1 is better IMHO.

 

Higher MP cameras:

  1. 5Dsr – Sharp and detailed. Moire reduces fine details.
  2. A7rII – Less sharp than 1. Moire also makes fine details disappear.

Rank

1st: 5Dsr

2ndA7rII

 

Overall Ranking (sharpness and clarity):

1: Pentax K-1 (with pixel shifting)

2: Canon 5Dsr

3: Sony A7rII

4: Pentax K-1 (standard mode)

5: Nikon D810

6: Sony A6300 and Xpro2

7: Canon 7D mark II

8: Nikon D500

 

X-Trans does NOT reduce details

 

So among its peers (20-24MP crop cameras) the Xpro2 holds it’s own.  Lack of Megapixels seems to affect the details of course.  So I decided to compare the Xpro2 (24MP) to a couple of Full Frame cameras that are similar in MP count.  The Canon 6D (20MP) and the Nikon D750 (24MP):

 

Similar MP (will Full Frame make it better?)

similar MP

 

Again I tried to do as blind a test as possible. I had no idea which one was the Nikon and which one the Canon.  I did know the Xpro2 because I know its look.

Nikon D750: Little soft. Major moiré affects details.

Canon 6D: Softer than 1 lacks fine details. Moiré affects details.

Xpro2: Sharper than 2, slightly softer than 1. No moiré. Fine details is similar to the Full Frame cameras.

So the fact the Xpro2 is not Full Frame doesn’t necessarily affect details.

 

 

Conclusions for part 1

X-Trans has a distinct advantage of being able to avoid moiré.  This helps it maintain details in areas of fine lines and details.  Due to lack of pure Megapixels it can’t compare to a higher MP camera of course, so if you need the extra resolution go for one of those.  HOWEVER, there is nothing to suggest the X-Trans sensor is a disadvantage for details over traditional bayer sensors, in fact I’d say it improves them in a lot of cases.

 

I’ll add you have to be quite careful with Fuji files in Lightroom.  Odd artifacts are easy to introduce because I think Adobe haven’t quite worked out the demosaicing process.  You can use the Iridient X Transformer plugin if this is an issue, but I’d rather Adobe got their act together.

 

Thanks for reading,

signature

 

Part 2 will be about noise and artifacts, especially in Lightroom. I’ll try and get that famous paint effect.