Sunday, July 19, 2015

Fujifilm X-T1 vs Nikon D7200


Fujfilm X-T1

Time for a check-up. Fujifilm's X-Trans sensor was introduced back in 2012 and allowed for a 16mp APS-C sensor to punch above its weight in terms of image quality. If you took a X-Pro1 when it first came out, there was something about it that made it stand out from the the Nikon D7000... and a case could be made that it could stand its ground against the 1st generation of APS-C sensors from the likes of the Sony NEX-7 and Nikon D3200.

However, amongst enthusiasts, the benefits of the image output was contentious depending on who you talked to, but amongst the general population the X-Trans sensor is a crowd pleaser. Since then, there have been several iterations of higher resolution 24mp sensors from the competition, whereas X-Trans has remained at 16mp. If it were purely about the numbers, it would be one story, but when it comes to image quality it is never that simple.



Nikon D7200

Practical Differences


Setting aside the differences in image quality, these are the major usability factors in favour of the Nikon D7200:

  • More lens choice
  • Better flash photography options
  • Longer battery life
  • Better motion tracking autofocus
  • More comfortable hand holding position
  • Better video quality

In favour of the X-T1:

  • Smaller and lighter body
  • Lens choice is better suited to travel and street photography
  • Electronic viewfinder is physically larger than optical finder in D7200
  • WiFi app is better featured
  • Focus peaking. (Seriously, Nikon...when?)

What follows is largely a discussion of how well Fujifilm's sensor has aged since its debut, but ultimately, your choice between the two comes down to form factor and functionality. The X-T1 is not as good as the D7200 for bird and professional wedding photography, but the Nikon is simply not as small and portable as the Fujifilm.

Bayer Battern vs X-Trans


If there is one thing to remember about the difference between the X-Trans sensor and the traditional Bayer pattern, it's that they produce different renditions; each has its strengths and advantages, but each respective strength also depends on where you move the goalpost. Some other things to remember:

  • The basic structure of the X-Trans sensor is provided by Sony; Fujifilm's proprietary colour filter array is then added afterwards. 
  • X-Trans is a repeating 3x3 pattern that rotates 90 degrees as it repeats, whereas the Bayer pattern is a rotating 2x2 block. This means that the Fujifilm sensor has 5 green pixels for every 2 red and 2 blue pixels, whereas in the Bayer pattern, the ratio is 2:1:1. In other words, the X-trans sensor is more sensitive to green light, but is less sensitive to red and blue. This is a conscious choice on the part of Fujifilm, as the majority of what we see is in the green portion of the colour spectrum, and is what the human eye is naturally sensitive to.
  • The extra amount of green pixels means that the X-Trans sensor is also more sensitive to luminescence information than the equivalent Bayer pattern sensor. This has two implications. The first is that the quality of high(er) ISO output tends to be subjectively pleasing, as the image noise has less colour speckling than with other cameras. The second is that X-Trans cameras also tend to produce good-looking black and white images. 
  • Compared to the Bayer pattern, there is at least some green, red and blue pixel on every row and column of the sensor, meaning that the sensor output can be tuned to produce less false-colour detail (moire), This allows the 16mp sensor to shed the anti-aliasing filter, whereas the conventional implementation of the base sensor keeps the filter (Nikon D7000, etc.)
  • The disadvantage of the X-Trans sensor is that the red and green pixels are physically spaced further apart than on the Bayer pattern. This means that the X-Trans sensor is worse at producing fine detail in image sections that have higher amounts of red and blue information (fabrics, flower petals etc.)
  • Another consequence of the more sparse spacing of the red and blue pixels is that the sensor can at times produce false detail artifacts on intricate subject patterns.

Bayer Pattern Colour Filter Array
X-Trans Colour Filter Array

However, the X-trans sensor was first launched in 2012. Undoubtedly Fujifilm has refined the concept since its launch, but as of 2015 it now competes against the likes of the Nikon D7200, which has more pixels and does not have an anti-aliasing filter. Recall that one of the main benefits of the X-Trans sensors is that it can drop the anti-aliasing filter at 16mp in order to produce sharp-looking images. The D7200 (like the D7100 before it) take this advantage away. The question then: If the X-Trans sensor was able to make 16mp overachieve, is it still competitive with 24mp on a more level playing field?

Image Noise and Subjective Quality



The following is an ad hoc demonstration of the camera's JPEG output quality through the ISO range. Pay attention to the legibility of the soda bottles for detail retention (or lack thereof) as the ISO value rises, speckling in the broad colour patches for overall image noise, as well as the harshness of the reflections and shadows as dynamic range decreases. These samples are not directly comparable to similar samples found elsewhere in this blog because of variable ambient lighting conditions.

Naturally, there is more processing leeway in the original RAW files, but comparing RAW files from different cameras can give one a false sense of fairness. Because each RAW format is different, the best and most representative results depend on each file being processed for best results. Often, different RAW files are compared at default converter settings, which is a bit like comparing default JPEG settings. Hence, for quick comparative purposes we are using the JPEG files out of the camera, as they can give insight into how the competing imaging engineers view the output of their respective cameras.This is especially important when comparing X-Trans files to other RAW formats; going by the default conversion setting in Adobe et al will not give the best renderings for each camera.

One thing to note: Fujifilm cameras tend to meter exposure differently from other manufacturers. You can see that at the same settings,  the X-Trans image is darker than the Nikon image. Some of this is due to how the Fujifilm tone-curve is set up, but the difference in exposure metering is consistent across all of the X-mount cameras. In other words, the stated ISO is not as sensitive to light on the X-T1 as it is on the D7200 in this case. While this is within the bounds of how ISO can be defined, some see this as a means of making the low-light quality of the Fujifilm cameras "feel" better than it is... in other words, you will get clean results at ISO 6400, but the equivalent setting on another camera would be closer to ISO 4000 or lower.

Click on images for 100% crop view.

ISO 200, Native Size (Note: File is mislabelled, this is a D7200)
ISO 6400, Native Size (Note: File is mislabelled, this is a D7200)

At the native size, the Fujifilm output subjectively does very well against the Nikon. Edges are crisp and the image noise has less colour speckling than with the D7200. However, the X-T1 is producing a darker image (you can see more detail in the black of the refrigerator frame on the Nikon). However, even if the edge acuity isn't as crisp in the D7200 sample, there is more detail being resolved by virtue of there being more pixels. If you downsample the Nikon to 16mp to get the same image magnification as the Fujifilm, you get this:

ISO 200, Normalized to 16mp
ISO 6400, Normalized to 16mp

By putting 24mp worth of detail into a 16mp space, you condense image information and make it more "dense". In other words, if you have more information in a smaller space, you have more detail, which you can see as the general overall bump in crispness in the D7200 image. Recall that X-Trans is not as good as the Bayer pattern at intricate red and blue detail... you can see this in the labelling of the pop bottles where the lettering doesn't jump out as much. Except for possibly the Ricoh GR carefully processed, the X-T1 is producing results that are class leading for 16mp APS-C sensors in this instance.

Regardless, this is a good showing for the X-T1; even with the pixel deficit it does a good job keeping with the Nikon, which has the benefit of more pixels and the virtue of being newer on the market. Remember, this is a 100% crop and therefore a small portion of the total image. You likely won't be looking for this much detail in so small a space for most applications. Also recall that if instead of using the same exposure settings and if each camera were allowed to meter the scene as it saw fit, the Fujifilm image would look brighter and punchier.

Concluding Thoughts


Based on image quality alone, the Nikon D7200 likely produces the best results of any APS-C camera as of 2015. However, image quality does not a camera make.While it is true that a DSLR system is more versatile than a mirrorless system by the sheer size of user base and the history of the legacy lenses, Fujifilm has made great strides in developing an appropriately matched system of lenses for its X-system cameras. If anything, the X-Mount is better and more complete than Nikon's offerings if you only look at the APS-C DX lenses. Nikon simple has no answer to lenses like the XF 23mm f/1.4R or the XF 50-140mm f/2.8R

In a more practical sense, the lesser amount of pixels of the X-Trans sensor is not a meaningful deterrent for the majority of people. You can still produce great-looking printed images at 11" by 14", even at the 19" by 14" size with 16mp.  What matters more is how the camera fits into your shooting needs, and for that, even though the X-T1 is Fufjifilm's closest offering to what a mirrorless DSLR would be, it is still quite different and suited to a different type of photography. Ideally, one would have both with an unlimited budget.




With thanks to Broadway Camera



2 comments:

  1. I'm a little confused...The first set of comparative images of the soda cooler are labeled, "Fujifilm X-T1" and "Nikon D750." Is this possibly a misprint or am I missing something? The article was comparing the X-T1 to a D7200 NOT a D750. Anyway, great article and thanks for the effort to conduct it.

    ReplyDelete
  2. This is definitely a misprint (fixed). Back when I was working on this I was working on multiple projects at the same time.

    ReplyDelete