Review:http://www.martinzimelka.com/homepage/Blog/Blog.html

NAVIGATION:

The Image and Image Editor:

One of the biggest shortcomings of the X-Pro 1, in my opinion, is it’s new X-Trans sensor. The justification of which is to increase the randomisation of pixels in order to combat Moiré, thus also eliminating the need for an anti-aliasing filter. Fujifilm also claims that the sensor fully exploits the exceptional performance of the XF lenses, providing new heights in resolution quality.


I must say that it sounds very exotic, and no doubt that it is, but it translates into a less exotic user experience for me. The biggest point of contention is that I fail to see good detail in the RAW files when using LR4. LR4 is my RAW converter of choice due to practicality and speed. Unfortunately, at higher magnifications, LR4 conversions have a mottled/painterly feel to them, surrounded by colour artefacts and chroma smearing.

RAW conversions with RPP are much more detailed than LR4, but colours are often unpleasant, and detail ends up looking messy with colour artefacts as well. The best samples are indeed from the camera’s JPG output and in second place Silkypix conversions, but these also have a subtle mottled and painterly feel, which isn’t always very attractive. (See image on right which illustrates this look)


Best interpolation happens if you can sample from pixels close by, and on the X-Trans sensor, pixels are too spread out, unevenly spaced and not aligned! At the end of the day, the RAW files need quite complicated demosaicing algorithms to convert the files into images, and these algorithms require more processing power than those for conventional bayer arrays. Quite obviously, shooting in RAW requires longer post-processing times, and RAW converters consequently make compromised decisions regarding speed over quality.


(See the diagram below illustrating the differences between a conventional Bayer array sensor and the Fuji X-Trans CMOS sensor)

Difficulties aside, the X-Pro 1 is capable of delivering very good image quality.

The workable dynamic range that can be extracted from the RAW files is very satisfactory. To be honest, it’s really good, and should be more than sufficient for whatever scene you may come across.


The shadows allow for a lot of recovery, and I haven't noticed any odd pattern noise whilst pushing the shadows of my images. Highlights offer a fair amount of recovery, but nothing out of the ordinary. I was pleased with how clean colour transitions were kept between highlights and clipped highlights. This is not the case with most digital cameras.


The 24 colour chart calibration profile didn’t look all too different from the standard Adobe profile, and for those interested, the profile can be downloaded HERE. The profile was created using Adobe’s DNG Profile Editor rather that with the XRITE’s software, the latter having the annoying habit of over saturating colours and having no influence over the tone curve.

Below are image comparisons that show the different RAW conversions of LR4, RPP and the X-Pro 1 JPG output.

The LR4 and RPP were converted with zero sharpening, but a 80% 0,3 rad Smart Sharpen was applied in CS5 to attempt matching the “Normal” sharpening applied with the in-camera conversion. In the images below ACR may compare favourably to the X-Pro1 Jpg, but in many cases fine details are surrounded by colour artefacts which the X-Pro1 Jpg doesn’t show.


(The Silky Pix software that came with the camera was deleted very soon after I installed it. I didn’t like the software at all, and I wasn’t prepared to familiarise myself with it for the short loan period)

Fujifilm claim that moiré is eliminated with the use of the X-Trans sensor, but this is simply not the case. Yes, it does reduce it to a large extent, but I’ve come across moiré artefacts in my images on rare occasions. Moiré is not all that tricky to overcome in the field, and reducing the visual impact of it in post-processing has become easier and easier these days. I would therefore have preferred the X-Pro 1 to have a CMOS sensor with a regular Bayer array, which would make post processing not only less time consuming, but it would probably have made the camera a little snappier too. The X-Pro1 RAW files have been very frustrating to edit in Lightroom 4.1 (ACR 7.1)


Below are some time based figures of how long it takes my Macbook Pro to render six (6) 1:1 previews in LR4 from RAW files with only default Lightroom settings applied:


Nikon D700:           25 seconds

Nikon D7000:         45 seconds

Nikon D800:           109 seconds

Fujifilm X-Pro 1:     115 seconds


It’s fairly evident to see the impact of the X-trans sensor, especially when compared to a camera with a comparable sensor size and resolution, ie the Nikon D7000.


Below are a few sample images that are linked to larger versions. These larger versions also show pixel-level crops to illustrate finer details of the image.

The X-Pro 1 has fantastic JPG output, and in my opinion, it’s the best I’ve seen with any camera I’ve tried. Most of the out-of-camera JPG files are close to perfect, and I would consider shooting the camera in JPG mode if I had to. The JPG engine falls a little short of stellar at higher ISO values and/or long exposures. In these circumstances, LR 4.1 did render better images from RAW files, and in addition, all the odd/lazy/dead pixels were removed from the long exposures. I’m not particularly fond of LeNR on digital cameras, so I avoided using it on the X-Pro 1 during my loan period.