5
Oct
2013

X-A1 vs. X-M1: the Shootout

 photo mastering_zps877477b0.png

 photo X-A1_zpsc7ed59fb.png
 photo X-M1_zpsa401e256.png

Mastering the X-PRO1 and X-E1: book here and Kindle edition (AVAILABLE) here)

Fujifilm X-A1 in stock check

USA: AmazonUS (black in stock)/ eBay / BHphoto (blue in stock)/ Adorama (black in stock)/ Pictureline / DigitalRev EUROPE: AmazonUK (via DR) DigitalRev / eBay

Fujifilm X-M1

USA: AmazonUS (save $72 on the silver version + kit lens) / BHphoto / Adorama / PicturelineeBay ($70 price drop) / DigitalRev EUROPE: eBay / DigitalRev / AmazonDE (via DR) / wexcamerasDE / AmazonUK (via DR) / PCHstore

_ _ _

by Rico Pfirstinger

Talk to Rico (open forum for questions & feedback)Rico’s Flickr sets – X-A1 sample images set – X-M1 sample images set –  X-M1 vs. X-A1 shooutout set (private Flickr set, MUST use this link!)Original X-A1 & X-M1 shootout RAW files for download (incl. Lightroom XMP files) – Mastering the Fujifilm X-Pro1 reading samples (65 free pages) – Pre-order my very soon to be released NEW book: Mastering the Fujifilm X-E1 and X-Pro1

Edit: I have updated the shootout with Photo Ninja samples, please read this separate article.

Welcome to this weekend edition of X-Pert Corner! Comparing the X-A1 and X-M1 is pretty popular these days. Since the hardware of both cameras is pretty much identical, such comparisons usually boil down to an X-Trans vs. Bayer color filter array (CFA) pissing contest. Since we are on the Internet, it is a fact of life that much of what has been and will be written on this issue is inaccurate or just plain wrong, mostly due to flawed methodology. For example: Instead of comparing the output of different sensor CFAs, many “reviewers” are simply showing differences between the two cameras’ JPEG engine default settings.

DSCF2914 – Lightroom 5.2
X-M1

There are indeed visible differences between the X-A1′s and X-M1′s default JPEG processing: The X-A1 is targeting entry-level customers who prefer sharp, clean, brilliant, contrasty and noise-free JPEGs straight out of the box (at least that’s what Fuji thinks), and this preference is mirrored by the camera’s default JPEG engine calibration at its factory settings. In other words: With everything set to factory defaults, the X-A1 delivers sharper (aka more sharpened) results with a bit more contrast and stronger noise reduction than the X-M1. It is however possible (and advisable) to change the default JPEG settings in both cameras to achieve different results that are much harder to distinguish. This is where things become more interesting, but this is also where most “expert comparisons” come to a frustrating halt.

DSCF0991 – Lightroom 5.2
X-A1

In any real-world scenario, it is impossible to compare the performance of different sensor CFAs without processing the RAW images on which we want to base such a comparison. The processing can either be performed in-camera or with an external RAW converter. It is very difficult to remove this processing part from any sensor comparison equation, because different CFAs require different processing in order to achieve the same or at least similar results. Using identical RAW converter settings on comparable images from the X-A1 and X-M1 won’t do the job, quite to the contrary. In order to get comparable results, one has to apply different parameter settings that correspond to the unique qualities and properties if each sensor’s CFA.

DSCF0283 – Lightroom 5.2
X-A1

Please forget comparisons that are based on factory default settings of cameras and RAW converters. They are a waste of our time, because they won’t tell us how each camera performs with optimized real-world settings that most of us would choose to achieve specific precessing results, such as revealing “maximum detail”.

The Setup

In order to compare both cameras (and CFAs), I shot several different subjects in manual mode, each with exactly the same exposure settings and the same lens. This setup made sure that each camera’s sensor was exposed with the same amount of light, at least in theory. Practically, there may still be subtle brightness differences due to the fact that corresponding shots had to be taken in sequence (exchanging cameras and the lens on a tripod and reframing the shot could take a minute or two).

DSCF1141 – Silkypix 5
X-M1

Using this setup, I found that my X-A1 production camera produced slightly darker images than my pre-production X-M1. The difference (somewhere around 0.1 EV) is within specifications and could easily be adjusted by moving the exposure slider in the RAW converter. By the way: The reason why I couldn’t get my hands on an X-M1 production sample for this shootout is that Fujifilm Germany has 250 of these cameras tied up in their X-M1 roadshow workshops held in Germany, Austria and the Netherlands.

DSCF1100 – Lightroom 5.2
X-A1

In order to achieve optimal input quality, all four sample subjects were shot with optically corrected Fujinon XF14mm and XF23mm prime lenses. I understand that it’s not very realistic to use lenses that cost about twice as much as the X-A1 camera body, but using a less expensive kit zoom lens could at least theoretically obscure some image detail.

The fact that I actually had to use expensive primes is a testament to the similarity of both cameras’ output quality. This is major pixel peeping territory. While X-Trans output certainly looks different from Bayer output at 100% magnification, I am unable to say which CFA’s output looks better. It may well depend on the specific subject and can quickly boil down to your personal tastes and RAW processing skills. Of course, it will also make a difference which RAW conversion software you are using to process a particular image file. In my opinion, these factors are more important in determining the final quality of processed images than a sensor’s CFA, but hey, that’s just me trying to inject some sanity in this often quite emotional discussion.

DSCF2116
X-M1

That said, there’s no reason for you to trust my opinions or processing skills. Please make up your own mind! Click here to access a Dropbox folder containing the RAW files that I used for this little shootout, complete with the respective XMP files, so you can open the RAWs in Lightroom with the settings I have used to process each image. I am sure that you will soon find different settings that you like better.

DSCF1960
X-M1

Click here to access a private Flickr set with SOOC JPEGs and processed Lightroom versions of these RAWs. In Flickr, you can inspect and download full-size (100%) images: Just pick “Original” as a viewing option when right-clicking on an image. You can also look at the EXIF information by clicking on the “show more” link next to the “Additional info” text below each Flickr image.

DSCF0882 – Lightroom 5.2
X-A1

The SOOC JPEGs in the Flickr set come in two versions: Version 1 (named “JPEG standard” in the respective image titles) are renderings using each camera’s default settings for sharpness and noise reduction. Version 2 (named “JPEG custom” in the respective image titles) are custom renderings with noise reduction set to -2 for the X-A1 and X-M1, plus an additional sharpness setting of +1 only for the X-M1. Setting the in-camera noise reduction to a minimum reveals and preserves as much detail as possible, while the additional sharpening for the X-M1 partly compensates for the stronger default sharpening of the X-A1′s JPEG engine.

The Shootout

Our first test subject is a common scenario with plenty of dynamic range (the cameras were set to DR200% and hence ISO 400) and an abundance of detail:


Camera A


Camera B

Click on one of the images to access the Flickr set with several full-size versions of this shot.

First observation: The white balance tint of X-M1 files is always leaning more towards magenta than the white balance tint setting of corresponding X-A1 files. I assume this is to compensate for the relative overweight of green pixels in the X-Trans CFA compared to the Bayer CFA. That’s why you should never compare X-A1 and X-M1 images by choosing identical white balance tint settings in a RAW converter. You may well choose the same color temperature, but do keep the differences in the tint settings. They are there for a reason.

Second observation: The default colors of the X-A1 and X-M1 are similar, but not identical. This applies to both the internal JPEG engine and Adobe Lightroom. It may also apply to other RAW converters (at the time of this writing, only Lightroom/ACR offered support for both camera models), but it’s eventually up to the makers of each RAW converter to determine the default color gradation when importing RAW files of a specific camera model. As for Lightroom, it’s possible to achieve a pretty accurate color match between both cameras by properly adjusting the Camera Calibration sliders.

Third observation: Since in-camera noise reduction appears to work selectively, dark areas (shadows) lose significantly more detail than midtones and highlights. This effect is particularly pronounced in the X-A1. Using Lightroom, I feel that I was able to extract more (shadow) detail from both sensors than with the cameras’ respective built-in JPEG engines.


X-M1 vs. X-A1 Lightroom 5.2 (click on the image for a 100% view)

So again, not the sensor CFA but the processing is the major contributing factor for revealing maximum detail and eliminating noise. It’s an open secret that X-Trans RAWs require “different” RAW processing parameters, especially in Lightroom/ACR. For example, X-Trans needs less chroma noise reduction than Bayer, so Lightroom’s default color noise reduction setting of 25 is usually overkill. Depending on your subject and your ISO settings, a value of 0, 5 or 10 will usually do the job (and reveal more detail in the process).

In a similar fashion, X-Trans cameras require less luminance noise reduction. Luckily, Lightroom’s default setting for luminance noise reduction is zero, and you can usually keep it like that even up to ISO 6400. Not so with Bayer RAWs, where high-ISO shots will often require at least some luminance noise reduction in order to look pleasant.

Another frequently discussed issue is detail sharpening. With X-Trans, I usually crank up the detail slider (sometimes to the maximum), but I am quite conservative using the sharpness slider. With Bayer, my sharpening approach is more conventional.

Of course, these are just a few examples telling you how I like to process my images, so feel free to think that I’m an idiot. You may find completely different settings far more suitable. I have no problem with that, so let’s not even consider arguing what settings are “the best”. Simply pick the ones that work for you. Just be aware of the fact that identical settings will never result in identical looking results when applied to an X-Trans and a Bayer sensor version of the otherwise exactly same shot.

This brings us to the topic of using either the X-M1 or the X-A1 as an add-on body to an existing X-Trans camera like the X-E1 or X-Pro1. Obviously, this is easier with an X-M1, because to a very large degree, you will be able to apply the same (or very similar) RAW converter settings in order to get similar results. With the X-A1, you may still be able to get great looking results, but you should be aware of the fact that in order to get there, you’ll have to use different settings.

Let’s have a look at our second sample:


Camera A


Camera B

Once again, I chose a subject with plenty of dynamic range in order to preempt any rumors that X-Trans offered more (or less) dynamic range than Bayer. This time, I was able to shoot at base ISO 200 (DR100%) and still capture the subject’s tonal range. Click on any of the two images to access the Flickr set showing different JPEG and Lightroom versions of this shot in full-size views.

First observation: Even at base ISO, the default noise reduction of both cameras’ JPEG engines is already kicking in and smoothing shadow areas. This is mitigated in the custom JPEGs, and my custom-processed Lightroom versions appear to reveal the most shadow detail.

Second observation: You will notice the different handling of certain (but not all) kinds of foliage, not only in the Lightroom versions, but also in the SOOC JPEG versions. Using a different converter like Iridient, Aperture, Silkypix or Capture One can lead to different (more Bayer-like) results.


X-M1 vs. X-A1 SOOC JPEG custom (click on the image for a 100% view)

Our third sample is a pretty much “impossible” subject with huge dynamic range, large of out-of-focus areas and difficult shadow detail thanks to DR400% (and hence ISO 800).


Camera A


Camera B

Both cameras’ built-in JPEG engines smear DR400% shadow details. Applying less noise reduction helps, but once again, only the external RAW processing reveals maximum detail.


X-M1 vs. X-A1 Lightroom 5.2 (click on the image for a 100% view)

To maintain comparable noise levels, I had to apply some luminance noise reduction to the X-A1 file in Lightroom. No luminance noise reduction was required for the X-Trans version from the X-M1.

Our fourth and final subject illustrates how the JPEG engine of the X-A1 works differently from the engine in the X-M1. This is also our high-ISO test, as I shot the same subject with ISOs 800, 3200 and 6400 (using DR400% to capture the scene’s dynamic range).


Camera A


Camera B

The X-A1′s JPEG engine appears to work like the popular Topaz Clean plug-in: Processing noisy areas (like walls or ceilings), is will smoothen everything below a certain threshold and sharpen everything above it. This leads to pretty clean out-of-camera JPEGs, but is also results in a dramatic loss of fine detail. Have a look at this direct comparison at ISO 800:


X-M1 vs. X-A1 SOOC JPEG standard ISO 800 (click on the image for a 100% view)

To view additional direct comparisons at higher ISO levels and with different JPEG noise reduction settings, please visit my private Flickr set.

I would have loved to include Lightroom versions of this final series, but as it turns out, Adobe has messed up its RAW support for the X-M1 in a very peculiar way, importing high-ISO DR400% RAWs with a wrongly calibrated digital gain, over-amplifying the ISO 6400 shot one full stop in the process. Ironically, the newer X-A1 is perfectly supported.

That said, feel free to download the RAWs from my Dropbox and process them with any RAW converter of your choice. If you insist in using Lightroom, just make sure to manually correct the exposure of each important X-M1 sample to match the brightness of the corresponding X-A1 image sample.

Conclusion

The different results you get from either the X-A1 or X-M1 depend more on the settings of the JPEG engine (or the external RAW converter) than on the sensor CFA or leaving out the anti-alias (AA) filter. While lab tests and test charts will certainly prove a resolution and noise advantage for the AA filterless X-Trans sensor, carefully processed real-world shots may paint a different picture. At views below 100%, properly processed samples from both camera will be almost indistinguishable. In fact, if you are looking for identical results from both cameras/sensors, the most challenging part may be perfectly matching the color gradation rather than matching detail, sharpness and noise response.

As I already pointed out in my previous column, the X-A1 appears to the better deal due to its price advantage, at least in an isolated comparison. However, if you are looking for a second body as an addition to your existing X-Pro1 or X-E1, you may be better off getting an X-M1. Don’t underestimate the convenience of an X-Trans-only workflow.

What’s next?

I have just received word from my publisher that the shipment of my new book “Mastering the X-E1 and X-Pro1” has been received at the warehouse. I expect pre-orders to be fulfilled as early as next week. Thank you all for your continued support!

The next edition of this column should be up in the wee hours of 18OCT, covering another second-generation X-Trans sensor camera. That’s when I am also going to announce my next book project here on Fujirumors. Stay tuned, and don’t spend too much time with pixel peeping. Also look at the big picture!

Fujifilm X-A1 in stock check

USA: AmazonUS (black in stock)/ eBay / BHphoto (blue in stock)/ Adorama (black in stock)/ Pictureline / DigitalRev EUROPE: AmazonUK (via DR) DigitalRev / eBay

Fujifilm X-M1

USA: AmazonUS (save $72 on the silver version + kit lens) / BHphoto / Adorama / PicturelineeBay ($70 price drop) / DigitalRev EUROPE: eBay / DigitalRev / AmazonDE (via DR) / wexcamerasDE / AmazonUK (via DR) / PCHstore

For your convenience, here’s a TOC with links to my previous X-PERT CORNER articles:

Rico Pfirstinger studied communications and has been working as journalist, publicist, and photographer since the mid-80s. He has written a number of books on topics as diverse as Adobe PageMaker and sled dogs, and produced a beautiful book of photographs titled Huskies in Action (German version). He has spent time working as the head of a department with the German Burda-Publishing Company and served as chief editor for a winter sports website. After eight years as a freelance film critic and entertainment writer in Los Angeles, Rico now lives in Germany and devotes his time to digital photography and compact camera systems. His book “Mastering the FUJIFILM X-Pro1” (Kindle Edition) (Apple iBook Store) (German version) is available on Amazon and offers a plethora of tips, secrets and background information on successfully using Fuji’s X-Pro1 and X-E1 system cameras, lenses and key accessories.

  • Renato S.

    The part that takes the differences into consideration is what I liked the most, it seems trivial and obvious but I’m pretty sure almost no reviewer will be doing this when comparing these two. Actually a lot of reviewers don’t do that when comparing X-Trans files to other cameras. They try to make things equal but they fail to understand that different CFA requires different approaches and that making that part equal is not actually equal.

    I’m looking forward to see what will Kaizen bring with the next announcement. Each new camera announcement usually brings some news feats that were brought through user’s feedback and I’m really looking forward to see Fuji’s improvements.

    • http://www.fujixspot.com/f34/ Rico Pfirstinger

      Fuji has indeed fixed or improved many points on our X-Pro1 and X-E1 feedback lists, so I think you will be pleased.

  • David Whittley

    Where do I sign up for the X-trans Lightroom 5.2 master class.
    I have a X100s and cannot seem to get the results you have achieved with these images. Not with regard to developing foliage at any rate. Trees at 100% (and sometimes farther away) resemble seaweed or melted wax!
    What am I doing wrong?

    • http://www.fujixspot.com/f34/ Rico Pfirstinger

      This is exactly why I included the XMP files in my Dropbox download folder.

      • David Whittley

        Thanks Rico
        I’d missed that.

    • Snikkel

      You can’t get the detail you want from this sensor. You either need a Sigma DP Merrill or a Sony NEX-7.

  • Donsantos

    No 3200 iso comparision?

    • http://www.fujixspot.com/f34/ Rico Pfirstinger

      The ISO series contains RAWs and processed JPEGs of the test subject in ISOs 800, 3200 and 6400 (this is also mentioned in the article).

      • Donsantos

        Rico!!!

        Thanks for the raws! On the 3200 image in raw I prefered the sharpness color and noise of the x-a1. (This was also a blind test because I didn’t even look at the exifs in which was which)

  • Mrz

    So bayer = x-trans, but bayer – more noise and more resolution, x-trans – less noise and less resolution.

    • http://www.fujixspot.com/f34/ Rico Pfirstinger

      No, X-Trans has better resolution, but Bayer looks sharper SOOC due to more sharpening. In Lightroom, the additional luminance noise of the Bayer gives the impression of (false) detail. In any case, download the RAWs and process them yourself with the converter of your choice. Iridient will soon support the X-A1, for example.

  • jonnie

    Thanks for the comparison and your hardwork in bringing this to us. I really how you processed the images to get maximum quality from each file instead of using default settings. As far as results I would be more than happy with either camera, for me it comes down to convenience of a single workflow. Cheers jon

  • Ros

    very interesting, thanks!

  • G.

    …now we know that xtrans sensor is only marketing…as I always state…and to my eyes bayer sensor is better…

    • Andrew Williams

      You should try to read the article. It is really most interesting and absolutely cannot be summed up in one sentence.

      • http://deleted G.

        …conclusion:”…. At views below 100%, properly processed samples from both camera will be almost indistinguishable. In fact, if you are looking for identical results from both cameras/sensors, the most challenging part may be perfectly matching the color gradation rather than matching detail, sharpness and noise response.”
        You are so right Andrew Williams…:)

  • otaki

    It is not hat easy, unfortunatelly Rico choosed Lightroom, wich offers the worst of all trans demosaicing.
    Silkypix er recently Photo Ninja are creating significantly Bettler trans results.

    • http://www.fujixspot.com/f34/ Rico Pfirstinger

      This is basically a comparison of the JPEG engines and Lightroom, but since I made the RAWs avilable, everybody can reprocess the samples with any converter of their pleasing. However, I expect at least 90-95% of the target audience of the A1/M1 to use JPEGs from the camera or Lightroom/ACR.

    • Bemused

      Oh please, it’s hardly the worst.

  • Milan

    Thanks for the very informative comparison. I think that the conclusion is quite easy: the performance of the X-Trans sensor is close enough to that of a Bayer sensor to make it almost indistinguishable in the real world. Hardcore pixel peepers might see the slight loss of resolution, but that’s really irrelevant for photography.

    What I always wondered is why the marketing department of Fuji communicates so poorly with the technical department. Why on earth are they trying to sell (and insist ad nauseam) on the much superior (30%?) resolution advantage of the X-trans sensor and that it matches a Bayer Full Frame sensor in resolution and noise??? Why this silly lie??

    From a technical point of view, the X-trans sensor is designed (and ONLY designed) to avoid moiré. And at that, it delivers quite well. Of course, you have to pay a price (the very slight loss of resolution/accuracy being the least important of all):

    - You will need a more powerful processor to achieve the same performance (speed) as you would with a Bayer sensor (this was clearly stated when the X-Pro1 was released: they admitted they had to develop a more powerful processor because of the X-trans CFA), while using more battery for it.
    - Most important, the RAW support has been a disaster for over a year and the biggest problem for real photographers wanting to use these wonderful cameras but having to change their workflow (or give it up ans stick to JPEGs) in order to do so.

    So the final question would be: is moiré such a big issue in real world photography as to go through these pains in order to reduce its incidence? My own answer is NO. That’s why I think that Fuji shot in its own foot with all this X-Trans affair. Should they have used the sensor on the X-A1 from day 1, the X-System would have been a much better alternative for any photographer/enthusiast.

    But marketing won.

    • G.

      it’s difficult not to be agree with you…but fuji fan boys believe only in Fuji God!

    • http://www.fujixspot.com/f34/ Rico Pfirstinger

      X-Trans also results in less luminance and chroma noise, which is clearly reflected by lab tests and required LR settings. However, is also has its disadvantages.

      This is like odering the otherwise exact same SUV mith either AWD or 2WD. The maker of the car will tell you the advantages of AWD, how great it is in snow and off-road, but he will not tell you about the extra cost, the higher fuel-burn, the added weight and the reduced acceleration and top speed. Most drivers in a city would never or rarely need the AWD, but they would experience its drawbacks every day. And yet, AWD is very popular and people pay the extra price.

      • G.

        From Dpreview:”…Using our standard processing, with Adobe Camera Raw’s noise reduction set to a minimum, the X-E1′s images look rather different to those from Bayer sensor cameras, and specifically show much lower chroma noise. This means that direct comparisons have to be treated with caution; in effect the demosaicing required by the X-Trans CMOS sensor includes a degree of chroma NR…”
        That’s why x trans files look cleaner….so again and again x trans=marketing!

        • http://www.fujixspot.com/f34/ Rico Pfirstinger

          It looks cleaner because it’s cleaner, and still the details are there. Btw, DPR folks aren’t infallible, they are only people, too, (and at least some of them seriously flawed ones considering the major censorship campaign that’s going on in their forums). In any case, here’s what DPR really wrote:

          “As with the the X-Pro1 Adobe Camera Raw’s processing of the X-M1′s files exhibit remarkably low noise figures, when compared with its Bayer sensor rivals. Chroma noise is strikingly low, and detail retention is impressively high – very much like the camera’s JPEGs, in fact. Because of this, direct comparisons have to be treated with a degree of caution – it’s best to assume that the de-mosaicing process of the X-Trans CMOS sensor behaves as though it’s doing substantial chroma noise reduction relative to ACR’s standard treatment of Bayer sensors.”

          So the de-mosaicing of the X-Trans sensor behaves AS THOUGH it’s doing chroma noise reduction. You can always use different converters that don’t do any NR at all, you will still see major differences between Bayer and X-Trans noise.

          • Robert

            This is exactly what I find as the best thing with X-Trans, i.e. that there is less croma noise. I tend to like the SOOC jpegs from the X-E1 with standard NR up to ISO 3200 (above that I see too much noise and some color desaturation).

            Generally I appreciate a mix between keeping detail and getting some noise that goes in the noise reduction direction. The X-M1 SOOC standard jpegs suits my taste, while the X-A1 SOOC jpegs looks as if they have too much NR and sharpening applied.

            I can live with some luminance noise (guess I am used to it from the film days) but do not want any croma noise, which I find very ugly. Therefore I prefer the X-Trans over the Bayer CFA. That together with the best light weight optics available (mounted on a light weight body with viewfinder) and a beautiful image rendering were my reasons to select Fuji X-E1 + 18-55 for a truly portable camera that at the same time has excellent IQ.

          • Robert

            I should mention that on the X-E1 i use saturation (called color in the Q menu) +1 for SOOC jpegs. I think that the colors look a bit flat with saturation 0. On the X100S that is not necessary, likely because Fuji tuned this in the second generation JPEG engine.

          • G.

            I posted what dpreview wrote about xe1(same sensor of xm-1) page 20 of xe-1 review.
            If I run xtrans files via PN I see more details and more noise too, again x trans do not do a better work then Bayer sensor but noticeable worse in the highlights…where x trans really suck.

          • http://www.fujixspot.com/f34/ Rico Pfirstinger

            There is no highlight performance difference between the different CFAs. If PN “sucks”, it is because PN does not yet support highlight recovery with X-Trans, according to their own release notes. As of now, the PN guys don’t seem to be overly interested in me covering their software, though, and I certainly do not intend to press them. Maybe they change their mind once PN offers full support for X-Trans, I’ll happily wait for them to contact me once they feel PN is up to the task of being evaluated her on X-Pert Corner.

          • MJr

            Some people just don’t want to hear it Rico, no matter how many ways you explain it. Like right-wing politics, immune to reason.

          • http://www.picturecode.com picturecode

            Hi Rico, absolutely, we’re interested to have you cover Photo Ninja. Also, I’d like to understand where the disconnect occurred that caused you to think otherwise. (Did we miss an email? Or did the person who replied mishandle it? Or something else?)

            Please send me a PM or email and let me know how I can help.

            Regards,
            Jim Christian
            Founder, PictureCode

          • G.

            I expressed myself badly…I mean the problem with the highlight is with all the program not only with PN ( …by the way i think this is a great program..)…x trans sensor is more prone to blow highlight then the Bayer sensor..with conventional sensor you can recover higlight much much better…..

          • http://www.fujixspot.com/f34/ Rico Pfirstinger

            Thanks, Jim, you’ve got mail.

            As for highlight recovers, I totally cannot confirm that, actually, I would flatly deny it based on my extensive experience with all X-Trans cameras and 8 different RAW converters. You would have to prove your assertion with identically exposed RAWs from an A1 and an M1 that show that the A1 offers more highlight recovery headroom. I’d be happy to learn and try to keep an open mind, but this claim needs proof until I even think about buying it.

  • otaki

    The advantage sold to us was that the AA filter can be obsolete with transx, i wonder now that the bayer with AA delivers same sharpness with aa filter
    from algorithm side noise is better to eliminate considering the transx bigger pixel array.
    Rico should repeat the comparison using one of the better converters such as photo ninja, silkypix, capture one.
    LIGHTROOM hasnt developed the heart of their rawconverter – the demosaicing – color separation – modern adaptive denoising – for years now and the transx development is for sure the weakest on the market.

    • Gab

      The X-A1 seems to have a weak AA filter, which means moire will be a problem with sharp lenses. (the new kit lens is so so, it could be moire free)

    • http://www.fujixspot.com/f34/ Rico Pfirstinger

      Why don’t you (and I mean all of you) repeat the comparison with the converters of your choice once they support the A1? It can’t be long, and wouldn’t it be great to see what other users can come up with? The RAWs are also a great test to see how well different converters deal with the DR function. As we have seen, Adobe is totally messing up the digital gain of some X-M1 high-ISO samples. So do/did other converters, like the early Capture One, and it did not take long until some idi… I mean expert blogger blamed it on Fuji and them “inflating ISO”, causing a medium sized shitstorm. I fondly remember this incident, and I also remember that nobody cared to listen to my injunction that this expert blogger simply measured the calibration (or rather miscalibration) of his RAW converter (which was a beta release featuring preliminary(!) X-Trans support).

  • Max

    Rico, thanks for sharing this GREAT post! I have a X-M1 and I’m rather disappointed of the IQ. Maybe my XTRANS-expectations where too high… Especially the high-ISO performance seems to be poor. Nice to see that your comparison also somehow disenchants the XTRANS technology. Additionally, I have the impression, that its hard to find good Lightroom adjustments which outperform the results of the in-camara raw converter. Would you be willing to share your X-M1 as well as Lightroom presets which could serve as all-day starting point for fine adjustments?

    • http://www.fujixspot.com/f34/ Rico Pfirstinger

      I always start with factory settings. I do not have a single camera-specific preset for any RAW converter.

  • David Whittley

    I wonder just hdw damaging this comparison will end up being for the x-trans. Fujifilm will need to have a really good argument to convince people now that the x-trans is worth the trouble.
    What would happen if the X-E2 launched with a choice of bayer filter and no AA filter or x-trans with no AA?

    • http://deleted tim

      Simple, nobody “out there” cares, or understands. They just see 200 less or 200 more, its a nice distraction from the other short comings with the camera’s.

      I don’t care or understand either. If they would sell a camera variant for 200 less, I would take it. If I have to stuff around with (one of ten) RAW converters to get a good image … then the joke would be on me … I guess.

  • me
  • Dima

    Rico, you are very smart and practical mind!
    Thank you for this great comparison and explanation :-)

  • 078Adam

    Rico, thank you for your efforts on this. For some I had no preference, but on the 3 occasions I did, I preferred the A1, I went through a few times and each time picked the same favourites. I only peaked at the info once I’d finished. A little sad as I’ve got an XE1 with XTran. It wouldn’t be sad if I loved the results from the XE1, but I don’t. I love the fact that I’m getting low light high ISO pictures, which otherwise I wouldn’t have got. But in good light I agree with those who call them flat. I hope the XE1s or 2 has more depth to the photos.

    • http://www.fujixspot.com/f34/ Rico Pfirstinger

      New cameras feature the 2nd generation JPEG engine that is also found in the X-M1. Obviously, any X-E2/E1S would also work with 14 bits, like the X100S.

  • http://www.outbackphoto.com Uwe Steinmueller

    If I use PhotoNinja for the X-Trans sensor the details look good to me. PhotoNinja does not seem to support the A1 properly (bad WB)

  • Tv

    Great article – I have just purchased Rico’s new book on Kindle – Highly recommend it for all x-e1 and x-pro1 owners. Can’t wait to see what fuji comes up with this month. The competition is getting even stronger with Sony rumoured to be releasing full frame mirror-less cameras (I hope the lenses are not too big &/or slow) and the new micro 4/3 cameras having great autofocus and LCD refresh rates. The X-pro 2, which i assume will be out next year, will have to be a major step up to keep up with the competition. In the end we consumers are the winners.

  • otaki6

    Photo Ninja Results

    I could not find all of the comparison images in the dropbox folder so i did what i could.

    Basically i see no difference in detail and noise, just the Bayer noise (ISO 800) is smoother while the Transx Noise is rough.
    In fact not much “plus” for the Transx here to see.

    I converted one of the lake pictures of the X-M1 with photo ninja – i could not find the A1 version, so maybe Rico can add my PN Version to his crop comparison.
    13MB so i had to use a file hoster
    http://www.datenkeule.de/dl.php?file=thumb_file1380986982DSCF2932.jpg

    PN can read the A1 files but colors and everything has to be adjusted manually, i tried my best for the tree shots, but it’s not perfect.
    X-M1 (Photoninja)
    http://www.fotos-hochladen.net/uploads/dscf2937v1pn7gf9kc2a68.jpg
    X-A1 (Photoninja)
    http://www.fotos-hochladen.net/uploads/dscf0108v2pnm3jk1cqhd4.jpg

    • http://www.fujixspot.com/f34/ Rico Pfirstinger

      All 12 RAWs are in the Dropbox folder linked above.

  • otaki6

    Sorry, here again the full size link to the lake photo of the X-M1
    http://www.datenkeule.de/dl.php?file=file1380987772DSCF2932.jpg

  • otaki6

    So, last one for today.

    Rico is doing a great job, i could develop the A1 version of the lake photo in PN, but colors are strange.
    Nevertheless it shows that the X-M1 Version shows little more detail, os Rico seems right in what he is telling us.
    X-M1 has little more detail & different noise patterns, at the end there is very little difference, i had wished that Fuji would give us at least a little enclosure advantage for the X-M1 (for 200,- more, some real metal parts) but htey decided to safe production costs by implementing a standard routine.
    Considering the price the X-A1 would be my choice now, considering the transx demosaicing challenge, the X-A1 would be my choice now.
    I bought the X-M1 already.

  • souk1501

    Thank you so much for such thorough review! I do have a question about the X1A OOC JPEG in caparison with X100 OOC JPEG. Many people complaint that the X100s OOC JPEG is too contrasty when compared to X100 and doesn’t have the original film look. Do you find the JPEG from X-1A closer to X100 or X100s (or any other XTrans CFA for that matter)? And if the latter, can it be easily tweak with in-camera settings to get back the X100 look? Thanks!

    • http://www.fujixspot.com/f34/ Rico Pfirstinger

      The A1 engine is even a bit more contrasty SOOC than the M1. This is also mentioned in the article and can be seen in the examples, just compare the histograms if you do not believe your own eyes or your monitor.

      Of course, it is easy to change the settings when so required. For example, the JPEGs with the ISO 200 lake panorama was processed with highlights -1 and shadows -1 in both cameras.

      • Dima

        Rico, thank you for your patience and your education work for Fuji X customers! :-)

        • http://www.fujixspot.com/f34/ Rico Pfirstinger

          Always happy to help!

  • KAKAROTH

    Ive just retourned my xpro1 and xe1 because xtrans is a crap sensor at all! waiting for a real camera and a no toy object… Nex FF on the way and zeiss, real objects!

  • me

    It’s time to say it.
    X-Trans is dead, enough with the pointless comparisons of which PP software loses less details.
    Fuji, go back to your drawing board, give us x-e2s with bayer sensor.

  • http://www.ishootalpha.com B.Allen

    Great job Rico with the comparisons. This really gives potential customers a clear view of the capabilities and limitations of the X-trans v.ii and Bayer sensors. IHMO, I would choose the bayer of the XA-1 (better detail especially in the foliage). I think Fuji shot themselves in the foot on this one and all future X-trans v.ii cameras. They put a better sensor (with respects to IQ) in an entry model camera than a more expensive model. The overall Achilles heel of the X-trans is that lack of a standardized third party raw converter and smearing of greens. Hopefully, Fuji sorts this all out.

    • Denis

      The most optimal color mosaic is bayer … where greens and blues are swapped :)
      Anyway, XTrans is less optimal than traditional bayer, so demosaicing it you either lose colors or resolution compared to bayer.

  • Jonas

    I’m surprised Fujifilm did an X-A1 when the difference is this slim to their pricier X-Trans cameras when comparing the RAW’s. They could have just left the “budget” crowd (relatively speaking…) with the X-M1 and let it stay like that.

    But no — they actually went further and released a camera that is shown to be very controversial due to how little worse the less sophisticated sensor is, and how it even takes the waterpaint foilage out of the equation when it comes to super popular post-processing tools like Lightroom.

    It’s kind of astounding that Fujifilm went there in my opinion. I don’t understand the point of the X-A1 other than to introduce confusion and doubt about their other cameras.

  • ShawnS

    Why are people so down on the X-trans? How long has it been around compared to the Bayer? Give it a couple of years before you can call it dead. It’s the start of a new idea. It will be a while before it is perfected. I want a camera that performs differently so I can take photos that other people can’t.

    • Denis

      http://infoscience.epfl.ch/record/50200/files/ASH05.pdf
      Quote: “This model also shows that the Bayer CFA is the most optimal spatial arrangement of three colors on a square grid”.
      So, it’s not a coincidence there are still no good demosaicing for xtrans. It’s just impossible to match bayer output. In other words: if you go to resolution with XTrans, you’ll end with crappy colors. If you go to colors, you’ll end with tons of details smearing.
      BTW, detail wise bayer+AMAZE demosaicing is far above everything I have seen from xtrans. https://dl.dropboxusercontent.com/u/12196364/gallery/different/boring-comparison.png — NEX-5n on the left, XPro-1 on the right.

      • Waspy

        The Academy of Sciences said also the earth is a disc.

        Open your mind for new insights.
        Believe me, the Bayern pattern will be not the most optimal arrangement in the next future.
        Also not XTrans.
        But it’s the most best newcomer for fresh and new way of thinking.
        The battle is now open for Canon, Nikon, Fuji, Sony for a new and reinvented pattern.

      • Jack

        As I pointed out in the other thread, that article does not say what you think it says. It’s talking about very specific square arrays that existed or were proposed in 2005, not about patterns that were proposed later, and it explicitly states that non-square arrays might be better, especially hexagonal.

        • ShawnS

          Precisely. Let’s see some more innovation. I’m happy to be at the cutting edge, and if that edge isn’t as sharp as we all hope, let’s try something else. I’m not in this for the well trodden path.

  • Alan W

    Just curious—there’s a certain “too sharp” quality that I see in most photos these days. Mine included. I shot film for 35 years or so before pretty much just shooting digital. (Say 99% of the time.) I began to think about the idea of why people consider the A1 sensor “better” than the M1 and a specific aspect of what constitutes a good image began to become more critical in this discernment. I’m wondering to what extent photographers who look at the A1 IQ and prefer it to the M1 are those raised on digital. Please understand that I am not saying that growing up with film as a sine qua non for photographic images is in any way better than pure digital. If I felt that, I’d still have my darkroom. When looking at smaller (not Leaf, etc.) sensor output, I think that the X Trans sensor array has a more film like quality to it. Granted—with a wider gamut, much longer dynamic range, huge adjustment to white balance, etc. that we could only dream about with film. (But all digital has this additional capacity.) I’ve assumed that the more “random” color array was the reason behind this, as Fuji’s marketing department blathered about. I end up liking the quality of the Fuji images quite a lot. Like many others, I was waiting to see what the raws had in them, and I think we’re just beginning to get a better idea of this. It’s still very new.

    • http://www.fujixspot.com/f34/ Rico Pfirstinger

      You are raising an interesting point. In my experience, people often seem to confuse resolution and detail with sharpness. In fact, more sharpness leads to less detail, and resolution has little to do with how sharp an image looks.

      Per default, the X-A1 is oversharpening its output. This can be seen when taking pictures of a newspaper, the letters will show sharpening artefacts. Fuji showed this in a presentation I was able to get my hands on. This also leads to softer out-of-focus areas in X-M1 output, less of the image appears to be in focus, the bokeh begins earlier.

      You can experience a similar effect when you switch between standard sharpening and Pure Detail sharpening in Silkypix. With pure Detail, the parts of the image that appear to be in perfect focus diminish, and the out-of-focus areas grow. At the same time, you can see more fine detail and experience maximum resolution.

  • Alex Dodis

    Rico, do you mean is the “detail” sharpening of SilkyPix correspond to “clarifty” in LightRoom?

    • http://www.fujixspot.com/f34/ Rico Pfirstinger

      Clarity is mostly midtone microcontrast and can give the impression if more or less sharpness. However, this is about applying different sharpening methods (algorithms). Lightroom only offers one, Silkypix offers two to choose from, and they are very different.

  • Waspy

    All this discussion about pixel peeping reminds me of the per default implemented ClearType setting in Win7.
    Someone hate it, someone love it.
    Sharpness is not everthing for overall impression.

  • http://www.fujixspot.com/f34/ Rico Pfirstinger

    Inasir’s sample from the DPR thread clearly demonstrates the resolution advantage of the X-Trans sensor (w/o AA filter) over the same sensor with a Bayer CFA (and an AA filter).

    Using a pre-release version of Photo Ninja that supports the X-E1 the new X-A1, I developed both RAW files and uploaded them on Flickr as maximum resolution JPEGs.

    I disabled luminance noise reduction and applied the same amount of sharpening. Have a look, the difference between the sensor is quite obvious in the details, like roofs or foliage.

    Here’s a direct comparison of 100% crops: http://farm6.staticflickr.com/5453/10128926884_4c1a097bd1_o.png

    Here’s the full-size X-A1 image: http://farm8.staticflickr.com/7345/10129037135_28e5627be2_o.jpg

    Here’s the full-size X-E1 image: http://farm4.staticflickr.com/3760/10129158256_bc7a16a643_o.jpg

  • http://www.fujixspot.com/f34/ Rico Pfirstinger

    As a reference, here are my Lightroom versions of both files:

    Direct comparison of 100% crops: http://farm8.staticflickr.com/7384/10129745515_1dbafe9b7e_o.png

    X-A1 Lightroom image: http://farm3.staticflickr.com/2835/10129862366_b44001c0e3_o.jpg

    X-E1 Lightroom image: http://farm3.staticflickr.com/2857/10129964143_6c15d9e969_o.jpg

  • Dirk

    Of course, this is an interesting comparison. But I’m bit surprised and disappointed in the same time. First of all, as far as I know Fujifilm developed this new filter layout for two reasons, one was indeed the moiré-reduction (however this phenomenon still exists to a certain degree) and secondly, to match up in a better way with film. Btw, if you look a recent patents, Sony also seems to have plans to leave the traditional Bayer arrangement for (some?) sensors. What still stands for me, is that X-Trans is one of the only APS/C-sensors that almost equals FF in overall IQ-performance (leaving aside the geometry/crop factor discussion). And very true, the output is excellent. There is something special with the colours I can’t explain (let some people like Steve Huff call them ‘flat’, I completely disagree, there better than anything I’ve seen before including Leica). But in the mean time, I also feel how critical the RAW-conversion is. And there you go: LR is obviously not the best choice and was in some pixel peeping exercises very surprised by the difference LR, C1, SP, Aperture & Iridient have. The conclusion is silly, but true, Fujifilm X-owners better buy a Mac if they want to have the best conversion…?