Fuji Issues with X-Trans sensors

[Tomp]

I have been reading posts by Diglloyd re the X-Pro 2. diglloyd - Guide to Mirrorless Cameras
The excerpts quoted below describe Lloyd Chambers' findings (this is a subscription based site so the details quoted below are not available to non-subscribers).

The issues are backed up with ample proof and clear images. They involve colour balance; grid artifacts; and, 'worm-like' artifacts - exacerbated with sharpening.

1. "...the X-Pro2 consistently delivers in-camera JPEGs with a green tint..." The recommended remedy is a global +1Magenta (B=1 R=1) setting.
2. "...the X-Pro2 can generate images with extensive grid artifacts over large areas of the frame, causing severe damage to image quality. The issue occurs with in-camera JPEGs from raw or when processing raw RAF files in ACR. Presumably it occurs with straight JPEG-only as well."
3. "Fujifilm camera bodies preceding the X-Pro2 [and including the new X-Pro 2 sensor] have all had ... fractal-like artifacts which crop up regularly, particularly with natural textures."

Has anyone come across these issues? The reason I am posting this is because I am seriously thinking of upgrading from my X100s to a T2 later this year.

 

[Adam Bonn]

1. AWB is dependent on a lot of things... The light when you shoot, what's in the frame.. Some times it does a better job than others. FWIW I find the AWB better on the XP2 than the XT1 but others don't

2. Some have reported this issue in VERY particular circumstances eg, a very heavily back lit subject with a lot of flare. Personally I haven't seen it... But I've seen it in examples, those reporting in the things I've read, say it's occasional not every shot.

3. Well he's talking all X-Trans sensors here... And not everybody likes them... How are you finding your X100S? Do you like images you posted see from the X-Pro2?

 

[Tomp]

Thanks for your comments. The only issue that really concerns me is (3) about digital 'worm' or 'fractal' types of artifacts. I have found something very similar with RAW files from my 100s. I have to be very careful not to overharpen as they will quickly start to form these types of artifacts. I use PhotoNinja for RAW conversion and also use its smoothing feature in the Noise Reduction settings which seems to help. I think there is some kind of edge sharpening happening in camera that promotes the development of these artifacts - even in RAW files.

 

[Adam Bonn]

Most people complaining about artefacts point the finger at ACR, with PN and Irridient being the good guys

IMO too much as been said about this, by too many people...

I'm NOT saying it's not a thing... But people do like to repeat what they've read as fact or confuse things like OOF areas with smudged details...

That said, the XP2/XT2 files are 30% bigger than on previous X cameras, so if someone looks for problems, they're easier to find

But, if you're not that happy with your X100S files, then....

Perhaps it's worth downloading some raw XP2 files (there's probably some XT2 files out there by now to) and running then through your workflow and seeing what you think.

 

[flysurfer]

After 4.5 years of X-Trans in the X-Pro1, X-Pro2, X-E1, X-E2, X-E2S, X-M1, X-T1, X-T2 and X-T10 as well as the X20, X30, XQ1, XQ2, X100S, X100T and X70, there are countless (I'd estimate hundreds of thousands) sample images available. If most or many of these images exhibit artifacts or color deviations that aren't acceptable for a particular user, that's certainly a good reason to look elsewhere. However, one also has to make sure that the X10, X-S1, XF1, X100, X-A1 and X-A2 (all using more conventional Bayer sensors) do not exhibit the same or similar artifacts or color deviations. Only then can you assume that X-Trans may be to blame for the difference. Plus, in order to prove that artifacts or deviations of any kind have their roots in the camera and not the RAW converter, it's important to test each camera with several different RAW converters that use different demosaicing engines.

 

[Tomp]

After 4.5 years of X-Trans in the X-Pro1, X-Pro2, X-E1, X-E2, X-E2S, X-M1, X-T1, X-T2 and X-T10 as well as the X20, X30, XQ1, XQ2, X100S, X100T and X70, there are countless (I'd estimate hundreds of thousands) sample images available. If most or many of these images exhibit artifacts or color deviations that aren't acceptable for a particular user, that's certainly a good reason to look elsewhere. However, one also has to make sure that the X10, X-S1, XF1, X100, X-A1 and X-A2 (all using more conventional Bayer sensors) do not exhibit the same or similar artifacts or color deviations. Only then can you assume that X-Trans may be to blame for the difference. Plus, in order to prove that artifacts or deviations of any kind have their roots in the camera and not the RAW converter, it's important to test each camera with several different RAW converters that use different demosaicing engines.

Thanks Rico. The research in the article I quoted from was based on raw conversion with ACR and Capture One. Both raw converters exhibited the artifacts although C1 was less pronounced. The article used images from XPro 2 which has the same sensor as the T2 - the camera I am aspiring to purchase. My own experience has been with the 100S with PhotoNinja, which I find intuitive to use and produces good results for me. But what I see in my images might be just my faulty post processing - I am still learning here. My post is trying to establish whether the appearance of these artifacts as evidenced in Lloyd Chambers' article is a characteristic of the X Trans sensor or perhaps more of an isolated issue. He seems to think it is the former. If no-one else has experienced these, then I guess it is not a game changer.

 

[Lightmancer]

Thing is, there is an optical chain. It starts with you. You set your camera up in a particular way. You then point it at a chosen subject, you hold it as steady as you are able and you choose when to press the shutter release. You then edit the image in your chosen way, with your chosen tools. Some will be better suited to what you are trying to do than others.

Personally I use SooC jpg much of the time, and the built-in raw converter on occasion. I don't use a third party converter because that introduces another link into the optical chain. I use PSE for editing and occasionally Topaz B&W or Nik for a particular end result. I restrict myself to straightening and cropping, tweaking levels and hue, converting to mono and dodging and burning. I very, very, very rarely tweak sharpness. I firmly believe that it is really easy to get carried away in post, and to make your life much harder than it needs to be. A file that has been over-manipulated breaks down in the same way as a poorly made sauce; the end result becomes unpalatable.

In much the same way that a poor workman blames their tools I do wonder to what extent those who complain about the "issues" with X-Trans files are serial post tinkerers. I also note that there is not a lot of motivation for the likes of Adobe to invest the development man-hours in getting the best out of what is still a minority file type.

Whatever. I can only speak for myself. Happy bunny here.

 

[wt21]

I, too, am considering trying Fuji, so I'll let the others comment. The only thing I wanted to add was I don't trust Digiloyd. I've found him off base on m43 and Sony postings. I stopped reading him a while ago.

 

[Adam Bonn]

Plus, in order to prove that artifacts or deviations of any kind have their roots in the camera and not the RAW converter, it's important to test each camera with several different RAW converters that use different demosaicing engines.

This! Because we often read about the best settings for X-Trans sharpening, and I'm certainly NOT saying that these tips aren't helpful and valid, but as Rico says, it's the demosaicing that is the foundation of your image from RAW, if this process doesn't suit your image or image tastes, then it's hard to resolve (like trying to uncook a cake to add more egg whites!)

Not only do different RAW convertors use different demosaicing algorithms, but different images with different content suit different post processing methods

. But what I see in my images might be just my faulty post processing - I am still learning here. My post is trying to establish whether the appearance of these artifacts as evidenced in Lloyd Chambers' article is a characteristic of the X Trans sensor or perhaps more of an isolated issue.

The 'problem' (sic) with post processing is that many of the things that we like in our images are actually worsened if we're not careful... I mean who doesn't want a sharp, noise free, great clarity shot with rich colour? But sharpening, noise reduction, extra clarity and over saturation can lead to false details, smeared details and colour casts, which leads me nicely to the excellent point below

Thing is, there is an optical chain. It starts with you. You set your camera up in a particular way. You then point it at a chosen subject, you hold it as steady as you are able and you choose when to press the shutter release. You then edit the image in your chosen way, with your chosen tools. Some will be better suited to what you are trying to do than others.

Personally I use SooC jpg much of the time, and the built-in raw converter on occasion. I don't use a third party converter because that introduces another link into the optical chain. I use PSE for editing and occasionally Topaz B&W or Nik for a particular end result. I restrict myself to straightening and cropping, tweaking levels and hue, converting to mono and dodging and burning. I very, very, very rarely tweak sharpness. I firmly believe that it is really easy to get carried away in post, and to make your life much harder than it needs to be. A file that has been over-manipulated breaks down in the same way as a poorly made sauce; the end result becomes unpalatable.

In much the same way that a poor workman blames their tools I do wonder to what extent those who complain about the "issues" with X-Trans files are serial post tinkerers. I also note that there is not a lot of motivation for the likes of Adobe to invest the development man-hours in getting the best out of what is still a minority file type.

Whatever. I can only speak for myself. Happy bunny here.

Which is the long and short of it! Except I'll add, that images have end uses... There is a world of difference between a high dpi ultra print on gloss, and an image on a computer screen, not only is the ability of the post processor's screen to resolve detail relevant in the process, but also the screen of the viewer...

As stated above, when a camera has a non standard image solution (eg X-Trans or the method of lossy raw compression of the first generation Sony A7) then it's very easy to look to this non standard feature as the source of ALL problems.

The reality is that non standard solutions, generically speaking, excel in some scenarios and not others.

The challenge is in deciding whether or not a non standard solution works for your particular type of photography or not.

I'm not claiming X-Trans is perfect, but there's enough people out there making enough great images, making their living from it etc that clearly X-Trans is a perfectly valid solution.

There's enough helpful articles and YT videos, not to mention RAW samples to download, that I think people can do their research and form their own opinion and all without investing in the Fuji system.

 

[flysurfer]

Oversharpening artifacts have different shapes with X-Trans than with Bayer. X-Trans tends to create worms (it's a larger 6x6 pattern, after all), Bayer creates blotches that are often confused for false detail. Just like noise can look nice (similar to grain that people also like), sharpening artifacts (aka false detail) can look nice, too.

 

[Adam Bonn]

Personally I use SooC jpg much of the time, and the built-in raw converter on occasion.

slightly OT...

I've only had my XP2 for 3 weeks...

and the SOOC jpgs... Jeez, I'm totally amazed, I mean there's a bit more from RAW (in context of me, what I like/want/am able to get - others MMV) for my colour stuff...

But acros, take the SD card out of the 'oven' and stick a fork in it - it's done

I think Fuji just fixed my "Hmmm Adam, your BW pp game isn't all that" problem

Anyway, sorry for the OT, I'm just playing with a shot of the cat, first the raw went through my PP, then I keep making SOOC jpgs and comparing them to my RAW effort!

 

[Adam Bonn]

I have been reading posts by Diglloyd re the X-Pro 2. diglloyd - Guide to Mirrorless Cameras
The excerpts quoted below describe Lloyd Chambers' findings (this is a subscription based site so the details quoted below are not available to non-subscribers).

The issues are backed up with ample proof and clear images. They involve colour balance; grid artifacts; and, 'worm-like' artifacts - exacerbated with sharpening.

1. "...the X-Pro2 consistently delivers in-camera JPEGs with a green tint..." The recommended remedy is a global +1Magenta (B=1 R=1) setting.
2. "...the X-Pro2 can generate images with extensive grid artifacts over large areas of the frame, causing severe damage to image quality. The issue occurs with in-camera JPEGs from raw or when processing raw RAF files in ACR. Presumably it occurs with straight JPEG-only as well."
3. "Fujifilm camera bodies preceding the X-Pro2 [and including the new X-Pro 2 sensor] have all had ... fractal-like artifacts which crop up regularly, particularly with natural textures."

Has anyone come across these issues? The reason I am posting this is because I am seriously thinking of upgrading from my X100s to a T2 later this year.

Back OT... And here's a (IMO) decent article that goes some way to address some of your concerns....

Problem solving: How to deal with the Fujifilm X-Pro2 “grid” artefacts - MirrorLessons - The Best Mirrorless Camera Reviews

The chambers article is mentioned in the comments, and the comments (as of 03/08/16) are worth reading

 

[Tomp]

Thanks Adam. The article you referenced is very useful and indicates that this particular issue is easy to remedy; this is good news. Personally, I am still very keen on the T2 and will seriously look at it when it is available. For me, it will be a toss up between the T2 and the Nikon D500. The Nikon is bigger, as are the lenses available for it. The use of full-frame lenses on APS-C sensors can lead to loss of quality so this detracts from the Nikon with its limited range of lenses for that format. So, pending a hands-on experience, the Fuji still has my preference.