Fuji INCONSISTENT Autofocus issue Pt. 3

[Nielk Mike]

Let's continue the discussion with the aim of trying to figure out what's wrong and how this could be addressed by Fuji.

Fo those who have not seen AF-S issues on the X-T5 or any other models: That's fine, but I think it wouldn't be productive to continue the back and forth between "there is/there is not". No need to defend Fuji - great company, great cameras, great lenses.

Advice on how to obtain sharp images despite the AF issues has also been given in abundance. We know the work arounds, and yes, Fuji cameras can deliever very sharp images.

So where are we now:

There are reports that the X-T5 (and other models starting with the X-T1, and including the x100v, X-Pro2, and X-E3 and 4) do not reliably focus in AF-S, in particular at apertures f5.6 and up.

Mostly with lenses wider than 35mm. Almost (or completly) absent with lenses like the 50f2 or 90f2.

Aiming at the same target and repeatedly half pressing the shutter button can result in distances between 1,5m and infintiy.

The use of LM motors doesn't make a difference.

On older models like the X-T1, using just the CDAF focus points outside of the PDAF focus points doesn't make a difference.

AF-C (if the shutter is pressed without half pressign first, and AF-C set to Focus Priority) returns better results.

Checked my Sony cameras (including the trusted NEX5R) - no such issue.

What can cause this behavior (hopefully someone with technical background can chime in):

The lens? Well, while the 14f2.8 is among the worst when it comes to "AF distance jumping), the latest and greatest 18f1.4 isn't immune to the issue. But then there are some which do not show the issue at all. Prime example: The 90f2.

X-Trans sensor? In another thread (Future of X-Trans sensor) someone said he doubts that.

Large DoF with wide angle lenses stopped down? Some else mentioned this and it sounds plausible, The 14f2.8 @f8, everything from 60cm to infintiy has "acceptable" sharpness if one uses standard DoF claculators (Film Based in the Fuji world. Pixel Based is a different stoy. DoF is half, if not less, of the Film Based DoF.

So the theory is that with a large DoF the camera just can't decide where to put the focus within the area that appears sharp to the camera. So it just puts it somewhere within that range.

But it does make a big difference (in particular on teh 40MP sensor) where the camera puts the focus. 3m is different form infitniy. The are covered by critical focus changes. At Pixel Format DoF, areas that should be in critical focus may look good if not cropped, but lack sharpness when cropped in.

Two questions arise: Why does the aperture close for AF? If it creates an area in which the AF struggles to distinguish levels of sharpness, why not focus with the lens wide open? May be at the time it was thought to be a good thing if lenses displayed focus shift.

Why do Sony cameras not struggle to detect the highest level of sharpness within a wide DoF? The NEX5R closes the aperture, too. Yet it finds correct focus without "jumping". So what are the technical differences between the two AF systems?

Would find it most helpful and interesting if members with better technical knowledge could chime in.

 

[Massinissa]

I just add that, apart having some images slightly soft, sometimes the error is so big that nothing is in focus (OOF).

Tried also different box sizes and no joy.

 

[BklynBill]

I use "M" and the "AF-ON" button to focus(back button focusing), works better, but not perfectly(the lens at least is wide open for focusing). With the 16-55 zoom, at 55mm focusing is perfectly consistent, at 16mm it can vary widely, but being on "M" allows one to change that. One would think changing to a small focusing box would help, but it doesn't. Interestingly though, I've had no problems with the 14mm-but maybe DOF hides them. As you said, the 90mm is perfect, as is my 55-200 zoom and both my 30mm & 60mm macros.

 

[DJF55]

Let's continue the discussion with the aim of trying to figure out what's wrong and how this could be addressed by Fuji.

Fo those who have not seen AF-S issues on the X-T5 or any other models: That's fine, but I think it wouldn't be productive to continue the back and forth between "there is/there is not". No need to defend Fuji - great company, great cameras, great lenses.

Advice on how to obtain sharp images despite the AF issues has also been given in abundance. We know the work arounds, and yes, Fuji cameras can deliever very sharp images.

So where are we now:

There are reports that the X-T5 (and other models starting with the X-T1, and including the x100v, X-Pro2, and X-E3 and 4) do not reliably focus in AF-S, in particular at apertures f5.6 and up.

Mostly with lenses wider than 35mm. Almost (or completly) absent with lenses like the 50f2 or 90f2.

Aiming at the same target and repeatedly half pressing the shutter button can result in distances between 1,5m and infintiy.

The use of LM motors doesn't make a difference.

On older models like the X-T1, using just the CDAF focus points outside of the PDAF focus points doesn't make a difference.

AF-C (if the shutter is pressed without half pressign first, and AF-C set to Focus Priority) returns better results.

Checked my Sony cameras (including the trusted NEX5R) - no such issue.

What can cause this behavior (hopefully someone with technical background can chime in):

The lens? Well, while the 14f2.8 is among the worst when it comes to "AF distance jumping), the latest and greatest 18f1.4 isn't immune to the issue. But then there are some which do not show the issue at all. Prime example: The 90f2.

X-Trans sensor? In another thread (Future of X-Trans sensor) someone said he doubts that.

Large DoF with wide angle lenses stopped down? Some else mentioned this and it sounds plausible, The 14f2.8 @f8, everything from 60cm to infintiy has "acceptable" sharpness if one uses standard DoF claculators (Film Based in the Fuji world. Pixel Based is a different stoy. DoF is half, if not less, of the Film Based DoF.

So the theory is that with a large DoF the camera just can't decide where to put the focus within the area that appears sharp to the camera. So it just puts it somewhere within that range.

But it does make a big difference (in particular on teh 40MP sensor) where the camera puts the focus. 3m is different form infitniy. The are covered by critical focus changes. At Pixel Format DoF, areas that should be in critical focus may look good if not cropped, but lack sharpness when cropped in.

Two questions arise: Why does the aperture close for AF? If it creates an area in which the AF struggles to distinguish levels of sharpness, why not focus with the lens wide open? May be at the time it was thought to be a good thing if lenses displayed focus shift.

Why do Sony cameras not struggle to detect the highest level of sharpness within a wide DoF? The NEX5R closes the aperture, too. Yet it finds correct focus without "jumping". So what are the technical differences between the two AF systems?

Would find it most helpful and interesting if members with better technical knowledge could chime in.

Thx for the summary. I can think of further factors that may lead to differences in observed lack of sharpness:

- OOC or RAW? If OOC, settings, including film sim; if RAW, software and settings.

- Shutter type: MS, EF or ES, also tripod yes/ no

- Experience with hand-held shooting

- Same lens variations

In other words, I think there are still insuffucient information from from both camps (sharp/not sharp) to objectively conclude that there is a problem. Not saying there is none, though.

 

[Nielk Mike]

I just add that, apart having some images slightly soft, sometimes the error is so big that nothing is in focus (OOF).

Tried also different box sizes and no joy.

That may happen when the camera focusses beyond infinity. Seems to be a subset of the "AF jumping" issue.

 

[Nielk Mike]

Thx for the summary. I can think of further factors that may lead to differences in observed lack of sharpness:

- OOC or RAW? If OOC, settings, including film sim; if RAW, software and settings.

RAW - but I don't see why JPEGs should be affected differently. After all, the "jumping takes palce" before the JPEG is created.

- Shutter type: MS, EF or ES, also tripod yes/ no

All shutter types. No tripod. But neither shutter type nor tripod would make difference as the "jumping" can already be seen on teh distance indicator. The result is not camera shake or subject movement.

- Experience with hand-held shooting

Me? A lot. It is neither camera shake nor subject movement.

- Same lens variations

Lens variation? Given that this is not limited to one specific lens, unlikely.

In other words, I think there are still insuffucient information from from both camps (sharp/not sharp) to objectively conclude that there is a problem. Not saying there is none, though.

 

[Massinissa]

Thx for the summary. I can think of further factors that may lead to differences in observed lack of sharpness:

- OOC or RAW? If OOC, settings, including film sim; if RAW, software and settings.

- Shutter type: MS, EF or ES, also tripod yes/ no

- Experience with hand-held shooting

- Same lens variations

In other words, I think there are still insuffucient information from from both camps (sharp/not sharp) to objectively conclude that there is a problem. Not saying there is none, though.

- OOC or RAW doesn't matter

- Shutter type doesn't matter

- tripod or not doesn't matter

- OIS or IBIS doesn't matter

tested with 2 X-T5 and 2 23 f1.4 LM WR and 1 10-24 f4 Mk1

I wouldn't say lack of sharpness but slighly or heavily out of focus, even with large DOF as the 10-24 at 10mm f5.6 or f 8. (Diffraction is not the issue).

You can check my video and download RAF/jpegs samples from the links in the description

 

[PLShooter]

There are those of us who purchased the X-T5 or X-H2 for their high res sensors to be able to crop wide angle images. That is when the OOF issue rears it's ugly head. I've noticed that a few responders have submitted images that they state look OK, but once these are cropped, look like crap.

FWIW, I took my Z fc with the new Z 12-28 lens out yesterday and did not get one (1) OOF image. I paid much less for this combo than I did for my X-T5 body alone. I want my X-T5 to perform at least as good as my much less expensive Nikon. :-|

--
"AMF"

 

[jano50]

My son`s a6000 is also much cheaper (and older), but the S-AF is dead-on even with cheapo lenses...

 

[DJF55]

Thx for the summary. I can think of further factors that may lead to differences in observed lack of sharpness:

- OOC or RAW? If OOC, settings, including film sim; if RAW, software and settings.

- Shutter type: MS, EF or ES, also tripod yes/ no

- Experience with hand-held shooting

- Same lens variations

In other words, I think there are still insuffucient information from from both camps (sharp/not sharp) to objectively conclude that there is a problem. Not saying there is none, though.

- OOC or RAW doesn't matter

- Shutter type doesn't matter

- tripod or not doesn't matter

- OIS or IBIS doesn't matter

tested with 2 X-T5 and 2 23 f1.4 LM WR and 1 10-24 f4 Mk1

I wouldn't say lack of sharpness but slighly or heavily out of focus, even with large DOF as the 10-24 at 10mm f5.6 or f 8. (Diffraction is not the issue).

You can check my video and download RAF/jpegs samples from the links in the description

Well then, good luck with making your point with Fujifilm.

 

[Massinissa]

Another odd behaviour that happened with BFF in AF/M too with the 10-24 at 24mm f8

Subject at 50m, press BFF, focus distance went to 10m

Pressed BFF again, focus distance went to 5-10m

Pressed BFF again, focus distance went to 5m

Each time was like the AF alghoritm was subtracting the same value, focus peak gave that everything was in focus but zooming looked soft.

Turned OFF/ON the camera, pressed BFF and this time the focus distance was correct at first try and focus was sharp when zoomed in.

 

[zurubi]

I'd like to comment that I verified this behavior with the wide angle lenses on the XT5 (10-24, 16-55 at short range, etc). Initially I was skeptical but you are right. I sold all my previous XT models so I can't test this on the older cameras. I don't shoot too much wide angle these days so maybe that's why I missed this. The only thing that saves the day is that at short FL's and closed aperture almost everything is in focus....

 

[addieleman]

Checked this on my X100V with the latest firmware V3.00. AF-C, single point, subject (book case) at about 5m, 10 shots at each aperture from f/4 through f/16. Camera on tripod.

At f/4 and f/5.6 focus seems fairly reliable, only small variations around optimal focus. At f/8 focus is starting to become unreliable, at least 3 shots are too unsharp to my taste. At f/11 and f/16 the focus scale was jumping around like crazy, anywhere between 2m and infinity and hardly any picture was adequately sharp. Ugh.

At least it's good to know that I have to avoid apertures smaller than f/5.6 for my street photography and I understand now why I had to resort to MF for static shots at f/8.

This is my first Fuji camera besides my Sony system around the A7R4 and the experience has been less than stellar when it comes to image quality. I suspect this is largely due to this focus problem. Might be the last Fuji camera too.

 

[Nielk Mike]

Checked this on my X100V with the latest firmware V3.00. AF-C, single point, subject (book case) at about 5m, 10 shots at each aperture from f/4 through f/16. Camera on tripod.

At f/4 and f/5.6 focus seems fairly reliable, only small variations around optimal focus. At f/8 focus is starting to become unreliable, at least 3 shots are too unsharp to my taste. At f/11 and f/16 the focus scale was jumping around like crazy, anywhere between 2m and infinity and hardly any picture was adequately sharp. Ugh.

At least it's good to know that I have to avoid apertures smaller than f/5.6 for my street photography and I understand now why I had to resort to MF for static shots at f/8.

This is my first Fuji camera besides my Sony system around the A7R4 and the experience has been less than stellar when it comes to image quality. I suspect this is largely due to this focus problem. Might be the last Fuji camera too.

Thanks for sharing. One thing you need to look out for with the X100v is focus beyond infinity. Caused quite a lot of soft images when I used the camera.

 

[Kyle Logan]

This got me curious so I got out my XT30II and 14 and tested various distances and apertures. The distance AF indicator was accurate every time. So I’m not seeing these issues.

 

[Alan Sh]

You are not the only one.

 

[Samuraidog]

This got me curious so I got out my XT30II and 14 and tested various distances and apertures. The distance AF indicator was accurate every time. So I’m not seeing these issues.

This is super important. If they can get it right with at least one camera, it should mean they can fix it with firmware update.

 

[Kyle Logan]

I don’t know how “scientific” my testing was but I shot at infinity, 30ft, 10ft and 3ft at various apertures and went back and forth between those distances - every time it registered the correct distance.

 

[PLShooter]

I don’t know how “scientific” my testing was but I shot at infinity, 30ft, 10ft and 3ft at various apertures and went back and forth between those distances - every time it registered the correct distance.

What lens did you test with??? What AF did you use??? BBF???

--
"AMF"

 

[Nielk Mike]

I don’t know how “scientific” my testing was but I shot at infinity, 30ft, 10ft and 3ft at various apertures and went back and forth between those distances - every time it registered the correct distance.

What lens did you test with??? What AF did you use???

He used the 14f2.8.

There are always reports from those who don't have issues and those who do. Personally, I would be very surprised if the issue was fixed on the X-T30II - and not on the X-T5.

This thread should be more about what could cause the behavior than starting the discussion again about whether there is a problem or not.