Observations from 20 months with Fujifilm X-series

BTW, I will not recommend any Fuji X camera with this strange 40MP sensor. With this 40MP move, Fujifilm has become a victim of the smartphone-inspired marketing hallucinations. Probably they used the ancient version of ChatGPT when developing their marketing strategy instead of human expert brains. The only new X System body that is worth buying today is the X-H2S with 26MP but blazingly fast sensor (thus, zero rolling shutter and 2x more sensitive AF system).

I can explain why 24-26MP is perfectly enough.
Thats a short sighted conclusion. I have the H2s and it's a great camera. But I also have a XT 50 with 40 meg sensor. I find it's a great lightwieght walkaround camera and I'm happy with the purchase. I think there's room for both camps to be happy.

Mark
 
I also have a XT 50 with 40 meg sensor. I find it's a great lightwieght walkaround camera and I'm happy with the purchase.
My go-everywhere kit is a 24MP X-E3 + Sigma 18-50/2.8 (wide open 90% of the time, aperture priority mode, floating ISO). I bet my kit is more lightweight than yours :-)

Even if I equip the X-E3 with the XF 23/2 and put the XF 50/2 in my pocket, too, the total weight will be Ok in comparison :-)
I think there's room for both camps to be happy.
Sure :-)
 
Last edited:
Sigh.

This litany reminds me of when I was first learning about DPI vs LPI when going from digital image to print. There was an otherwise solid rule that if you were printing at, say, 150 lines per inch, then you would need double that (300) for the dpi. I regularly printed at 175 lpi, and suffered myself terribly when I didn't have enough dpi.

But after a few production runs of a few different things, run with a variety of dpi's, including as low as 250, and not once saw any pixellation, I stopped worrying.

So that experience should make me agree with the OP that 25 mp (a lower number) is enough resolution for sharpness. But dammit, I have now invested thousands of dollars in an X-T5 and six different lenses, and I will defend 40 mp to the death, even if I can't see any difference.

So there.
 
For the still photo. I am not at all familiar with X-T5,

Regards, Andreas.
Wait! So after suffering through all those engineering numbers, now we get to the punchline. What is the point of your post if you're not even familiar with the X-T5 (which I would presume includes the actual images produced by the X-T5)????

Call me flummoxed.
 
For the still photo. I am not at all familiar with X-T5,

Regards, Andreas.
Wait! So after suffering through all those engineering numbers, now we get to the punchline. What is the point of your post if you're not even familiar with the X-T5 (which I would presume includes the actual images produced by the X-T5)????

Call me flummoxed.
Yes, I noticed that too.
 
For the still photo. I am not at all familiar with X-T5,

Regards, Andreas.
Wait! So after suffering through all those engineering numbers, now we get to the punchline. What is the point of your post if you're not even familiar with the X-T5 (which I would presume includes the actual images produced by the X-T5)????
I said that I am not familiar with X-T5, not that I don't know it at all. I had my hands on it and took a few photos and videos with my own, familiar lenses that I use on X-E3 and X-T20. I was not impressed at all. AF feels slower and more sluggish than on my two 8-year-old cameras. Files are big. IQ is visibly the same. From my memories, X-T3 performed observably better overall. Also, I had a quick try at X-H2S - this is an entirely different class! I was able to do quick panning in video! Fuji X cameras are not popular on our island, so I had only a few chances to make a brief trial.

That's why I started my comparison research, been curious about these unexpected results and impressions I got.

Obviously, I am exploring this topic from my personal interest PoV - I feel the need to upgrade, and add a high-quality, video-oriented camera to my gear, capable of UHD 4k 4:2:2 10-bit with no crop.

Changing the system is not an option for me; I did it once in my life, and I have no intention of repeating this experience. I am happy with the Fuji X system, with the lenses I have, but the only upgrade path I consider is to purchase a used, near-mint X-H2S next year. Fujifilm left me no other choice, unfortunately.

For paid video work, I'd rather rent a Canon R5 Mk2 with a pair of lenses for a few days.
Call me flummoxed.
:-)
 
Last edited:
I understand the need for engineers to do all the tests that they do. But with all due respect, for an end user (or what used to be called "a photographer") to bother with it is way beyond my understanding.

Nothing is going to be perfect, but I'll assure you that you'll discover imperfections that you wouldn't ordinarily observe by looking at only the numbers, or even at the numbers first, or even looking at the numbers at all.

For all I know, you could be right about 25mp is enough. And I (and a thousand others) could have been sold a bill of goods with a 40+mp sensor. But then by that logic, full frame cameras must give you the creeps and medium format must give you the vapors.

I bought into the XT5 only because I impulsively sold my X100V when the VI came out, and then couldn't get the VI. My only real gripe about it is that it won't fit in my pocket like the 100V did, even with a pancake lens on it.

But as angry as I am with Fuji for not being able to keep up with production for their X100 cameras (and a few lenses now too), I'm not ready to make a wild claim that they are pulling a fast one with producing a 40mp sensor. And for at least one project I have in mind, it will not only come in handy, but necessary.

And I would much rather have to downsample, than upsample.
 
Last edited:
...

For all I know, you could be right about 25mp is enough.
Thank you. For completeness, I'd add the "for most applications" clause.
And I (and a thousand others) could have been sold a bill of goods with a 40+mp sensor. But then by that logic, full frame cameras must give you the creeps and medium format must give you the vapors.
Sensors are not a bottleneck anymore; optics is. With a good (or even the very best) lens that is capable of ~80 LP/mm frame-average (show me one!), you get
  • ~1250 LP/PH on a Fuji X Super35 frame,
  • ~1920 LP/PH on a photo 24 x 36 mm frame,
  • ~2630 LP/PH on a GFX pseudo-MF 43.8×32.9 mm frame,
  • ~3,200 LP/PH on PhaseOne 100MP almost-MF 53.4×40.0 mm frame (close to my beloved 645 from my old days),
  • ~13,440 LP/PH if scanning a 6x9 negative low-ISO film shot (56 × 84 mm) with a good industrial drum scanner (12000 dpi, if I recall correctly) - I still own one 6x9 camera.
With downsampling these even to a 4k screen (not FHD) ~8MP, that is 1080 LP/PH, you will observe better details from the larger sensors. How can I disagree? The difference will drastically increase when moving to A3 print @ 300 dpi (I've done it and seen it many times in the past with my now-forgotten Canons).

However, the difference in detail between 40MP and 26MP, both from the APS-C frame downsampled for monitor viewing, will be somewhat observable, but not crucial; the lens limits LP/PH, and it increases only slightly between ~133 LP/mm and ~167 LP/mm Nyquist limit, which was my point.
I bought into the XT5 only because I impulsively sold my X100V when the VI came out, and then couldn't get the VI. My only real gripe about it is that it won't fit in my pocket like the 100V did, even with a pancake lens on it.

But as angry as I am with Fuji for not being able to keep up with production for their X100 cameras (and a few lenses now too), I'm not ready to make a wild claim that they are pulling a fast one with producing a 40mp sensor. And for at least one project I have in mind, it will not only come in handy, but necessary.
For a project where it matters, I'd rather take the Panasonic S1R and the sharpest lens I can get for it (think Sigma 105mm f/2.8 Macro Art, stopped to f/4 - f/5.6 - sharpness and resolution king of L-mount) and make this 8-frame multi-shot high-resolution capture, which uses the body IS mechanism to shift the sensor. The S1R outputs a single, monolithic RAW file — not eight separate files. This pixel-shifted RAW is effectively equivalent to four full frames worth of data processed from the eight source images. It provides a merged high-resolution of 16,736 × 11,168 pixels, or about 187 megapixels. Files are huge, you know. But in this specific case, the size is justified. I can't say the same about the Fuji 40MP file sizes. A colleague of mine often uses this trick for low-budget shooting of paintings; for heavyweight projects, he uses more serious tools.

Some say this trick is also possible for landscapes, but I doubt that - you need zero wind for it to work (and, obviously, bring a heavy, massive pro tripod to the location, which is often boring).
I would much rather have to downsample, than upsample.
How can I disagree :-)
 
Last edited:
Very nice write up. As a dad whose primary interest in photography has been documenting my kids adventures I have a very similar layout. Sub out the Helios for a Minolta rokkor 58 1.2 and the 50 - 140 for the 100-400 and we're pretty similar. Depending on your kids ages I would recommend keeping the 50-140. Sports...school events...various graduations...having a faster zoom lens for situations like that where you won't know the lighting or the distance can be very useful. Very specific use cases but generally pretty important ones. I have the 90 too and while it is probably one of my favorite lenses it is more for head shots after haircuts. Side line sports it's ok while they are on the kids fields but once they move up to larger playing areas you wish you had more reach.

I sold the 16-55...I already had the 18-55. I don't really miss it. It was an awesome lens. It is definitely above the kit lens in regards to image quality (and colors I found noticeably better) but I found I preferred the 35 1.4 for image quality and speed and 18-55 has great stabilization making it a bit better for low light and shooting the odd video.
 
The main advantage of the 40mp sensor imo, all things being equal on net gain of 26vs40 has to be the dynamic range,
I think it is improbable.
Well, I think Fuji did release their best 26mp sensor implementation around the X-T4/X-S10 period, with dr comparable at iso160 base to the X-T5 and X-H2 iso125 (which is according to PtoP(photons-to-photos) slightly better still. In any event, my observation with the X-H2s and X-S20 was that both clipped highlights very quickly and although PtoP doesn't have results for X-S20, it does show the X-H2s as significantly below the X-T5 at base. The X-H2s is not the best example due to the stacked sensor, also responsible for some loss of dr at base too.
The underlying tech under the photo sensors is the same; the sensor pixel pitch is higher (the pixel is smaller). I don't see any prerequisites for a higher dynamic range.
It doesn't really matter what pre-requisites you see, the test results from lenstip and PtoP all show dr has changed in the latest 26mp cameras, specifically the X-S20 and the in particular the X-H2s, due to its stacked architecture, and not for the better or even as good as the X-T4/X-S10.
base iso of 125
Yes — both Fuji’s 26 MP non‑stacked BSI APS‑C sensors (e.g., X‑T4, X‑S20) and the 40 MP X‑Trans CMOS 5 HR sensor also feature dual-gain (dual native ISO) architectures, but there are slight distinctions in their implementation and switch points.
Dual gain has been around forever, we all know about dual gain!
Dual-gain ISO means the sensor has two different analog amplifier circuits (gain stages) engaged at different ISO settings. This switching improves dynamic range and noise performance in low-light or highlight-heavy scenes by shifting to a higher-gain circuit beyond a certain ISO threshold.
  • Below the switch point: lower gain, more dynamic range
  • Above the switch point: higher gain, lower noise in shadows
This architecture is sometimes called dual native ISO, although strictly speaking, it’s dual-gain, and not necessarily full native ISO performance on both stages.

Fuji’s 26 MP BSI sensor uses dual hardware gain: one mode optimum for highlight/dynamic range (base ISO 160, and up to < 800), and another extending shadow performance at higher ISOs (ISO 800 and above).

I haven't seen any official Fujifilm statement explicitly calling the X‑Trans CMOS 5 HR “dual-gain.” Still, user testing and behavior in DR modes suggest it retains the dual-gain design. The most probable is the combination of the base 125 up to < 400, and 400+ modes.

In practice, this means a pretty simple rule: "If you can't shoot at the low (first) base ISO, then switch ISO to AUTO with the second-stage base ISO as a low threshold, maybe setting a top ISO at 12800".

So I don't perceive your point. What did you want to tell me? For me, either 125 first base ISO or 200 makes no difference.
For you, for others base iso in landscape is important, the base iso of 100/125 is a good compromise, the base iso of 64 for Nikon Z7ii/8 and before that d850 is probably a bit less convenient but iso100/125 is good balance for landscape and sport. Iso200 is too fast basically, its why I moved away from m43 and although Panasonic base has moved closer to iso100 I believe, Olympus (OM) is still iso200 I think
and imo Fuji's improved functionality. The X-H2s at base iso has a low base dr compared to the alternatives
I don't use DR settings other than 100%. For still photos, it does not influence RAFs. For video... honestly, I am not sure.
Yes, ok, I don't use other than 100% either.
and although this improves "relative" to the other X-T cameras as the iso increases, nevertheless the X-H2s is not really a landscape, day to day camera where you want the best image quality possible imo, it's tuned for action/sport first.
Look at the Part 1 and Part 2 explanations above. No lens exists that is capable of exceeding 1200 LP/PH on the APC-C frame. 26MP is enough.
It doesn't really matter, if you don't like 40mp, it's ok, I think the convenience of having 40mp shot to shot is huge for Fuji, it really puts them on a par with 45mp FF sensors, I will share some comparisons if you need them and the 40mp sensor is really delivering on the X-T5 especially, if it delivers in the X-E5 as well it puts Fuji in a very strong position. Combined with their excellent object/people/fave/eye/animal detection it's a very strong offering. The X-H2s is big, X-T4 previous gen didn't have the capabilities of the X-T5 and just wasn't as reliable in eye detection and was larger too, the X-T5 is imo a far better camera than the X-T4 in so many ways. X-H2s is a big, FF body size like the X-H2 and despite the better evf, overall if I had to use a body of that size I'd go FF, and use my R5ii.
The base iso of 125 can also be useful for getting more light on the sensor for longer and the extra features like 4:3/5:4 ratio are also great for visualizing crop's in-camera.
Why? I see neither theoretical nor practical use of this wisdom of yours.
Being able to visualize different aspect ratios is very useful, Fuji agree too, they included it and they make a huge thing of it on their new RF camera too, so unless Fuji are badly informed they would not include it, or indeed the dedicated ratio dial on the RF.
Certainly the X-T5 rear tilt screen is more practical too than the fully-articulated screens as well.
I don't care. For stills, I use the viewfinder 100% of the time. For video, I simply don't care about whether it's tilted or full-articulated.
In certain situations, camera on a tripod etc, it can be very useful to see the rear screen for framing and in low level situations. Keeping the screen behind the camera rather than off to one side is very useful as its just so much more practical and logical to have it behind too.
I think the X-H2s has its place but Fuji has paved a new path with the 40mp sensor which some will embrace and others want to dismiss.
I will dismiss.
I have dismissed 26mp aps-c, 40mp is so much more useful.
I think the X-T5 in most practical cases proves the equal to my Canon R5ii in iq and dr most of the time, aside from af tracking and speed/buffer ruggedness, so that's pretty impressive for what is a very compact camera.
...except the weaknesses of the 40MP sensor, as mentioned above, ahha, Ok.
I haven't see many weaknesses of the 40mp sensor, probably video, which I don't do much any more, is the biggest issue with these cameras and the slow readout in e-shutter. In any event, 4k/50p with a tiny 1.14x crop has very low rs and is probably excellent for most users purposes imo. For static based footage you can go for 6k or oversampled 4k if you need it, even my R5ii doesn't have 6k video :( just 8k.
If the 16-55ii had proven to be as good as I was expecting I would probably keep my Canon R5ii for purely wildlife/action.
I apologize, but I am neither familiar with the 16-55/2.8 Mk2 nor do I intend to purchase it.
Probably a good idea as I haven't been too impressed with either of the two lenses I've had so far, both have been returned.
Regards, Andreas
 
The main advantage of the 40mp sensor imo, all things being equal on net gain of 26vs40 has to be the dynamic range,
I think it is improbable.
Well, I think Fuji did release their best 26mp sensor implementation around the X-T4/X-S10 period, with dr comparable at iso160 base to the X-T5 and X-H2 iso125 (which is according to PtoP(photons-to-photos) slightly better still. In any event, my observation with the X-H2s and X-S20 was that both clipped highlights very quickly and although PtoP doesn't have results for X-S20,
Hope you don't mind me jumping in with a reply that's not pertinent to the main discussion topic.

I've been looking to get an X-S20 and all the reviews that I've read (and watched) all say that it uses the same sensor as the X-S20.

And I'm guessing that PtoP may not have the X-S20 data for that reason.

Do you have anything you can point me to that says that the sensors are different?
 
You forgot to do one for the 60mp Sony FF sensor and the Fuji 100mp MF sensor.....

--
------------------
RWN Photo
 
Last edited:
The main advantage of the 40mp sensor imo, all things being equal on net gain of 26vs40 has to be the dynamic range,
I think it is improbable.
Well, I think Fuji did release their best 26mp sensor implementation around the X-T4/X-S10 period, with dr comparable at iso160 base to the X-T5 and X-H2 iso125 (which is according to PtoP(photons-to-photos) slightly better still. In any event, my observation with the X-H2s and X-S20 was that both clipped highlights very quickly and although PtoP doesn't have results for X-S20,
Hope you don't mind me jumping in with a reply that's not pertinent to the main discussion topic.

I've been looking to get an X-S20 and all the reviews that I've read (and watched) all say that it uses the same sensor as the X-S20.

And I'm guessing that PtoP may not have the X-S20 data for that reason.

Do you have anything you can point me to that says that the sensors are different?
Sure, as you can see here the X-S10 dr was excellent, which I'd agree with too, certainly on a par with X-T4.


Now, look at how it changed on the X-S20?

over 1ev less at base than the X-S10, something changed, alot by Fuji!


was it to do with faster scanning for video/stills, not sure but it's quite disappointing, x-m5 is similarly disappointing compared to the 26mp sensors in the X-T4/X-S10!


So, right now, we don't really have a great modern 26mp sensor implementation from Fuji imo outside of the X-H2s, so we can argue about 26 vs 40 until the cows come home but the fact's are the fact's, you probably do get similar dr in some circumstances from the X-H2s but again I did find it below the X-T4, which is correct and the X-T5 much better, which is probably not entirely accurate but I do prefer the X-T5 resolution and overall functionality by some margin, especially size, of the X-T5 to the X-H2s, ymmv.

 
You forgot to do one for the 60mp Sony FF sensor and the Fuji 100mp MF sensor.....
Don't encourage him.
I appreciate "Andreas Undone's" research, not least because it agrees completely with my own lived experience of these last two years with the X-H2! There's convenience to be had in the cropping ability of all those wasted pixels and terabytes on disk, but the photos (while always fatter) are very rarely any better and occasionally worse than the output I get from my x-Pro3. The AF is somehow simultaneously better and worse than X-Pro3, and the number of photos I have littering my hard drives that are ever so slightly out of focus is appalling. I grab the X-H2 for video or birds, but for most other things I grab the X-Pro3 - It's just more reliable and consistent for me. I shipped my X-H2 off to Keh last week and have been deciding what to replace it with and have been having a hard time deciding between the X-T4 or X-H2s. I'm not a fan of chunky DSLR ergonomics, but that stacked sensor has a sex appeal all its' own ;-)

Update: I split the difference and went with the X-S20! Sensor from the X-T4, processor from the X-H2s. Much more svelte and comfortable in the hand and on the shoulder strap...
 
Last edited:

Keyboard shortcuts

Back
Top