Fujifilm Q&A: COVID impact, GFX strategy, shrinking IBIS, secrets of a 300,000-cycle shutter and more

 

 

This is the third in a series of interviews I conducted while in Japan in early March of this year, a trip I barely managed to sneak in before the Coronavirus travel restrictions. Between the COVID situation, the sale of IR, and a number of other lingering content projects I needed to finish, I'm only just now managing to post it. World events have overtaken my questions at the time about the projected impact of COVID on Fujifilm's production capability, so I reached out to them for an update before posting this story, and have replaced the earlier questions and answers on the topic with the fresh information below. They also answered a couple of questions that were left open during the interview itself.

 

Toshihisa Iida
(Now former) General Manager
Optical Device & Electronic Imaging Products Division
Fujifilm Corp.

My interview here was with Mr. Toshi Iida, the now former General Manager of the Optical Device and Electronic Imaging Products Division (he has since transitioned to managing Fujifilm's European operations), Mr. Shinichiro 'Shin' Udono, Senior Manager of the Sales and Marketing Group of the same divison and Mr. Jun Watanabe, Manager of the Sales and Marketing Group, also within the Optical Device and Electronic Imaging Products Division. (I unfortunately neglected to take a group photo of the executives, so have used images of Mr. Iida and Mr. Udono from previous interviews with those gentlemen; deep apologies to Mr. Watanabe that I don't have an image of him to use here :-/)

We covered a lot of ground, from general questions about how their lineup is evolving and where the strongest user interest is to a fairly deep technical discussion about the improvements they made in their IBIS technology in the X-T4. Read on for all the details!

Dave Etchells (updated question/answer): A question that's in everyone's mind, and the main thing that needs updating since we met in March is of course the COVID situation. It’s now five months since I was there; what is Fujifilm’s current situation, in terms of your manufacturing, your supply chain, and your ability to continue shipping cameras? Have things largely normalized, or are there still issues with some parts of your supply chain?

 

Shinichiro 'Shin' Udono
Senior Manager, Sales & Marketing Group
Optical Device & Electronic Imaging Products Div. Fujifilm Corp.

FUJIFILM: Our factory operation and supply chain already went back to normal. Our China factory, located in Suzhou, stopped their operation for 2 weeks in February. Also, we have a factory in the Philippines and the factory was closed in April to follow the lockdown direction by the government. Both factories went back to normal operation. Also, Fujifilm worked very hard to secure parts supply. We had some minor trouble, but now we don’t have any concerns regarding supply chain of parts.

Dave Etchells (updated question/answer): Could you also comment on how the global camera market looks from your perspective? Are sales coming back in some areas while continuing to decrease in others? What is the general direction globally? (That is, for the world taken as a whole, are sales recovering yet, have they flattened, or are they still declining?) What is your projection for 2020 as a whole?

FUJIFILM: In regards to the Mirrorless and D-SLR camera demand, we see that April was the worst month due to retail shop closures and less shooting opportunities due to shelter-in-place or stay-at-home orders in many countries. According to CIPA, the April mirrorless demand (in value) was -68% of last year, it was the bottom. Then gradually recovering, May was around -58%, June -37%. We don’t have the July number, but we feel July became better than June.

 

Jun Watanabe
Manager, Sales & Marketing Group
Optical Device & Electronic Imaging Products Div.
Fujifilm Corp.
(Mr. Watanabe is a key product planner for the XT series of cameras, particularly the X-T4)

In the case of Fujifilm, we believe that we fared better than the market trend. The X-T4, launched in late April, and is well accepted in the market especially for its high movie function and AF improvement. The X-T4 demand is very strong. Also, X100V, launched in February, performed better than we expected. Customers buy cameras even under difficult situations, if the products are unique and innovative.

It is very difficult to predict 2020 whole year camera demand due to the unique circumstances. Probably the demand will not go back to the same level as last year, but we expect that the situation will get better towards the holiday season.

[Those were the two updated questions and answers about how COVID has impacted their production and the overall global market; the discussion below is from the original interview.]

Dave Etchells: I was trying to figure out how the camera business was represented on your financial reports. It looks like you're doing OK on the electronic imaging line item on the financial reports. Is that basically your business, or are there other parts?

Toshi Iida: If you go to our financial report, there is a big financial segment we're calling [the] Imaging segment, that's one of the biggest three categories, as a whole. Within this imaging segment, there are three business domains: Photo imaging, which includes Instax, all the print film business; electronic imaging is predominantly digital cameras and interchangeable lenses; the other part is optical devices. So you will identify the three business domains included within the imaging segment. Electronic imaging is almost 100% digital cameras and interchangeable lenses.

Dave Etchells: Ah. So then, electronic imaging looked like it's doing OK. You're down on revenue, but you're making a profit though, so...

Toshi Iida: I think probably we don't disclose profit for individual segments, but overall for the imaging segment, I think the profit is still relatively OK.

Dave Etchells: Yeah, it looks like imaging solutions dropped, but only like 12.4% revenue, and actually, with constant currency basis it's only 9.2% then. But that also includes Instax.

Toshi Iida: Instax and printing and other optical devices.

Dave Etchells: Does that include Fujinon lenses also?

Toshi Iida: Yeah, yes.

Dave Etchells: Ah, OK. That's across the board. Ah. What are your general projection at this point? I mean, I know maybe coronavirus just changed everything, but what had you been projecting for this fiscal year? Because it's like last year, people were projecting that, and then it did that very suddenly. <hand gestures, indicating that people were projecting a more gradual drop, but the actual decrease was much greater> Where do you think, how do you think this year will be compared to last? Or what were you projecting?

Toshi Iida: The industry association, CIPA, collects the predictions from individual camera companies. and they make some kind of consensus. And they already reported a nearly 20% further drop, even compared to last year. We think it's a little bit too pessimistic, and it all depends on which categories you are looking at.

Dave Etchells: Mmm, mmm.

Toshi Iida: If you look at the overall camera market, maybe a further drop [can be expected]. But I think if you just focus on the mirrorless segment, or if you look at the high-end segment, I think the numbers should not be as pessimistic as CIPA stated. So the important thing is which segment you are looking at, and we are focusing on [more the high-end].

Dave Etchells: Yeah, I have a feeling that we'll continue down, but then flatten at some point. The long-term photo enthusiasts, they'll still be there.

Toshi Iida: Yeah, yeah.

Dave Etchells: It's the consumers that went away. Entry-level models are gone, but the people that it's really a passion for will stay.

Toshi Iida: Sure.

Dave Etchells: Yeah, it's interesting when I was talking with Yamaki-san [at Sigma], he thought back in the film days, and noted that SLRs were about four million units a year or so in their heyday. I think that we had been much higher than that, and are maybe somewhere around eight million, 8.5 million now. His thinking is that it will maybe level out at six million. Because so many more people have phones and are taking pictures, maybe the number of enthusiasts is higher. I think we're close to leveling out maybe; if so, that's good news.

For many companies, even the mirrorless segment is facing severe conditions, but it seems like Fujifilm's X-T3 and GFX 100 are doing pretty well. Can you comment on the sales performance of those two cameras, and how do you view Fujifilm's place in the market generally?

 

Fuji's X-T3 was a hugely successful model, popular with IR readers, and in fact won top honors in our 2018 Camera of the Year awards. It apparently continues to do well, and will remain in the market, even following the release of its successor, the X-T4.

Toshi Iida: We've been glad to see the results of the GFX 100. Since the introduction in June last year, sales have been much higher than we predicted. And consistently, the demand for X-T3 is very strong.

Dave Etchells: Mmm! Mmm-hmm.

Toshi Iida: I think you know a simple answer [as to why that's the case] is that we are different from any competition. Last year everybody except Fuji, pretty much, was focused on full-frame mirrorless. We intentionally ignored full-frame mirrorless as a category. We have concentrated the focus of our resources on GFX and APS-C. Not only the cameras, but also the lens development.

Dave Etchells: Mmm, mmm.

Toshi Iida: So rather than just [doing the same as everyone else], we took a different approach.

Dave Etchells: Yeah, the other manufacturers have really almost abandoned the APS-C market, and I think there's a strong need. And Fujifilm's products are different, too. I think things like Film Simulations (extensive article) are significant for a lot of people.

Toshi Iida: Yeah. I also think that.

Dave Etchells: And yeah, I'm also not surprised the GFX 100 would be selling well. It performs very well, and its competitors are two or three times more expensive. You have a very broad price range with the GFX line now. In the US market, we've seen the 50R on sale for only $3,500, which is very inexpensive. In the US, it just says instant savings of $1,000. In the UK, it says the discount is for a limited time and expires March 31. Is that pricepoint a long-term strategy for that product, in terms of where you see it in the market?

[Ed. Note: The 50R is still selling at this price in the US, so this question is a bit dated at this point. Their strategy relative to full-frame cameras is still illuminating though.]

Shin Udono: We think that price range, from $3,000 to 5,000 is kind of the top of the full-frame camera price range. We want to expand our large-format customer base to new customers. So that is the target audience from our side.

Dave Etchells: Mmm, mmm. So if you can hit that price point then, many people that are considering full-frame...

Shin Udono: Yeah, so they can choose the top-end full-frame or can upgrade to medium-format.

Dave Etchells: Yeah, yeah. I think that's very smart. Someone can get a kind of toehold in medium-format that way.

Shin Udono: Yes.

Dave Etchells: I forget, can you use XF lenses with an adapter and crop on GFX?

Toshi Iida: We don't have an adapter, I don't know third-party. The flange-back distance is different, so I think XF lenses can't adapt.

[Ed. note: Oh, DUH! The GF mount's flange focal distance is 26.7mm, while the X-Mount is only 17.7. I should have thought for a moment before asking that question :-/] 

Dave Etchells: Yeah. I was just thinking of inexpensive ways for people to get lenses to use with it.

Toshi Iida: Of course, there are a lot of adapters between the EF lens to GFX...

Dave Etchells: Ahhh! [Pentax K and Nikon F mounts would also have plenty of backfocus length to work with, as well as pre-mirrorless medium format mounts.]

Toshi Iida: That's popular, and there are third-party adapters, but they're good quality, they're working with autofocus...

Dave Etchells: Yeah, so many people that have been Canon EF shooters that are thinking of full-frame mirrorless, maybe you can get them to come to GFX.

Toshi Iida: Yes.

 

There are a variety of adapters that allow the use of full-frame lenses on GFX cameras. This one from Fotodiox provides full AF and aperture communication between the camera and the lens. Of course, full-frame lenses generally won't cover the full extent of the GFX's larger sensors, so you'll need to crop into the images. But with 50 or 100 megapixels image sensors, you'll still end up with high-resolution images. Some adapters include intermediary lenses though, that spread the light between the lens and sensor, allowing full-frame lenses to cover the entirety of the GFX line's sensors.

Dave Etchells: I guess one question is how has that $3,500 price been working for you, can you tell if that's happening in your customer base, that people are coming from full-frame to GFX?

Toshi Iida: Before that price change, we checked our customer registration database. According to that data, I think 70-80% of the customers were coming from non-Fuji cameras.

Dave Etchells: 70%? Wow!

Toshi Iida: It's probably the SLR side, full-frame SLR. Actually, the average coming from our X-series cameras is much, much smaller. That is one of the reasons we made that the pricing for the GFX 50R. We thought GFX will be adopted by full-frame SLR users.

Dave Etchells: Yeah. Ah, that's very interesting. Most of your customers are coming from...

Toshi Iida: Sure, because full-frame SLR users are now thinking which mirrorless they should upgrade to. Probably, GFX will be one of their options.

Dave Etchells: Yeah. How long ago did you make that price change?

Toshi Iida: I think we started the current pricing in February.

Dave Etchells: Ah, so it's recent.

Toshi Iida: Yeah, yeah. I heard from US and Europe, that we made a quick sell-through. The sell-through was actually much, much better. But it’s been only two to three weeks, so we don't yet know the whole picture.

Dave Etchells: So you have three models, the GFX 50R, the 50S and the 100. When people are coming into the line, where do you see them coming in, and what types of users are they? I mean, presumably different people are entering with the GFX 100 than the others.

Toshi Iida: Especially the GFX 50R, we're capturing street photographers, because of the camera’s style. And of course the GFX 100 is more serious or professional photographers: Fashion, commercial, fashion portrait...

Shin Udono: Interestingly, we started to receive lots of inquiries from B2B industrial-type customers.

Dave Etchells: Oh, really?

Shin Udono: Yeah, for archival. For example, museums and libraries; they have to archive.

Dave Etchells: Ah, I see. It feels to me like people would either do the GFX 50R or GFX 100. Is the 50S still seeing sales, or is it really splitting to the other two?

Shin Udono: The current sales, in terms of volume per month, GFX 100 is top.

 

I was quite surprised to learn that the 100-megapixel GFX-100 is actually the highest-volume seller in the GFX lineup. On the one hand, it's surprising that the most expensive body is the one that's selling in the greatest numbers - but on the other hand, there's a significant market for ultimate-resolution cameras for applications like archival imaging of historic artifacts, and the GFX-100 is dramatically less expensive than other 100-megapixel cameras on the market.

Dave Etchells: Really?! [I was very surprised to hear that the most expensive model is the actually the one selling in the highest volume.]

Shin Udono: It has similar volume to GFX 50R, and 50S is maybe half of 50R, something like that.

Dave Etchells: Wow, that's kind of amazing to me, that the GFX 100 would actually be tops in terms of volume. Jeez!

Shin Udono: Yeah, the GFX 100 sales volume is more than 50% bigger than we predicted in our original plan.

Dave Etchells: 50%, wow. You have to make a lot more of them than you planned, but that's a good kind of problem to have.

Shin Udono: Sure, sure.

Dave Etchells: And you are able to keep up with production; availability is good?

Toshi Iida: Yes, yes. We are working very hard. Yeah.

Shin Udono: We are catching up.

Dave Etchells: And so you said that for the 50R, maybe 70% of users are new to the system. Is GFX 100 also a very high percentage?

Shin Udono: It's a similar trend.

Dave Etchells: Mmm. Your lenses are a really important part of the system. How does the lens attach rate look between the different products within GFX?

Shin Udono: I think for the lens attachment rate, we don't have the data. But GFX 100 is higher than 50R. 50S and 100 are similar, but 50R is smaller. But in total, I think the attachment ratio for GFX system is more than three lenses sold per camera body.

Dave Etchells: Really?! Wow, that's very good. [This may be typical of high-end bodies in general, but it’s very high compared to full-product-line numbers I’ve heard from various manufacturers.]

Shin Udono: That's much higher than for X-series cameras.

Dave Etchells: Wow. And those are more expensive lenses generally, too, so it sounds like the GFX line is a really good business for you guys.

Shin Udono: Sure, yeah.

Dave Etchells: I had a question about how you make the case for people to step up to medium-format, rather than just buying a high-res full-frame. But it sounds like you're at price parity -- or close to it -- with the highest-end, so maybe the value proposition is just "Yeah, it's a similar price, but we've got a higher-quality image." How do you make the case and do the marketing to full-frame photographers?

Shin Udono: I think the key differentiation factor is image quality, as always. It's not just a number of pixels.

Dave Etchells: Right.

Shin Udono: It's the medium-format look. That's something striking to full-frame customers. I think that the three-dimensional feeling from the higher resolution is one point of differentiation. Dynamic range is another. And the benefit from the high sensor resolution can be realized by very good-quality lenses. GF lenses have the power to deliver 100 megapixel resolution. So as a whole, the system is really attractive for customers who pursue the best image quality. I think this is the best choice for that kind of people.

Dave Etchells: Mmm, mmm.

Shin Udono: But, you know, we have to be realistic on the pricing and cost. And the other factor is maybe the size, weight and usability. [These are all the] kinds of things we're looking at, how to make medium-format attractive and realistic as a choice for customers.

Dave Etchells: Mmm, interesting. I'm curious how that translates into product plans, because it seems like you're seeing very strong activity at both ends of the line. Do you see the direction being "Okay, we need something like a 50R, but smaller" or do you need GFX 100 with more capabilities? What do you see the market asking for most?

Shin Udono: I think the product concept for 50R, it's many customers want to have a smaller GFX 50S. The other thing is for 100 megapixel, we decided, rather than compromising quality or usability [to make it smaller], let's put IBIS in it.

Dave Etchells: Mmm, mmm.

Shin Udono: Our original thought was to make the cameras get smaller, and we debated whether we should put the IBIS in them or not. Of course, with IBIS in it, the camera has to be bigger. But we decided let's make a perfect GFX camera. That's the GFX 100.

Dave Etchells: Mmm, mmm. Yeah, so in the case of the 100, those users, they are less concerned about portability.

Shin Udono: Yep, yep. It's quality, and, you know, handheld 100-megapixel shooting. That's the more important concept.

 

The GFX-100's IBIS system is a first for medium-format cameras, and an essential element for handheld shooting with a 100-megapixel camera. This is actually a shot of the X-T4's IBIS assembly, but I chose to put it here because it shows the key elements of the magnets (you can see one as the silvery rectangle at the bottom) and coil (the coppery reflection on the surface of the magnet). Moving a medium-format sensor with micron-level resolution is quite a challenge, but the GFX-100's IBIS system manages to do so and provide a level of shake reduction that's essential to take advantage of the sensor's resolution when shooting handheld.

Dave Etchells: Yeah, yeah. Handheld 100-megapixel is definitely important. Normally you really have to have the camera on a tripod or be shooting with fast strobes to take full advantage of 100 megapixels. So what do you think people are asking for next? Even smaller than 50R, or...

Shin Udono: Always, always smaller, cheaper, or better ease of use. It's always challenging, but if you look at the X-T4 versus X-H1, that shows what can be achieved in several years.

Dave Etchells: One thing I'm curious about in the industry in general is that we've been hearing for several years about artificial intelligence, deep learning or whatever, for autofocus. We've heard a lot about it, but we haven't really seen a lot of evidence in products. Is it there and we’re just not aware of it? How does AI fit in with what Fujifilm is doing?

Shin Udono: Of course, we are investigating what AI can do to help future picture-taking. It probably can be important part for future cameras. It's not only the AF, but also some other key features, which I cannot be specific about today, but you know, it can help in many ways for future photography.

[It's intriguing to me that Fuji is looking at applications of AI in photography beyond just autofocus, but I shouldn't be surprised.  Adobe has been using AI in its products since at least 2017, Apple supposedly uses AI for the portrait mode in iPhones and doubtless for exposure adjustment as well, and Google has been using AI in their Pixel phones for a several years now, most notably and integrally in their default HDR+ exposure mode. On the desktop, Topaz Labs has an entire suite of AI-based image enhancement products.

So there's clearly going on with AI and photography beyond just AF, but I wonder what form it might take in traditional, dedicated-purpose cameras? A lot will depend on what can be shoehorned into the cameras' image processors, and perhaps even more on what resources companies like Fujifilm can bring to bear on the problem. Outfits like Apple and even more so Google have vast troves of data to work with, and corporate budgets that are many multiples of Fuji's. On the other hand, Fujifilm has many decades of experience and large volumes of data of their own. Their Frontier mini-labs set the standard for the photo print industry for decades, and that business supported a lot of R&D about how to digitize images, color-space mechanics and what makes photos look good to human eyeballs. That's a pretty unique data set and knowledge base, and I can easily imagine it being used to train deep-learning systems with an aim toward improving things like color rendering, in-frame exposure and contrast control, image sharpness and noise reduction.

What's beyond that though? As I sit here writing this, I have a strong feeling that the above just scratch the surface of the sorts of things Fujifilm's engineers are considering. How about you, the reader? I'd love to hear your thoughts in the comments below about outside-the-box ways that AI might impact photography in the future.]

Dave Etchells: So for Fujifilm you see it's coming, it'll have value, but it's still in the future a little bit.

Shin Udono: Mmm. So we need to look at both the hardware side and the software side. Yeah. The current processor is not powerful enough to cope with AI. But we think it will be included in cameras in the future.

Dave Etchells: Another technology that we seem to be on the threshold of is 5G. What's that going to do to photography?

Shin Udono: 5G sounds very exciting, but we need to be realistic as well. There are a lot of limitations for 5G. When it comes to how 5G can help future image capturing or sharing, we may see it first for more industrial fields. Some close circumstances [meaning short-range, close to the cell sites], maybe 5G can be adopted earlier.

Dave Etchells: Ah, so 5G range is limited for in terms of the highest bandwidth, and so where you're in a local area...

Shin Udono: Yeah. We call it local 5G, that could be the first application, before it goes to the more consumer level.

Dave Etchells: Mmm, mmm-hmm. Yeah, I've thought that there will be a big change in latency and bandwidth and everything, and so I thought "Oh, maybe you will push more processing to the cloud, and the camera can talk back and forth." But then if you're out on a hike or in a rural area, all of a sudden your camera’s best features go away if it's dependent on 5G, so... yeah.

Shin Udono: Yeah. Mmm.

Dave Etchells: It's not clear to me at all how that will play out for consumers...

I have many questions about IS. One big one is whether the body IBIS can cooperate with the lens, so that they are aware of each other, or the body is aware of the lens. With the GFX, we've heard both that it does and does not cooperate. And likewise we haven't gotten a clear statement on the X-T4 IBIS. So can you clarify that whole area for us?

Toshi Iida: For the X-T4 and X-H1, they can cooperate to increase image stabilization performance.

Dave Etchells: The body IS does cooperate with lens IS.

Toshi Iida: Yeah. But for the GFX 100 we don't have the cooperative stabilization because OIS can deliver almost all the stabilization performance.

Dave Etchells: Ah, so if you have an OIS GF lens, it's as good as you would get anyway.
[I wonder about video shooting, though; cooperative IS seems to have a much bigger impact on handheld video shooting than it does on still-photo IS performance.]

Toshi Iida: Yeah.

Shin Udono: Yep. But in case of the X-T4 and XF lenses, their new IBIS capability is sometimes able to exceed OIS, so that's why we can push the number of stops of stabilization from 5 or 5.5 stops before to 6.5 now. It all depends how big the image circle individual lenses have.

Dave Etchells: Oh, *that's* why.

Shin Udono: That's the simple answer. Some lens [have a] big image circle, so the IBIS unit can [move further to stabilize the image]...

 

Iida-san's explanation that the maximum stabilization possible with an IBIS system depends on the lens's image circle answered a question I'd had for a long time, about why IBIS ratings varied depending on the lens in use. The image above illustrates the concept of an image circle. At some distance from the center of the image, the light projected by the lens begins to fall off, and image quality invariably suffers as well. In the illustration, the white rectangle represents a hypothetical image sensor, showing an image circle that's just barely enough to cover the full sensor area. Some lenses are able to deliver good image quality well outside the area required by the sensor. With such lenses, the IBIS system has more room to move to compensate for camera shake before image quality is impacted.
(Image by nagualdesign Original: Balkhovitin - Derivative of File:Ніжний ранковий світло.jpg, CC BY-SA 3.0, Link)

Dave Etchells: Ahhh! Yeah, that was a big question, because your specs will say that a body has 6.5 stops with this lens, but with another it’s 5.5. Ahhh! A simple explanation! That's very interesting. And that's a first: Other manufacturers will just say "up to 6.5 stops", but you don't know with a given lens...

Shin Udono: ...which lens can help you, yes.

Dave Etchells: Yeah. Very, very interesting. Huh. And then my next question is how are you able to make the image stabilization 30% smaller with X-T4.

Jun Watanabe: *shows stabilization units* This one is X-T4, this one for H1. The new unit is 30% smaller and 20% lighter. This is because we removed the coil spring. We had three here [that] we removed. We changed to magnetic coil for X-T4.

 

The IBIS module in the X-T4 (left) is dramatically smaller than that from the X-H1 (right). A number of separate advancements contributed to the size reduction.

Dave Etchells: Ohhh! So instead of coil springs, it's a magnetic coil.

Shin Udono: Yeah, So that's the biggest part.

Jun Watanabe: Another one is the inside structure. For X-H1, we have six magnets. Three for sensing, and three for moving position. Now we have only three magnets doing both jobs, sensing and position. Mainly, it is these two pieces which allowed the size and weight reduction. Another one is the circuits. We made the circuit inside for the X-T4, but for the X-H1 the circuit is here. So the space…

 

A bit part of the size decrease involved eliminating the separate magnets required for position control and sensing. As seen on the right, the X-H1's IBIS system used a total of six sets of magnets. In the middle, three pairs of magnets provide the field used by the Hall-effect sensors to precisely determine the current sensor position, while the three larger single magnets arranged on the left and bottom provide the magnetic field that the drive coils react against to move the sensor. On the left, the matching assembly from the X-T4 combines the two functions into one, with just three pairs of magnets arranged on the left and bottom. The same magnets used to help position the sensor also provide position information. I was amazed that the Fuji engineers could compensate for the field coming from the drive coils accurately enough to extract precise position information from the Hall-effect sensors.

Dave Etchells: Oh, I see. So the X-H1 has to fold and come down, but the new one is all inside. Ahhh, hai.

Jun Watanabe: This is the main reason we were able to downsize.

Shin Udono: So we downsized by 30%, and 20% lighter...

Jun Watanabe: But eight times more accurate.

 

I don't have a good comparison photo showing the difference relative to the X-H1's IBIS system, but this shot shows the flex circuits Udono-san mentioned, that let them fold the circuitry back on top of itself, to save both area and vlume.

Dave Etchells: Oh, eight times more accurate? Oh!

Jun Watanabe: Yes. Because we developed a new gyro sensor.

Dave Etchells: So will the eight times better accuracy that translate into sharper pictures with small vibration, or is it just that you get more stops of compensation?

Toshi Iida: More stops.

Dave Etchells: So it can detect much slower rotation than previously. Wow. Eight times though, that's amazing. That's three stops. Hmm!

Jun Watanabe: Eight times, three stops regarding the... [switches to Japanese]

Dave Etchells: Yeah, two, four, eight. Hmm, maybe sensor is three stops but system two stops, or...

Toshi Iida: We'll check what eight times means.

Follow-up answer from Fujifilm: 8x better performance means that X-T4 has 8x more detection accuracy of vibration than X-H1. Because we suppress the noise on the new gyro sensor works by 1/8 of current level. And IS performance depends on the size of image circle. By increasing the detection accuracy, it is now possible to correct blur even at the edge of the image circle.

[Ed. Note: It seems like the X-T4 would have 3 stops more IS improvement, but its official 6.5 stop rating is only one stop better than the 5.5 stops that the X-H1 claimed. I think it might be the image circle of the lens that's limiting the official IS rating using the CIPA test methodology, though, rather than the gyro noise. It's possible that you might see a fair bit more improvement between the X-H1 and X-T4 with wide-angle lenses, where the sensor won't have to move as far to compensate for a given amount of camera shake. The 8x improvement basically means that the new gyro has 8x lower drift, so it can keep things steady for 8x longer. It's just a matter of whether the maximum amount of camera movement (angular displacement, actually) during that time falls within the range that can be physically accommodated by the IS unit and the lens's image circle.]

Dave Etchells: Ah, OK. Well that's great; these are very concise answers to my questions! It must be very tricky, though, to have the position sensing and drive using the same magnets. Won't the drive signal interfere with what the Hall sensor is measuring? Or do you switch very quickly? How can the Hall sensor accurately detect the position when the drive coil is also contributing to the field? What was the trick they found to let them use just one set of magnets?

Toshi Iida: Maybe we'll ask our expert on it for an answer.

Dave Etchells: Yeah, I'm curious, because that's a big difference. I mean, as an engineer I think "That's really difficult"! Huh. Maybe the system knows what the strength of the field is from the coil, and then it can compensate. Ahhh, whatever. If you're able to tell me, that would be interesting.

Follow-up answer from Fujifilm: The IBIS unit is downsized by arranging the Hall element for position detection in the coil, but as you know, when a current is passed through the coil, a magnetic field is generated by that current, which affects position detection. However, since the amount of coil current is known by the control unit, the effect on the magnetic field from the coil current and the effect on position detection can be calculated and corrected by the controller.

[Wow, I'm surprised that this can work to the level of accuracy needed, and am sure there was some tricky engineering involved. In order to deliver sharp images, an IBIS system has to control the sensor position to within just a couple of microns or millionth's of a meter. (For our readers using English units, two microns is about 80 millionths of an inch.) So the Hall-effect sensor has to be able to detect changes in the IBIS system's magnetic corresponding to that tiny a movement. But at the same time that it's trying to measure the magnetic field from the magnets, the drive coils are creating their own magnetic field to move the sensor. So the system has to include the coil's field into the calculations and compensate for it. Helpfully, the field produced by the drive coils will be much smaller than that from the powerful rare-earth magnets, so it'll be a second- or third-order effect. But still, given the level of precision required, I'm amazed that they could subtract-out the effect of the drive coils accurately enough for the whole system to work. I suspect that part of the solution involved a careful arrangment of the magnets and Hall-effect sensor so the field from the drive coils would have the least possible effect.]

Dave Etchells: The shutter in the X-T4 seems like a significant advance. It manages to get to 15 frames per second from a mechanical shutter, but it’s very quiet. How did you accomplish that, while also delivering a cycle rating of 300,000 actuations?!

Jun Watanabe: We have three reasons for the X-T4's performance. The first one is that we have developed a new coreless DC motor for quick starting and stopping, and more torque. And the second one is more for more durability and reliability, we analyzed the X-T3's shutter unit deeply, to see which part is weak, etc. Based on the study, we changed the materials of the shutter unit, the plastic material. And we analyzed the weak points, then we changed the position of the injection into the mold.

 

The X-T4's increased continuous shooting speed and is largely the result of a newly-developed coreless DC motor that drives the shutter blades. (The motor is the large silvery cylinder on the left.) So-called coreless motors have less rotational inertia of their own to overcome, so are able to respond much more quickly to drive currents. 

Dave Etchells: Oh, really? Huh!

Jun Watanabe: In the mold. Difficult to break.

Dave Etchells: Ahhh! How interesting. So you're finding failures right near the injection point...

Jun Watanabe: Yes, we changed the injection point. We changed the mold performance; more rigidity.

Dave Etchells: That is really, really fascinating. Yeah, so I guess you can see where the injector pins pushed, but I don't think from just looking at it, you can tell where the injection point is? It doesn't show up.

Jun Watanabe: Yep, inside.

Dave Etchells: Oh, it's inside, you can’t see it. Ah, hai. So interesting!

 

The dramatic increase in shutter cycle life apparently came down to the subtlety of just where plastic was injected into the mold used to form the frame of the shutter assembly. This is a zoomed-in version of the shot shown above. The large circular depressions are where the "ejector pins" were located that push the part out of the mold once the plastic has cooled. What I find interesting here, though, is the light and dark striations generally radiating out from the lower right of the image. I think these show how plastic flowed into the mold when the part was made. Structural parts like this are made from so-called "glass-filled" plastics. This means that fine glass fibers are mixed with the plastic, greatly increasing its strength and stiffness. It makes sense that the glass fibers would align in the direction of the flowing plastic as it's injected, and I think that's what produces some of the subtle contrast we see here. Analyzing repair data, Fuji's engineers were able to trace end-of-life shutter failures to cracks in this plastic frame, related to how the plastic flowed into the mold. By changing the injection point, they were able to dramatically increase the longevity of the system as a whole. 

Jun Watanabe: And the third point is the spring, the design of the shutter.

Shin Udono: Because strong torque sometimes affects noise and shock. So we put a newly-designed spring to absorb the shock.

Dave Etchells: Ahhh!

Jun Watanabe: The cushion is here, the number is the  same, but we changed the position because we wanted to keep the same silence [noise] level. We changed the position, and we used a different type of spring for each position. Much nicer spring performance. So the [sound] level of the shutter is one-third that of the X-T3.

Shin Udono: Not only the sound, but it also reduced the shutter shock, shutter vibration.

 

When we got our hands on an X-T4 sample, we were amazed by how quiet the shutter was, especially when rattling off 15-fps continuous bursts. Apparently a big part of that came down to how the shutter assembly was mounted in the body. Engineers used springs to isolate shutter vibrations from the body, reducing both audible noise and also the incidence of shutter shock-induced image blurring. Compared to the X-H1, the X-T4 has a different arrangement of springs, and also springs of different stiffness. As a result much less of the vibration caused by shutter actuation is transmitted to the body, so we hear less of it as we're using the camera. It seems that the X-T4 has fewer springs, and at least one that we can see on the right in the image above appears to be much more open and thus likely less stiff. It's amazing the amount of engineering that goes into modern camera design: I can imagine an engineer spending months just modeling the performance of these vibration-isolating springs, trying out different configurations and spring strengths. His or her efforts were well worth it, though; the X-T4 has one of the quietest focal-plane shutters I've ever heard.

Dave Etchells: You said it was a different type of spring, how is it different?

Jun Watanabe: Just the material and the...

Shin Udono: Tension.

Dave Etchells: The material and tension, yeah. So not a different type, like a leaf spring versus a coil spring...

Toshi Iida: The type is the same. The material and strength differ.

Dave Etchells: Yeah, I don't think we have seen a sample of X-T4 yet, or at least I hadn't when I left the USA. But just from the presentation, you're running continuous shutter and it's so quiet! And the 300,000 actuations is very impressive too. So now that you have the more compact IBIS, will that eventually come to X-Pro also, do you think?

 

The X-T4 is a pretty camera too, at least in my eyes. I like the classic look of Fuji's cameras, and appreciate the number of functions that are readily available on top-panel dials.

Shin Udono: I think with the current IBIS unit it would probably be difficult to use it for the X-Pro, space-wise. We need to think even more for the future.

Dave Etchells: And the X-H1 was where you first deployed IBIS. Will that remain a one-model product line, or will there be an X-H2 at some point?

Shin Udono: We continue to investigate future X-H cameras. The X-T4 is not a replacement for X-H1, so we keep our minds open about future X-H models.

Dave Etchells: Ah, OK. So you can't say specifically, but it's not your intent for the X-T4 to replace the X-H1. You see a separate market.

Shin Udono: But we will clearly differentiate from the X-T line.

Dave Etchells: So what is that differentiation? What separates X-T and X-H?

Shin Udono: Difficult to say now; we need some sort of the breakthrough, probably.

Jun Watanabe: But actually, after we announced X-T4, studios received a lot of customer's requests that they want that H1 style.

Shin Udono: Bigger grip, and also...

Toshi Iida: ...the command dial control.

Shin Udono: But just having the bigger grip with the X-T4 is not enough for the X-H2. I think that if future X-H models come to reality, we have to have something more revolutionary.

Dave Etchells: Ah. So X-H1 is maybe where new technology enters the product line.

Shin Udono: Yep.

[This is interesting. Their saying that they’d need some sort of a breakthrough to create a new X-H model suggests that they do in fact view the X-H series to be a platform for moving new technology from the lab into real products. This was the case with their IBIS tech; the X-H1 was the first Fujifilm camera to incorporate IBIS, with refinements of that system later appearing in the GFX-100 and X-T4.].

Dave Etchells: Ah, hai. That's very interesting. Yeah. Speaking of the command dial, my guys were asking why there aren’t four-way buttons on the X-Pro3 or GFX 100.

Toshi Iida: Yes. This is for its balance of the grip holding and the number of controls...

Shin Udono: For all individual products, we have a lot of debate and investigation in terms of buttons and the thumb position, or [how easy it is] to hold, [and not] to press the four-way button accidentally. For a camera like the X100V, we decided to delete it because if you have a four-way button there it's easy to accidentally press it. But for the X-T4...

 

We're big fans of four-way button controllers, so were happy to see one on the X-T4. It seems that whether or not to include one on any given camera model is the subject of a lot of debate. Part of that decision has to do with available space, and how likely one of the controls might be to be activated accidentally. It didn't come out in the interview (I think we discussed it later, either during a lab tour or at dinner), but another consideration is that some users actually dislike them. So rather than trying to make the user interface identical on all their cameras, Fujifilm deliberately designs some to fit the preferences of a significant minority of users who prefer a different interface design. This was apparently partly why the X-Pro models don't use four-way controllers. (Space is limited on their rear panels as well, but I found the general idea of making cameras to fit different individual preferences rather than taking a one-size-fits-all approach interesting.)

Toshi Iida: We have space for the four-way controller.

Shin Udono: Yes, without misoperating.

Dave Etchells: It's just a little bit, little bit taller maybe, and then you have the little lip there, that kind of keeps your thumb up and off of it. But even the GFX 100... That has a very large back, but it’s the same issue in terms of where you might put your thumb.

Shin Udono: Yes.

Jun Watanabe: Also, this four-way keyboard layout is very important. That has an impact on the other controls. If we have the four-way key here, also we have to locate these other three buttons somewhere here. <gesturing, on the back of the camera>

Dave Etchells: Yeah, yeah, you have to put those someplace. Yeah.

Jun Watanabe: So with that, it's really messy.

Dave Etchells: Yeah. Yeah, we personally like four-way controls a lot, myself and the editors. But I understand the point. It's well taken.

So quickly on lenses, I'm wondering about what direction the GF lens line will go in? You just announced the 45-100mm, and that filled a good gap. The roadmap shows just one new lens in 2020, the 35mm f/3.5, and one for 2021. I'm struck by how much narrower the range of lenses is for GF versus XF. How do you view the X and GF lens lines?

Shin Udono: Yeah. Naturally, I think the GF requires a smaller number of lenses compared to the XF lens lineup, because with the XF we need big telephoto zooms, for example, to meet the demand for sports photographers. But for GF it's a slightly different type of customer who demands a smaller range [to be satisfied]. But with the current lineup, we understand the demand for more different types of lenses for the GF as well. So we keep studying and listening to customers, and we'll decide on [our] future roadmap.

Dave Etchells: So the fact that there are only two lenses shown over the next two years on the roadmap doesn't mean that you might not choose to add others; that could happen?

Shin Udono: We keep listening, yeah.

 

Early lens roadmaps included a 33mm f/1.0 lens, but that's now been dropped in favor of a 50mm version.  The reason for the change was a combination of the size and weight of the 33mm design, and also the much better bokeh the optical engineers could produce with the 50mm focal length. This was a rarely-seen glimpse into the process of lens design: It seems that some designs might proceed quite far in development before being scrapped in favor of other approaches. I'd love to know how many lenses are actually designed that never see the light of day as products?

Dave Etchells: Hai. We're curious, you have a very robust XF line, and the only new model on the roadmap is the 50mm f1.0, which I guess takes the place of the 33mm f/1.0. Why did the 33mm run into trouble, and how did moving to 50mm make it feasible?

Shin Udono: I think that when we started designing that lens, it turned out to be very big and very heavy, so size and weight is one reason. The other reason is that we found more benefit in terms of the bokeh effect. Compared to 33mm, the 50mm has a much more beautiful bokeh. So we are sorry to change, but we decided it's the right decision to provide customers a smaller, more easy to carry lens with better bokeh.

Dave Etchells: Mmm. So it wasn't a manufacturing problem. It was just that as you got into the design it just got so big, and then you discovered the bokeh would be better at 50mm. Very interesting.

Talking about camera design again, the X-Pro3 has a very different design on its back. It's got a little display, but then the LCD only tilts down. What led to that design? Was that something you were hearing from customers, or that you decided upon internally?

Shin Udono: The initial design came from our team, and we asked several times for opinions from photographers, especially for street  photographers or users of the X-Pro1, X-Pro2 or X100 series. Then we made a decision to go for that design. Because at that point, we had already started the project for X-T4. There is no point in making the X-T4 and X-Pro series too similar to each other. So X-Pro has a totally unique identity.

Jun Watanabe: When we asked photographers [about] that LCD style, it has a clear separation [between] like or dislike.

Shin Udono: Love or hate.

Jun Watanabe: Yeah, it was 50/50 or even disliked was maybe a little bit higher. But we thought that was a good signal, because we can get this [sizeable minority of customers who liked the design]. X-Pro3 is exactly the product for that type of customer.

Dave Etchells: That's really interesting. So there's customers that loved it or hated it. Customers that hated it would get an X-T4. People that love it, there's nothing else like it anywhere.

Shin Udono: Yes, sure.

Dave Etchells: Ah, that is really, really interesting. Yeah, I wouldn't say I hate it, but I'm probably more an X-T4 guy. I thought the little color display on the back panel is kind of cute. It reminds me of a film box.

Jun Watanabe: Yeah, yeah. That's exactly the film camera style.

 

I was happy to hear that the X-T3 will continue to be sold alongside the X-T4. The X-T3 was an excellent design with great capabilities and many advances. (I think it has one of the most sophisticated AF architectures of any model on the market, in terms of what's going on at the sensor level.) We at IR thought highly enough of it to crown it as our overall Camera of the Year in 2018.

Dave Etchells: Yeah, yeah. Film cameras used to have the little slot that you'd tear off the top of the film box and put it in, so you’d remember what film you had loaded. That's very interesting, I've never really thought about product planning and development like that, that instead of making every camera for everybody, you have specific designs for different groups. It’s OK to make a camera some people hate if it’s unique and some people love it.

Turning to the X-T line, will the X-T3 stay in the market now that the X-T4 is out, or is it going?

Shin Udono: It will continue as long as demand continues. We will keep manufacturing and supply.

Dave Etchells: So it'll be at a lower price-point.

Shin Udono: Yep.

Dave Etchells: Yeah, I think that makes sense. The X-T3 is a very good camera; it was our camera of the year just a year or two back.

I think I may have covered all of my questions. This was so interesting! Image circle size affects IBIS performance! When you said that, I was like "Ahhh!"

<laughter>

Shin Udono: Yeah, that is another benefit for smaller-format sensors, because we have the big-enough image circle so that the sensor can be shifted.

Dave Etchells: Yeah, yeah. Ah, that's really interesting. Although that's one area, I think, where for video, having the cooperative IS -- IBIS and IS working together -- helps a lot. You can get more motion.

Shin Udono: Yes.

Jun Watanabe: We have a new [Digital Image Stabilization] function for video. <demonstrates function on the camera>

Shin Udono: Especially for movie shooting and walking...

Dave Etchells: Yeah. That looks very good. Often, when I've seen digital IS from other people, there are many artifacts. It looks like very bad rolling shutter, but this looks very clean. Hmm, that seems quite good. And so DIS is in the X-T4?

Shin Udono: Yes.

Dave Etchells: Hai. And only that model, it will not be a firmware update to previous ones?

Jun Watanabe: No. It probably requires the new gyro sensor.

Dave Etchells: OK, I think that's finally all my questions, thank you very much!

Fuji (all): Thank you!

Summary

As always, this was an information-packed conversation; I've always been impressed by the extent to which even Fujifilm's upper managment understand the deep technical details of their products' engineering. 

Regarding the general market, I was happy to hear both that Fuji's production is back to full strength, and also that the market does seem to be recovering. Citing CIPA numbers for mirrorless sales, Fuji told me that the bottom occurred in April, with sales off 68%(!) relative to last year, recovering to -58% in May and -37% in June. There's still a long ways to go to a full recovery and I can only imagine how devastating month of sales off by 60 - 40% must be on a business, but it was good to hear at least some positive signs. In the longer term, CIPA has been projecting an overall 20% drop relative to last year, but Iida-san said that he felt this was overly pessimistic. (I think he's probably talking about the baseline trend, absent the impact of COVID.) 

Talking about their own product line, the X-T4 is apparently being very well-received, and I was happy to learn that the X-T3 will continue to be sold alongside it, as long as there's demand for it. On the GFX front, I was surprised to learn that the top-end GFX-100 is actually the highest-selling of their medium-format trio. While it's the most expensive model by a good margin, it's still vastly more affordable than any of its competitors in the 100MP+ market, and there appears to be a lot of demand for it, particularly in the archive-imaging business. (Think in terms of museums digitizing their collections.) 

In another bright spot for Fujifilm, it appears that 70-80% of their GFX customers are coming from other companies' platforms.

We talked quite a bit about image stabilization, and I not only learned a lot of details about advancements in Fuji's IBIS technology, but also a question that had been puzzling me for a while. Unlike most manufacturers, Fujifilm publishes IS ratings for various lenses separately (as opposed to just saying "up to X.X stops"). I expected IS performance to vary for different focal lengths, but was surprised to see different ratings for various Fuji lenses with identical or closely similar focal length ranges. Udono-san's explanation that it came down to the size of each lens's image circle was a revelation, not to mention a bit of a smack-my-forehead moment for me :-)

From their original IBIS system in the X-H1 to their latest in the X-T4, Fuji's managed to make a huge reduction in size and weight. They did several things to accomplish this, including folding some of the circuitry over on itself, replacing a mechanical spring with an electromagnetic coil, and cutting the number of magnets in half. This last was particularly interesting to me: Usually, you need a total of 6 magnets (well actualy twelve, in six pairs). Three magnets provide the fields that the drive coils work against to move the sensor, and the other three provide fields that high-resolution Hall-effect sensors measure, to determine exactly where the image sensor is located, moment by moment. Fuji's engineers found a way to use one set of magnets to provide both functions. This seemed incredible to me, because the magnet fields from the drive coils will be superimposed on those that the Hall-effect sensors will be trying to use to measure position. It seemed like it would be impossible to separate the two influences from each other, but after querying the engineers, Fujifilm told me that they do it by knowing precisely how much current is flowing through the drive coils, and then subtracting the calculated contribution of that from the position readings. I'm amazed that they can do that precisely enough to maintain the micron-level accuracy required to produce crisp images on a 26-megapixel APS-C sensor.

We also talked in some detail about the X-T4's remarkably quiet and long-lived shutter; blasting away at 15 fps, it's by far the quietest focal-plane shutter I've ever heard. A tiny new coreless-motor actuator simultaneously shrunk the size, increased the speed and reduced noise relative to earlier models. The other part of its quiet performance is thanks to a reconfiguration of the shock mounts that connect the shutter assembly to the camera body itself. A set of springs help keep vibration from the shutter mechanism from being transmitted to the body; by changing the arrangement of the springs, and the strength of some of them, Fujifilm engineers were able to significantly improve the shutter/body mechanical isolation, thereby reducting the noise we hear while operating the camera. (I haven't tested it, but it stands to reason that the reduction in body vibration should reduce shutter-shock as well.)

Another big advance in the X-T4's shutter mechanism was increasing its cycle life, from 150,000 cycles in the X-H1 to 300,000 in the X-T4. I was very surprised to learn how they did this: Analyzing many (hundreds? thousands?) of shutter mechanisms with end-of-life failures, the engineers discovered that many of them were the result of cracks developing in the structural-plastic frame that the various parts are mounted to. Many of these occurred in the same locations, and deep analysis led the engineers to trace the issue back to how the hot plastic flowed as it was injected into the precision molds used to form the part. By changing the injection point, they were able to double the rated life of the shutter. (I expect there may have been other design tweaks as well, but Fujifilm pointed to the change in injection point as having the greatest impact.)

We also spoke briefly about control layout and UI variations and I learned that, while we at IR have a strong preference for four-way controllers, some people actually dislike them. The decision about whether to use a 4-way controller or not is apparently a constant point of discussion within Fujifilm :-) Sometimes it just comes down to whether or not there's space on the rear panel to include a 4-way button group without risking accidental actuation by the user's thumb, but Fuji consciously caters to the sub-group of people who prefer a different control layout. This struck me as a unique approach: Rather than deciding on a single, corporate UI design, they intentionally create different camera models for different groups of users. Smart.

Finally, I was curious what led to the change from the 33mm f/1.0 XF lens that appeared on earlier lens roadmaps to the 50mm f/1.0 that now appears in its place. There were apparently two reasons for the change. As they got into the design, it turned out that the 33mm was going to be very large and heavy, to the point that it would be unweildy on Fuji's compact APS-C camera bodies. The other factor was that the optical engineers found they could achieve much better, more appealing bokeh with the 50mm focal length. Since much of the point of an f/1.0 optic comes down to its ability to throw background objects out of focus, greatly improved bokeh was a strong reason to switch to the 50mm focal length.

Phew, I think that about sums it up. This was an unusually informative discussion, with lots of the juicy technical tidbits I personally enjoy; I hope it was the same for you!

Let us (and everyone else :-) know your thoughts on all the above in the comments below, and feel free to pose any questions you'd like. As usual with my articles, I'll try to pay particular attention to the comments thread for the next week or two, to make sure I fully respond to any requests.