After I interviewed Fujifilm executives at their headquarters in Omiya, Japan again in July, there have been quite a lot of upcoming bulletins that they couldn’t but speak about, and so they recommended I drop by their X-Summit/Fujikina occasion in New York Metropolis in early September.
I did simply that, and it turned out to be a fairly important occasion, with bulletins together with the X-H2 high-resolution (40 megapixel) digicam physique with 20fps taking pictures, video as much as 8K/30p, digital shutter speeds as excessive as 1/180,000 second and way more) a brand new model of their already-excellent 56mm f/1.2 lens, a surprisingly compact and light-weight ultra-wide 20-35mm (35mm equal 15.8-27.7mm) zoom lens for the GF mount, and the shock announcement of a 110mm tilt/shift lens, additionally for the GF line.
I met with Yujiro Igarish, Divisional Supervisor, Skilled Imaging Group, Imaging Options Division, above proper, and Jun Watanabe, Product Planning Supervisor, Product Planning Group, Skilled Imaging Group, Imaging Options Division, above left.
I met with Yujiro Igarashi, Divisional Supervisor, Skilled Imaging Group, Imaging Options Division, and Jun Watanabe, Product Planning Supervisor, Product Planning Group, Skilled Imaging Group, Imaging Options Division. Igarashi-san has accountability for Fujifilm’s international digicam and lens enterprise, together with the Fujinon broadcast lenses. Watanabe-san is liable for product planning, so his group was liable for the design of the brand new X-H2 digicam.
I had loads of questions concerning the X-H2 and new 56mm lens specifically, however we additionally talked some concerning the new 20-35mm and 110mm tilt/shift lenses as effectively.
Learn on for all the main points, and depart any ideas or questions you may need within the feedback beneath. I’ll make a degree of checking the feedback frequently over the subsequent week or so, and can reply any questions you may need to the extent of my information.
(Within the beneath, in some locations I’ve summarized the dialogue for the sake of readability, whereas in others I’ve quoted their phrases straight. My summaries seek advice from Fujifilm within the third-person, whereas the direct quotes are first-person and set off in citation marks.)
Sign/noise ratio and dynamic vary of the X-H2 vs the X-H2S
Due to some intelligent pixel design, the X-H2 sports activities a base ISO of 125, which provides it a greater signal-to-noise ratio (SNR) than the X-H2 at it is base ISO of 200. Regardless of its smaller pixels, the X-H2 additionally does very effectively at increased ISOs, with solely barely decrease SNR at ISO 200, and almost equivalent SNR at ISO 1600.
RDE: The decrease base ISO of 125 looks like a major change. It’s spectacular sufficient that you simply managed to realize solely barely decrease SNR at ISO 200 than with the X-H2S, however the leap in SNR when going to ISO 125 is fairly dramatic. Are you able to say something about how that is completed within the sensor? (Does it imply a deeper construction for the photodiode? Another change within the pixel cell construction?)
Reply: “Despite the fact that the pixel pitch is narrower than earlier sensors, by rearranging the pixel structure (the amplifier circuitry, and so forth), and utilizing a wider construction for the photodiode, the pixel can now maintain extra electrons. This implies it will probably accumulate extra mild earlier than saturating, so we may obtain the ISO 125 functionality. These {hardware} modifications additionally helped hold the dynamic vary of nonetheless photographs at about the identical stage, regardless of the bodily smaller pixels.”
What’s the key to the 1/180,000 second shutter velocity?
A giant characteristic of the X-H2 is its astonishing 1/180,000 second most shutter velocity. (Expressed one other means, that is an publicity time of simply 5.5 millionths of a second.) This implies you’ll be able to shoot quick lenses large open even in full solar, for super-shallow depth of area.
RDE: Digital shutter speeds extending as excessive as 1/180,000(!) are astonishing. As you identified, that is going to be an amazing characteristic for individuals wanting to take advantage of the super-wide f/1.2 aperture of the 56mm underneath brighter lighting. What’s the key to reaching that? What’s the limiting issue for brief exposures with digital shutter, such that that is the primary time you’ve managed to realize this? (Is it a matter of shutter-signal propagation throughout the sensor’s floor?)
Reply: The brand new sensor provides way more precision in setting the digital shutter publicity time. That extra exact management is what enabled the extraordinarily excessive shutter speeds. Particularly, they stated: “Till now, the accuracy of publicity time management within the ultrahigh-speed vary has been low, and management in 1/3-step increments has been troublesome. This time, as a operate of the brand new sensor, we had been capable of enhance the power to exactly management the timing of reset and readout within the publicity time, and realized an digital shutter velocity of 1/180,000 sec. on the quickest.”
I used to be curious what it was concerning the new sensor that led to the way more exact timing management. Is there some extra primary architectural change in both the pixel construction or the help circuitry within the array that enabled the extra exact management, or is it one thing completely different altogether? My first thought was that possibly it needed to do with getting the shutter sign to propagate throughout the array quicker, however as I’ve thought of it extra, I believe that what they’ve accomplished is integrated some kind of extra subtle timing circuitry onto the sensor chip itself. I after all know nothing about both their present or former sensor designs, but it surely strikes me that they may have accomplished one thing like placing a higher-frequency clock oscillator and logic circuitry onto the sensor (or the stacked chip) itself, reasonably than sending the shutter sign from the principle processor straight. That is all blue-sky hypothesis, however I strongly suspect that the answer concerned placing timing circuitry within the sensor stack itself, with a purpose to obtain the dramatically improved precision they spoke of.
HR pixel-shift mode photographs should be assembled on a pc?
RDE: Pixel-shift excessive res appears to work very effectively (I used to be impressed by how far Jun-san may zoom in on the pattern picture he confirmed within the briefing, with pixel-level sharpness within the picture.) Jun confirmed it being pushed from a PC. I assume that the digicam can seize HR photographs by itself? Do you should assemble the excessive res picture on a pc, or can the digicam additionally do this itself?
Reply: “Sure, the digicam can seize the 20 units of photographs by itself, but it surely’s as much as a software program utility on a PC to sew them collectively; the digicam can’t do the stitching internally.”
This is sensible, the micro-alignment of the 20 separate exposures takes loads of computation. The outcomes are fairly spectacular although. In the course of the new-product briefing Jun Watanabe confirmed a picture taken utilizing the X-H2’s multi-shot high-resolution mode (video screenshots of it proven above), and I used to be stunned that it was as sharp because it was on the pixel stage. The lens after all has one thing to do with this, and Fujifilm’s design philosophy for lenses has all the time been to purpose for a single level of most sharpness, reasonably than a barely softer attribute at optimum focus that extends over a finite vary of journey of the main target ingredient (a extra frequent method amongst lens designers). I first discovered of this in an interview at CP+ in Japan a few years in the past, however couldn’t find the reference as I used to be penning this. A broader level of focus relaxes the calls for on the AF system, however ends in much less crispness than a design that goals for optimum focus to happen at a single level. Fujifilm’s method is well-suited to the trendy period wherein the picture sensor itself is used for focus dedication. In SLRs, the place the AF sensor was separate from the picture sensor, less-critical optimum focus helped to accommodate microscopic misalignments between the AF and imaging sensors.
Does the X-H2 use the identical processor because the X-H2S or is it completely different?
The X-H2 makes use of the X-Processor 5. Is that this the identical chip as within the X-H2S, or is it new?
Reply: It’s the identical.
How did they handle a blackout-free EVF with out a stacked sensor?
RDE: The X-H2 has 120fps blackout free dwell view: How did you handle that with out a stacked sensor? Is the dataflow structure the identical as within the X-H2S? (Within the X-H2S, the complete sensor is scanned and totally learn out at 120fps. The processor then outputs information individually for the dwell view/viewfinder and still-capture capabilities. The X-H2’s sensor isn’t a stacked design, so its readout goes to be slower than that of the X-H2S. In the event that they’re not scanning the total array at 120fps, how have they managed to provide a blackout-free Stay View that updates at 120fps?)
Reply: “The X-H2S can show live-view photographs at 120 fps as a result of the live-view photographs are skipped-reading, then again the X-H2S can show live-view photographs at 120 fps with full pixel depend.”
Ah, this is sensible: I consider that they imply that the X-H2 skips pixels for its live-view show. Since each the EVF and rear-panel LED are lower-resolution than the sensor itself, not the entire sensor pixels are required to create the live-view picture.
Why no 8K/60p possibility?
The X-H2 is not only a nice stills digicam, it gives a cornucopia of options for videographers as effectively. Codecs vary from 8K/24p to full HD at 240p. Whereas its dwell view show updates at 120fps, the utmost body fee for 8K video is 30p. The 30p restrict comes from the sensor’s readout velocity; the dwell view show is so quick as a result of it skips pixels; clearly not an possibility for video.
RDE: 8K/30fps is spectacular, however I’m curious why you’ll be able to’t do 8K/60, until you’ll be able to’t learn out the show that quick. (Or is it a processor limitation? – The EVF updates at 120fps, so why not video that quick?)
Reply: “The body fee is restricted to 8K/30P by the sensor’s efficiency limits (readout velocity).”
Once more, this is sensible: The dwell view show skips pixels to realize the 120fps refresh fee. This clearly isn’t an possibility for 8K video.
Is the body cropped when taking pictures stills at 20fps?
RDE: The digicam’s spec says it will probably do 20fps with digital shutter, however contains the notice (1.29x) after it. Does that imply that the 20fps frames are cropped by 1.29x from the total sensor body?
Reply: “Sure, at 20fps, the digicam shoots 24 megapixel photographs, with a 1.29x crop issue relative to the total body.”
It may appear odd that the digicam can seize 8K video utilizing the total sensor decision at 30fps, but it surely has to crop all the way down to 24 megapixels to shoot stills at solely 20fps. The distinction is the results of the very completely different picture processing and information pathways which might be used for nonetheless photographs vs movies. One distinction between the 2 forms of seize is that video compression and encoding is dealt with by devoted, extremely optimized {hardware}, which makes use of keyframes and frame-to-frame variations to vastly cut back the quantity of knowledge being handled early within the processing chain. Nonetheless seize has to deal with every body as a wholly separate picture, so it isn’t capable of benefit from the massive elements of the photographs that don’t change from one shot to the subsequent to achieve effectivity. That stated, twenty 24-megapixel frames a second would have been over-the-top efficiency for a “sports activities” digicam just some years in the past.
Does digital picture stabilization enhance when taking pictures cropped video?
RDE: Does utilizing a cropped video mode (4K or FHD) let the digicam’s IS system compensate higher for big movement, equivalent to when strolling with it handheld? (My thought right here is that digital picture stabilization ought to have extra sensor space to work with when it’s solely utilizing 50% of the sensor space, so it ought to be capable to compensate for extra digicam movement than the conventional IBIS.)
Reply: When doing digital picture stabilization, the X-H2 crops the sensor by 10%, then expands the video again out to regain the total picture dimension. Whereas it is likely to be doable to realize extra compensation in cropped-video mode, they don’t at present do it.
Interested by this query afterwards, I noticed that digital picture stabilization would possibly degrade the picture high quality considerably when utilizing 1:1 cropped video. Their regular method of cropping simply 10% of the picture after which increasing the outcome again out once more successfully provides them the power to make sub-pixel corrections for digicam motion, producing a lot smoother video.
Taking pictures 4K video utilizing the total sensor, there are barely greater than two sensor pixels vertically and horizontally for every pixel within the ultimate output. (2×2, that means a complete of 4 enter pixels for every output one.) Cropping by 10% means there are barely fewer than two sensor pixels in every path for every pixel within the output. This is not a big distinction although, so if the digicam shifts the readout location on the sensor’s floor by one pixel, it solely quantities to a bit greater than half a pixel within the output picture. This means to manage the readout place in roughly half-pixel increments makes for very easy output video.
All that stated, although, it’s not clear whether or not or not it is likely to be doable so as to add a particular digital picture stabilization mode sooner or later, that might benefit from the larger margin that cropped 4K video would offer.
Even lenses launched in 2021 had been designed with the 40MP sensor in thoughts!
This wasn’t a part of my questions, however I assumed it was value mentioning right here for many who may not have watched the rollout occasion. It seems that Fujifilm has been planning for the 40 megapixel sensor for some years now. The XF 18/1.4, 23/1.4 and 33/1.4 lenses launched in 2021 had been all designed to ship the upper decision demanded by the 40 megapixel imager. On condition that it will probably take 2-3 years for a brand new lens design to make it from preliminary idea to a delivery product, Fujifilm has been planning for the arrival of the X-H2’s sensor for fairly some time now.
Was the 70cm minimal focus of the unique 56/1.2 restricted by picture sharpness?
The unique XF 56mm f/1.2 lens was already super-sharp and one in all our very favourite XF lenses, however Fujifilm has additional improved it within the new XF 56mm F1.2 R WR mannequin. The brand new design is kind of completely different than the unique; it appears to be like to be an entire “clear sheet” redesign. The picture above reveals the structure of its lens components; the tan ingredient is created from ED glass, whereas the 2 blue ones are aspherical. In addition to improved optical efficiency, the brand new mannequin can focus all the way down to 50cm, vs the unique’s 70cm.
RDE: The press launch famous that the MOD (Minimal Object Distance) was lowered to 50cm attributable to “improved resolving functionality at minimal object distance (MOD)” Was the earlier model restricted to 70cm as a result of it could have simply gotten too gentle or had too-high aberrations at nearer focal distances?
Reply: Sure, it could have been bodily doable to let the unique 56/1.2 lens focus nearer, but when they’d accomplished so, it wouldn’t have delivered the extent of sharpness they really feel {that a} Fujifilm lens ought to present.
A fast facet notice right here, on the time period “minimal object distance”. I’m extra accustomed to listening to the time period “minimal focal distance”, so I Googled MOD and located conflicting definitions for it. Some sources say that MOD refers back to the minimal distance between the topic and the entrance ingredient of a lens, vs minimal focal distance which refers back to the minimal distance from the topic to the focal airplane. Different sources stated the 2 phrases are interchangeable. After I requested the Fujifilm of us about it, they stated that they deal with the 2 phrases as synonymous. – So the minimal focal distance for the brand new 56/1.2 is 50 cm from the digicam’s focal airplane to the topic.
What led to the improved bokeh of the brand new 56/1.2?
RDE: There was loads of emphasis within the presentation on the bokeh of the brand new lens. How a lot of the optical design (the essential association of components) was pushed by a quest for higher bokeh?
Reply: The brand new optical design may need had some impact on the bokeh, however the principle objective of the brand new design method was to cut back aberrations and improve most sharpness. The a part of the design that was particularly aimed toward bokeh was the 11-blade aperture.
I requested this query as a result of I do know there’s much more to bokeh look than simply how round the aperture is. Even with a wonderfully round aperture, you’ll be able to nonetheless get issues like bright-line or onion-ring bokeh or different unappealing traits relying on the optical design and fabrication of the weather. I believe that the designers of the brand new 56/1.2 gave no less than some thought to bokeh look once they had been developing with the optical system, however that apparently wasn’t the first motivation for the brand new design.
(The bokeh actually does look attractive, by the best way; it’s gentle, easy and featureless.)
Does the 56/1.2 II’s aperture transfer together with the main target components?
RDE: The “focus group” within the new 56mm f/1.2 really appears to be two separate lens teams, one in entrance of and one behind the aperture. Do these transfer as a single unit, or are they individually pushed? If the previous, does that imply that the aperture meeting strikes together with them? That should add much more mass…
Reply: “The main focus lens teams work in unison, together with the aperture. These lenses contribute vastly to bettering the minimal focusing distance and picture high quality at shut vary.”
Wow, a 120 gram focus group, however a change to DC motors. What’s AF velocity like?
A serious distinction between the previous and new is the dimensions and complexity of the focusing ingredient. Amped-up DC motors and improved management algorithms let the brand new mannequin focus about as quick as the unique although.
RDE: I used to be stunned to see that the main target group within the new 56/1.2 weighs 120 grams, that appears crazy-heavy for a spotlight group! However I additionally see that you simply’re utilizing DC motors. Are DC motors extra highly effective than different approaches? I suppose the underside line is focus velocity; are you able to inform me how the brand new 56/1.2 compares to the earlier model, or to different lenses usually?
Reply: “As for the AF velocity, it’s equal to the present lens. Though the load of the lens group is significantly heavier, the identical efficiency has been achieved by rising the torque of the DC motor and bettering the lens drive algorithm.”
From my conversations with lens designers over time, it seems to be troublesome to take care of excessive picture high quality at brief topic distances. The answer appears to be way more elaborate focus models with extra lens components in them. Because of this macro lenses are usually slow-focusing; the main target motor has loads of glass to maneuver round to focus the lens.
I’ve develop into so accustomed to listening to about linear motors, electromagnetic actuators, stepper motors and the like that once they stated “DC Motor”, I used to be considering it was one thing new or completely different. It seems that is simply an old school miniature DC motor with a spinning shaft coupled to the main target group through gearing. That is the best way most AF lenses had been constructed from the start, till the arrival of issues like ultrasonic motors and different newer actuator expertise. Due to fashionable motor tech although, DC motors can pack a heck of loads of torque for his or her dimension, and extra subtle electronics and firmware lets Fujifilm drive the motor shortly and precisely.
Will we be seeing extra WR (Water Resistant) lenses going ahead?
RDE: I additionally observed the advertising supplies for this lens additionally made a degree of it being climate resistant. Is {that a} characteristic we’re going to be seeing extra of in your lenses sooner or later?
Reply: “Typically talking, all our fashionable lenses are climate resistant. The primary technology weren’t, however sooner or later they we determined to make it an ordinary characteristic, so any current lenses are constructed that means, and future ones might be additionally.”
What’s the key to the 20-35mm’s secret, being so light-weight and compact?
The GF 20-35mm F4 R WR is remarkably compact for a wide-angle zoom overlaying a medium-format picture circle. It is nearly the identical dimension because the 23mm f/4 prime lens and solely barely costlier. It looks like an amazing possibility for wide-angle shooters.
RDE: The GF 20-35mm f/4 dimension and specs are superb: It’s really lighter than the 23mm f/4 and nearly the identical dimension (1mm smaller diameter, solely ~9% longer, and fewer than 10% costlier. What was the trick to have the ability to make it so compact? I believe I noticed within the presentation that it makes use of an unusually advanced single cam to maneuver the weather, the place there’d often be two cams. Was that an essential a part of making it so compact?
Reply: “A very powerful side was the appliance of the identical small-size optical system idea as within the GF35-70mm lens.”
“There have been a variety of issues that helped. First, we selected an internal-zoom design, to make it simpler to hold. That introduced its personal constraints although, so we needed to work laborious on each the optical and mechanical design. Optically, we used loads of aspherics and ED glass, with three aspherical lenses, three ED lenses and one aspherical ED ingredient. Nano-AR coatings and inner mild masks had been used to reduce flare and ghosting. Maybe the largest issue within the compact dimension of the lens is that it makes use of a single mechanical cam to maneuver all of the lens components as you flip the zoom ring. The barrel of the cam has two completely different diameters, and usually it could have required two separate cam assemblies, however the engineers managed to mix them right into a single extra advanced cam that handles all of the zoom actions by itself. This lowered each the lens diameter and its weight. To additional assist with the load, we used 4 completely different supplies within the physique of the lens (magnesium, aluminum, brass and plastic elements), letting the engineers make the optimum alternative of weight, rigidity, price and manufacturability for every part.”
A fast facet notice about supplies utilized in lens our bodies. You would possibly assume that the perfect factor to do can be to only machine the whole lot out of aluminum, but it surely seems “all metallic” doesn’t essentially imply “higher”. Fashionable structural plastics have near the identical energy per weight as aluminum does, and could be molded to dimensional accuracies (a couple of thousandths of a millimeter) that will be troublesome and really costly to take care of when machining aluminum at excessive manufacturing ranges. In contrast, upon getting the molds made, fully equivalent structural plastic elements could be cranked out in excessive volumes very cheaply. – So structural plastic elements can really be extra exact than metallic ones
Trying in the wrong way, Magnesium’s mild weight and excessive rigidity relative to aluminum or plastic make it an amazing materials for shaving weight and dimension from precision assemblies, but it surely’s fairly costly and tough to machine. Used strategically, although, it’s superb how a lot it will probably stiffen up an meeting whereas on the identical time lowering its weight.
Lastly, brass is straightforward to machine, and its mechanical properties make it nearly ideally suited to be used in lens flanges. It additionally takes laborious chrome plating very effectively, offering a easy and extremely wear-resistant floor for elements which have to slip in opposition to one another, as is the case with lens mounts and flanges.
Lens manufacturing is an intricate and engaging course of; for those who’d wish to be taught extra about it, take a look at my tour of Sigma’s Aizu manufacturing unit, and a follow-up piece once they added magnesium machining to their capabilities.
The 110mm tilt-shift was an enormous shock, what’s the story with that?
The announcement of a 110mm tilt-shift lens for the GF mount was an enormous shock. They’d been displaying a 30mm tilt-shift on the GF collection lens roadmap for some time, however had by no means talked about the prospect of a 110mm till Igarashi-san flashed one on-stage on the X-Summit, as “another factor”, within the spirit of Steve Jobs. Architectural photographers (amongst others) will rejoice at having not only one however two tilt-shift lenses, overlaying each very large and short-tele focal lengths.
RDE: I used to be very stunned to see you announce a 110mm tilt-shift lens for the GF mount; you’d proven a 30mm tilt/shift on the lens roadmap for a short while, however by no means a 110mm. How did that come about? Was the 110mm a newer resolution and extra shortly designed, so that you simply launched it with out speaking about it beforehand?
Reply: “After we had been wanting on the market and the completely different genres or use circumstances for tilt/shift lenses, we discovered that there was a necessity for an extended focal size as effectively. The 30mm was good for issues like architectural pictures, however the business market wanted one thing longer for issues like trend shoots and different studio work.” (30mm on the GFX platform has the angle of view of a 23.7mm lens on a 35mm physique, whereas a 110mm is equal to an 86.9mm lens on a 35mm physique.)
Apparently they had been contemplating each lenses from the beginning and started designing them each at across the identical time, however for no matter purpose they weren’t certain whether or not they’d really be making the 110mm or not till someday after their plans for the 30mm had solidified. Ultimately, after some preliminary delay, they arrived on the ultimate design of the 110mm pretty shortly, such that they had been capable of announce it on the Fall X-Summit.
(In the meantime, the discharge date for the 30mm nonetheless hasn’t been set – or they’re merely not disclosing it but.)
Abstract
As all the time, it was an fascinating dialog, and lots of because of Igarashi-san and Watanabe-san for taking the time to speak to me. The X-H2 appears to be like like an amazing digicam that’ll match the wants of many shooters. In my very own pictures, I’m way more serious about decision than excessive taking pictures speeds, though the X-H2’s 20fps/24 megapixel mode remains to be loads of decision and would have been thought-about cutting-edge for a “sports activities” digicam just a few years in the past. Whereas the X-H2S might be a no brainer for sports activities and maybe wildlife photographers, I believe that the X-H2 will finally outsell it. No matter which mannequin is extra widespread, the pair leaves Fujifilm with a superb place within the high-end APS-C market. On the identical time, their ever-growing medium-format GFX line contains 5 our bodies, 15 lenses (together with the 110mm tilt/shift, which doesn’t seem on their web site but) and a teleconverter. Mix all that with Fujifilm’s strong AF efficiency, their excellent optics and Fuji-only options like movie simulations, and so they appear poised for achievement for years to come back.
