What Will the Next Fujifilm Sensor Be?

According to Fujirumors, the next Fujifilm APS-C camera will be the X-H2, which won’t be released until early 2022, and it will have a new sensor that’s capable of 8K video. Not a whole lot else is known about it at this point. What will the new sensor be? What specs will it have? Absolutely nobody outside of Fujifilm has any idea, so it’s a fun opportunity to wildly speculate. To be clear, I have no inside information. This isn’t a rumor. What I’ll discuss below is a bad guess at best. I just thought it would be fun to talk about the possibilities.

The assumption is that the next sensor will be X-Trans (X-Trans V), which is logical—most likely it will be. I don’t know what would differentiate X-Trans V from X-Trans IV. The theory is that because Fujifilm has been developing sharper lenses with more resolving power, they’re preparing for a higher-resolution sensor (in fact, they’ve said as much). But how much more? 28-megapixel? 30? 32? 36? 50? Nobody knows, but don’t be surprised if it’s 36-megapixels. Unless you crop steeply or print largely, that extra resolution won’t do much for you. I personally wish that Fujifilm would focus less on megapixels and focus more on other advancements, but that’s just my opinion.

It’s possible that the new sensor inside the X-H2 won’t be X-Trans, or at least not a Sony X-Trans. Fujifilm has partnered with Samsung to create the ISOCELL technology that Samsung uses in their cellphone cameras. In an oversimplified explanation, ISOCELL allows pixels to more accurately capture light, which means that smaller pixels act more like larger pixels. Samsung uses ISOCELL in conjunction with Pixel Binning (“Tetracell”), a technology that uses a group of pixels to act as a singular larger pixel for improved dynamic range and high-ISO performance. This technology allows tiny cellphone sensors to perform better than they should. Why can’t this be applied to larger sensors? Remember when Samsung used to have a highly-acclaimed 28-megapixel APS-C sensor before their NX camera line went suddenly defunct? Maybe Fujifilm and Samsung will partner to bring some of Samsung’s innovative sensor technology to Fujifilm cameras.

I’d be surprised if Fujifilm included a Sony Bayer sensor in the X-H2, but it’s possible. Anything is possible. More likely, if Fujifilm were to move on from X-Trans, the sensor would have to have some unique marketing aspect to it. Fujifilm X cameras are the only cameras with X-Trans sensors, and all other current cameras use Bayer (except for some Sigma models). X-Trans has some advantages and disadvantages, but more importantly it’s unique, which Fujifilm takes advantage of, both in terms of technology and marketing. There’d have to be something especially special about a non-X-Trans sensor for Fujifilm to suddenly abandon what has brought them this far.

Now imagine this: a Fujifilm X-H2 with a 144-megapixel ISOCELL and Pixel Binning sensor, that “normally” captures 36-megapixel images, with the option to capture 144-megapixel images in good light and 9-megapixel images in very low light. That would stir a lot more attention than an ordinary 36-megapixel Bayer sensor, and would also have some advantages over it. It would certainly make headlines!

The way it would work is that under most conditions the camera would capture a 36-megapixel image that would perform, in dynamic range and high-ISO, similar to the 26-megapixel X-Trans IV sensor. When the ISO is set to 320 or lower, the camera would have the option to capture a full 144-megapixel image (with the limitation of DR400 not available). Of course, Fujifilm lenses, while exceptionally sharp, cannot resolve that much detail, so you’d likely get details more in line with 50-megapixel cameras (maybe more, maybe less, depending on the lens). The camera would also have the option at higher ISOs—perhaps ISO 3200 and above—to capture extraordinarily clean 9-megapixel images (and perhaps 1080p video). I know that 9-megapixels are hardly anything to get excited over, but think of this as being sort of like the Sony A7S, which has only 12-megapixels, but is highly regarded for its low-light capabilities. So, yeah, the picture might only have 9-megapixels of resolution, but it was captured at ISO 25,600 and looks as clean as ISO 800. Maybe pixel-shift could even be incorporated into this somehow.

There would be a whole host of issues if Fujifilm incorporated Samsung’s technology into the X-H2, most notably the RAW files. I don’t think my suggestion is likely, but since anything is possible, I thought that I’d wildly speculate, and this is as wild of a speculation as you’ll likely find on this topic. It will definitely be interesting to see what Fujifilm comes up with, and as soon as I know something, I’ll be sure to share it and my ideas about it with you.

Comparing X-Trans Sensors

Omar Gonzalez posted an interesting video yesterday that I want to share with you. It poses the question: which Fujifilm X-Trans sensor is the best? I don’t want to get too deeply into what’s “best” because that’s very subjective. Omar directly compares images from X-Trans II, III & IV sensor cameras to see what the differences are. There can be, in fact, some pretty significant differences between sensor generations! If you have a few free moments and haven’t already watched it, push play on Omar’s video above.

I’ve done some pretty similar experiments. I’ve done my own side-by-side comparisons in the past. I know the differences between the sensors, particularly regarding JPEG output, and I agree with most of what Omar says in his video. Each sensor generation produces slightly different results, and that’s largely due to Fujifilm’s programming.

X-Trans II is programmed warmer than the others, and fairly significantly so. For example, my X-T1 Kodachrome 64 recipe requires a White Balance Shift of 0 Red & -3 Blue while my X-T30 Kodachrome 64 recipe requires a WB shift of +2 Red and -5 Blue, so there’s definitely a difference. X-Trans III is slightly warmer than X-Trans IV, but not by much—it would require a decimal in the shift, such as around a 0.3 adjustment, to make them match, which unfortunately isn’t possible. Omar doesn’t discuss X-Trans I, but it’s more similar to X-Trans III and IV in regards to warmth.

Man in Red – Farmington, UT – Fujifilm X-T1

I don’t find X-Trans II to be punchier than X-Trans III or IV, but they are certainly not programmed the same. Omar detected more contrast and more vibrant colors from his X-Trans II camera, but I think I can explain what is happening. First, the luminosity curve isn’t identical, so highlights and shadows are rendered slightly different on X-Trans II. Second, adjustments in X-Trans II cameras max out at +2 and -2, while adjustments in X-Trans III & IV cameras max out at +4 and sometimes -4; however, +2 on X-Trans II isn’t the same as +2 on X-Trans III & IV. +1 on X-Trans II is roughly equivalent to +0.8 on X-Trans III & IV, so it might seem to produce punchier results, but +3 on the newer sensors goes beyond +2 on X-Trans II, allowing you to get more contrast and color vibrancy from the newer cameras. X-Trans I is more similar to X-Trans II in how it renders shadows, highlights and saturation, but it’s not identical.

High ISO is something else Omar looked at, which is definitely subjective—what one person finds acceptable another might find detestable. On X-Trans I & II cameras, I don’t like going above ISO 3200 for color photographs. On X-Trans III cameras, I sometimes find ISO 12800 to be acceptable for color photography, depending on the subject and settings. On X-Trans IV cameras, ISO 6400 is my upper limit for color pictures. This isn’t too dissimilar to what Omar found, although I believe that ISO 3200 is his preference for the upper ISO limit no matter the camera. There’s no right or wrong acceptable threshold, just what works for you. For B&W photography, I don’t mind using even higher ISOs—in fact, it might be preferable to do so.

100% – Salt Lake City, UT – Fujifilm X100V

In my opinion, of the different X-Trans sensors, X-Trans II is the most “different” looking, although mostly because it’s programmed to produce warmer pictures. In some ways X-Trans I is more similar to X-Trans II and in other ways it is more similar to X-Trans III & IV. There are some differences between X-Trans III and X-Trans IV, but to my eyes they’re the most similar. I don’t personally believe that any one sensor generation is inherently better than another, but it’s clear that they’re not completely identical.

Fujifilm continues to add new JPEG options to newer cameras, which allows you to further customize your straight-out-of-camera look. X-Trans I doesn’t have Classic Chrome. X-Trans II doesn’t have Acros. X-Trans III doesn’t have Color Chrome Effect. This is just scratching the surface! There are just so many more picture aesthetics that one can get straight-out-of-camera on the X-E4 than the X-E3, and the X-E3 can get more than the X-E2, and the X-E2 can get more than the X-E1. For many people, that makes the newer sensors “better” than the older ones, but if you prefer how an older sensor renders pictures, then that sensor is likely “better” for you. It just depends on your preferences—whatever works best for you and your photography. While one camera will render pictures a little different than another, and one might have more features than another, the most important thing is what you do with it. Using your gear to the best of your ability is much more important than the gear itself.

Thoughts On Samsung’s 108 Megapixel Sensor + How It Relates To Fujifilm

16770808038_edc1774179_c

We Will Deliver – Rosamond, CA – Nokia Lumia 1020

Samsung announced (in conjunction with Xiaomi) that they have made an 108-megapixel 1/1.33-inch camera sensor that will soon be found inside of cellphones. At first glance it sounds absurd. What kind of image quality could it possibly have? How ugly will it be above base ISO? How much resolution do you really need for social media posts? But there are some interesting innovations that might someday be applied to Fujifilm cameras, so let’s take a closer look.

How this new sensor directly relates to Fujifilm is that it’s an ISOCELL Plus sensor, which requires a materiel developed by Fujifilm, and only Fujifilm has this material. What Samsung did with it is develop a sensor that has less “cross talk” between pixels, which improves color accuracy, dynamic range, high-ISO capabilities and fine-detail rendering. Essentially, it allows smaller pixels to perform similar to larger pixels. You can put 108 million teeny-tiny light sensitive sensor elements on a small sensor with ISOCELL Plus, and it will perform similar to 108 million larger-but-still-quite-small light sensitive sensor elements on a little bit larger sensor without this technology. Whether the lens will be able to resolve that much detail, as it will need to be a heck-of-a-sharp lens, remains to be seen, but if it can, that would be quite the leap in cellphone camera technology.

I used to have a Nokia Lumia 1020 cellphone, and the phone itself wasn’t especially great, but the camera, with a 41-megapixel 1/1.5-inch sensor and Zeiss lens, was surprisingly good. Well, sort of. It had a very narrow margin, as you needed to stay close to base ISO, and the dynamic range was small, but in the right situations it delivered stunning pictures that you’d never guess came from a cellphone. I have no idea if Xiaomi’s phone with the new 108-megapixel sensor will be similar or not, but it might be, and it might even be better.

14357431707_458e7c4c5d_c

Energy – Tehachapi, CA – Nokia Lumia 1020

Aside from ISOCELL Plus, the other interesting innovation from Samsung with this sensor is quad-Bayer array. Instead of the typical two green, one red and one blue Bayer square arrangement, this has a four green times two, four red and four blue square arrangement, with the four pixels of the same color next to each other in a square. The idea is that the four same-color pixels can be merged through software into one pixel, turning the camera into a 27-megapixel traditional Bayer array. Why wouldn’t Samsung use larger light sensitive sensor elements and set the megapixel count at 27? Why do this weird tiny-pixel quad-Bayer pixel-merge thing? Well, it allows software to do some interesting tricks. For example, it can capture up to four independent 27-megapixel exposures simultaneously and blend them together, extending dynamic range, reducing noise, and/or increasing high-ISO capabilities. Or, if the dynamic range doesn’t need extended, and the noise doesn’t need to be reduced, and the ISO doesn’t need to be increased, it can produce a very large fine-detailed full-resolution picture.

Slowly the technological advancements of the small sensor world trickle up to larger sensors, and someday a version of ISOCELL Plus and pixel-merge could very well be found in Fujifilm cameras. What might this look like? If you were to take this same Samsung chip and increase it to APS-C size, it would have roughly 216-megapixels, and would deliver a pixel-merged 54-megapixel image. I’m sure, however, that there would be a reduction in noise performance, dynamic range and high-ISO over current X-Trans sensors, and, even with the excellent Fujinon lenses available, the question of whether that much detail can be resolved would still need to be answered. What I see more likely to happen is sensor elements being used that are twice as large as those on the tiny Samsung chip, and an APS-C sensor with 108-megapixels produced, which could be pixel-merged to 27-megapixels. I’m not 100% sure, but I believe a quad-X-Trans array is possible. Essentially, it might be feasible to have nearly identical resolution as X-Trans IV, but with improved dynamic range and high-ISO capabilities, and the option for full-resolution 108-megapixel pictures when the ISO is under a certain amount (say, ISO 640). It’s still questionable whether or not Fujinon lenses can take advantage of that much resolution, but even if it is “only” able to produce resolution equivalent to 50-megapixels, that’s still double what it is now. If ISOCELL Plus and pixel-merge ever do come to Fujifilm X, it could very well be a game-changer type of thing. Or perhaps the required processing power and heat dispersion are too difficult to overcome, and it never makes its way to larger sensor cameras. Time will tell.