Why the speed of Stacked CMOS is key to Nikon's pro mirrorless camera: Digital Photography Review

Why the speed of Stacked CMOS is key to Nikon’s pro mirrorless camera: Digital Photography Review


In an interview with DPReview, Nikon announced its next pro sports camera will be mirrorless. That in itself isn’t much of a surprise, perhaps, given how much investment has clearly been made in the Z-mount system. But the detail we didn’t expect to be revealed was that the camera will be based around a Stacked CMOS sensor.

What is Stacked CMOS?

Stacked CMOS chips are the next generation of chip designs after BSI (backside-illuminated) and provide the designers with greater flexibility, which in turn allows greater capability. BSI sensors are made by fabricating the sensor, then shaving off the silicon substrate upon which it was built. In effect, this allows the sensor to be turned round, and the ‘back side’ to be used for light collection. This means all the connection wiring ends up behind the light-sensitive part of the pixel, rather than getting in the way.

Stacked CMOS takes this process a step further: layers of sensor and circuitry are fabricated, shaved off the substrate and then carefully aligned and joined together. This means the chip designers can build even more complex structures behind the pixels. In the case of the stacked CMOS chips we’ve seen so far, this includes building RAM directly into the sensor: supporting super-fast readout by providing nearby storage for large amounts of data.

It is this fast data readout and handling that we believe makes Stacked CMOS the enabling technology for post-DSLR sports cameras. The news that Sony’s camera division won’t be the only one with access to this technology in future is excellent news for competition within the industry.

What does Stacked CMOS offer?

The only large Stacked CMOS sensors we’ve seen so far come from Sony Semiconductor. So the question is, how similar will the chip in Nikon’s camera be to the one used in the Sony a1?

In the examples we’ve seen so far, the benefits of stacked sensors haven’t come in the terms of image quality (where, at best, they match the already very high standard of existing BSI chips), but in terms of speed.

Sensor readout speed helps define mirrorless camera performance more than it did in DSLRs. Because the imaging sensor is totally central to the operation of mirrorless cameras (acting as both autofocus sensor and also the means of live view image composition), fast readout boosts every aspect of camera performance, and particularly those areas in which sports cameras need to excel.

Sensor readout speed helps define mirrorless camera performance more than it did in DSLRs

There are various tricks for squeezing the most performance out of existing sensors: using a lower resolution feed to provide faster AF updates or taking a dynamic range hit by dropping to lower bit-depth readout. But ultimately, the faster your sensor’s full-resolution, full-precision readout, the faster all these other modes become, too.

How readout speed underpins every aspect of performance

At its most basic, faster sensor readout boosts the camera’s maximum shooting rate. With no mirror to constantly move in and out of the way between exposures, mirrorless cameras have been pushing burst shooting rates upward for years.

Fast readout also means the camera’s autofocus system can be run faster. The more often the camera can take measurements from the scene, the more effectively it can react to changes in the subject it’s trying to focus on. This is a fundamental capability of a pro sports camera.

Interestingly, while the Stacked CMOS sensors from Sony Semiconductor are the ones setting the pace for readout speed, it’s worth noting how much Canon has managed to achieve without adopting such advanced (and expensive) techniques. The EOS R5 manages 20 fps shooting and 8K capture at up to 30p, which is pretty impressive. However, measuring the output of these modes suggests although it can deliver a very respectable readout time of ~16ms, it’s having to drop to 12-bit readout to do so. In other words, it’s enough to deliver a camera with strong specifications, but doesn’t provide as much of a boost to the camera’s capabilities as we’ve seen from the latest stacked sensors.

It also means the viewfinder can be refreshed more often and with less lag between something in the scene moving and that movement being detectable to the photographer with their eye to the finder. For professional photographers who need to anticipate how the action is going to unfold, this is an essential requirement if they’re going to move away from optical viewfinders.

As Sony’s Alpha 1 showed us, fast readout also improves rolling shutter performance, meaning that fully electronic shutters can be used with flash and can be used under a wider range of artificial lighting without having to worry about clashes between the flicker rate of the lights and the speed at which an exposure can be ended.

And, of course, Nikon also said its upcoming professional camera would offer a ‘high resolution’ sensor. Given Nikon’s history of using sensors from Sony Semiconductor Solutions, this raises the possibility that it will use a similar sensor to the a1 and that on top of all these performance improvements, it may be able to match that camera’s trick of providing sports camera speed, and landscape camera detail and image quality.

How fast will the Nikon camera be?

Beyond conventional stills photography, fast readout also underpins the ability to shoot high resolution video. Nikon mentioned 8K during our interview, which would immediately put the upcoming camera in a very select group. We’ll be even more excited if it’s got the processing power to downscale this 8K into super-detailed 4K (just because it keeps file sizes down and is arguably more useful) but for now we’re going to focus on what 8K means for the camera’s sports-shooting capabilities.

We’re going to resist the temptation to start thinking in terms of sideline shooters grabbing their best shots from video clips and argue that delivering 8K doesn’t tell us much about the camera’s prowess as a sports camera.

Canon achieved 8K video capture from the frontside-illuminated sensor in its EOS R5, but it had to drop to 12-bit readout to do so and it can’t match the readout speeds of the latest Stacked CMOS sensors.

The most basic 8K capture demands a sensor that can read-out at least 33 million pixels in 41ms or less and under 33ms to deliver 30p footage. But those numbers only equate to readout rates of 1/24 and 1/30 sec, which is a long time in which a lot can change, if you’re shooting sports. The distortion created by rolling shutter could still be quite significant at those speeds.

The key spec we’ll be looking for will be the e-shutter flash sync speed

What Nikon’s camera is capable of will depend on how much quicker than this its sensor can go. Sony’s a1 uses groups of readout channels to read out its whole sensor in under 5ms (proven by its ability to sync with flashes at up to 1/200 sec). This sets the benchmark for what’s currently possible in terms of readout and all the performance benefits that come from it.

Consequently the key specification we’ll be looking for, once Nikon reveals more information, will be the flash sync speed in e-shutter mode, because that’s primarily limited by how quickly the sensor can be read. This single line in the spec sheet will give the clearest insight into whether the sensor is able to read out fast enough to provide the performance boosts we hope to see across the whole shooting experience.


Related Posts

Leave a Reply

Your email address will not be published. Required fields are marked *