HDR imagery should allow a more lifelike version of images with interesting highlights.
Photo: Richard Butler
We often see concern expressed that there's no real innovation in the industry or that, horrors of horrors, there's more progress being made on the less-developed video side of cameras than on the fairly mature photo side of things.
But, despite these concerns, there have been innovations and the green shoots of progress making themselves visible in this year's releases. We've picked the ones that stood out for us as the ones we think are most likely to have some continued impact.
Combined conversion gain in partially stacked sensors
This one rather snuck under the radar for us: Panasonic had introduced a camera with dual gain output, capturing and combining both a high and low gain signal, simultaneously, as far back as 2022's GH6. But in that camera it was a mode that could only be used in certain circumstances. We saw the improvements in the G9 II and GH7 as addressing that camera's drawbacks, rather than recognizing them as something of a breakthrough.
The Panasonic DC-S1II is the first camera we've encountered to be able to read both the signal from a single exposure via both its high and low gain readout modes and then combine them.
Photo: Richard Butler
Panasonic pulled a similar trick again with this year's S1II, with a different but conceptually similar technology. The 'partially stacked' sensor in the S1II (which is a conventional BSI CMOS sensor with more complex readout circuitry around the edges) showed better dynamic range than the Nikon Z6III had gleaned from the same sensor, but the company was so tight-lipped about precisely what was going on that it took forum regular Adam Horshack to put it all together.
It transpires that the S1II and later Sony a7 V have a new variant of the dual conversion gain sensors that represented the last major step forward in image quality, back in 2014. The 'partially stacked' versions of the existing 24 and 33MP sensors are not only able to readout faster, delivering quicker burst rates and less rolling shutter, but they're also able to operate in their low gain mode and then re-read the same signal in their high gain mode, and combine the results. This takes longer, so isn't used in e-shutter modes, but means these cameras gain a dynamic range boost at their lower ISO settings (where DR is most meaningful).
This won't make much difference to a lot of photographers but the added ability to dig into the shadows will be useful for, say, sunrise and sunset images and will give more freedom for photographers trying to shoot with output on HDR displays in mind. Perhaps the most exciting aspect is that it appears this boost can be applied to existing sensor designs, without incurring the significant costs of truly stacked designs, so we may get to see boosted versions of familiar sensors.
HDR imagery came a bit closer
The Hasselblad X2D 100D II makes HDR photography easier, both through its high-brightness, wide-gamut rear screen and its use of JPEGs with a brightness map embedded in them.
Photo: Mitchell Clark
On the topic of outputting for HDR displays, it felt like we got another step closer to practical HDR workflows this year. Most of the major camera makers have already added true HDR capabilities to their cameras (ie: output for more lifelike playback on HDR displays, not the wide dynamic range capture being awkwardly squeezed into standard DR playback that we'd previously grown to know and dislike). However, the fragmented nature of the Internet means support for showing and sharing the HEIF files they've settled on remains patchy. Similarly, we've had very few instanced of manufacturer trying to tell the press about these capabilities at all.
As things stand, there's a major risk that smartphones, where the screens, underlying software and cameras are all controlled by the manufacturers, will continue to get better at using this approach, raising user's expectations of what photos should look like and leaving dedicated cameras looking dull by comparison.
With this as the background, we were delighted to see Sigma and then Hasselblad adopt HDR output as the default behavior of their most recent cameras. And, crucially, to do so using a filetype with full sharability and backward compatibility guaranteed. JPEGs.
Ultra HDR JPEGs, which are conventional JPEGs with a brightness map that delivers an HDR version on devices that can display it, can be readily shared and shown on the Internet (rather than being limited to specific platforms, such as Instagram), with the knowledge that anyone can open a version of the file.
With Adobe Camera Raw and Google's Pixel phones also supporting these files, it finally looks like there's a way to exploit the wide DR that large-sensor cameras inherently capture. It'll be interesting to see whether any of the big camera makers follow suit or if they're just going to continue to hope that HEIF gains more widespread support.
Content Credentials
Another long-heralded technology that finally started to appear more widely this year was the Content Credentials image authentication system. Originally developed by a vast consortium of stakeholders from media outlets to camera makers, it was originally intended as a cryptographically-backed chain of custody, tracing an image back to a specific camera and keeping track of the edits along the way.
With the increased proliferation of AI-generated nonsense on the Internet, we wonder whether a system for proving authenticity might find wider use. And, to that end, Sony extended its application to video files this year.
Nikon also attempted to add it to the Nikon Z6III, until it became apparent that you could use the camera's multi-exposure mode to get the camera to sign-off on an image that didn't originate with the camera. For now it's primarily Sony and Leica that are providing CC-capable cameras but both Canon and Nikon have been part of the effort, so we'd expect to see its support (and use) continue to spread.
Local AI models
While we look to systems like Content Credentials in the hope that they'll provide some bulwark against AI slop, it's probably worth acknowledging that not everything promoted as AI is a scam, an annoyance or a harbinger of the end of the usable Internet.
"AI" tools can increasingly run locally on your computer and allow you to make your chosen edits quickly and easily. They won't currently stop you creating something ill-advised, though.
Screengrab of Final Cut Pro's Magnetic Mask tool
In line with CIPA's 2024 statement about how AI should be applied to photography, there are some instances where it's being used to do something to support the creative process, rather than trying to supersede it. For instance, Adobe Photoshop now uses AI models running locally on your machine, to make it trivially easy to select and mask different parts of a photo. Nothing is invented or generated, it's not relying on untold additional processing at a server farm somewhere, it's just speeding up the editing process for you.
Similarly, the magnetic masking tool in Apple's Final Cut Pro video editing software is unbelievably quick and effective at selecting and cutting out objects or subjects (ie: particularly people) from their surroundings, even if they move and change shape, frame-to-frame. These are tools that were almost unimaginable a few years ago, that just let you get to the point of making the edits and adjustments you want, so much more quickly. Whether you're an enthusiast amateur or something like a working pro trying to power through a whole wedding's worth of images, these locally-running AI models can be a useful helping hand.
Future essentials or passing fads?
Ultimately, all these things are relatively new arrivals, and your first response my well be: I don't need that. But we've often found that new features and technologies can seem unnecessary at first, but once they start to find their way into your workflow, you one day find it frustrating to go without.
We can't yet know which of these innovations will catch on and bed in and which will seem as misguidedly hyped as 3D TVs and NFTs were. It'll be interesting to see where each of these stand, this time next year, and what other innovations and trends have become apparent in the meantime.
