Connexion
"I think a lot of people mistake our product as a gadget": Camera Intelligence at CP+
 |
| Some of the Camera Intelligence team, all holding Cairas. Left: Vik Kumar, COO. Middle: Vishal Kumar, CEO. Right: Liam Donovan, CTO. Photo: Mitchell Clark |
"I think a lot of people mistake our product as a gadget," says Vishal Kumar, CEO of Camera Intelligence, the company behind the Caira, a Micro Four Thirds camera that attaches to the back of an iPhone. "That's the wrong way to think about it," he says. "We're building the full technology stack, we believe, for the future of cameras."
While the company is currently wrapping up its Kickstarter campaign for the Caira, it has much bigger ambitions than selling an accessory that makes your phone a better camera, or even than selling dedicated cameras. Ultimately, it says, it's trying to change how cameras are made, and trying to convince the established camera companies that its tech is the way to photography's next step forward.
What is Caira?
 |
| The Caira has a Micro Four Thirds mount and an 11MP sensor. It doesn't work standalone – it requires an iPhone and has a corresponding app. Product photos: Mitchell Clark |
While Camera Intelligence's ambitions clearly go beyond its first product, it's worth looking at where they're taking root. If you didn't catch the announcement in October, the Caira mounts to the back of an iPhone using MagSafe and connects to it via Wi-Fi. From there, you get your image preview and can control it using an on-screen interface or via voice.
Its big trick, though, is that it's not just a lens mount and an 11MP Four Thirds quad-Bayer sensor from Sony Semiconductor (think Panasonic GH5S). It also has a Snapdragon processor, which it uses to power its AI autofocus and to make decisions about white balance and exposure. It also handles the processing for the camera's computational multi-shot modes.
"At the moment, it's mainly designed for denoising," says Liam Donovan, Camera Intelligence's CTO. "It's optimized for reducing noise in low light and giving you more dynamic range." He explains that the algorithm, which the company is still working on but is shipping on production cameras, takes up to 17 shots: eight before you press the shutter, one when you press the shutter, and eight after. "There's an algorithm that runs on all of those shots to align them all. It moves the frames about so that they match the central frame, and then blends them together. And one of the hardest bits of the algorithm is making sure that there are no artifacts."
 |
|
This Photo from the Caira was created by stacking eight frames together. Click here to see the original in HDR. Caira | OM System 20mm F1.4 | F2.8 | 1/710s | ISO 951 |
While the multi-shot capabilities are currently implemented as a separate low-light mode, the company is working on expanding what it can do with them. "We're working very hard on making it do super resolution, where you can actually increase the resolution of the images that it produces quite significantly," Donovan says. "Low-light mode is kind of a stepping stone on the way to that, because super resolution is much more difficult than denoising on its own."
The intensity of those algorithms is part of why the camera needs its own processor, despite being attached to a very powerful phone. "It takes quite a lot of time to transfer Raw photos off the camera, and it has to be the full Raws that go into the stacking algorithm," says Donovan. "So that all happens on the camera, and it's enabled by the fact that we have a very powerful smartphone-style processor from Qualcomm in the camera itself, which is pretty novel." He says it's much more powerful than the processors you'd typically find in more traditional cameras.
Influencing
While Camera Intelligence has obviously put a lot of work into the Caira and its predecessor, the Alice Camera, it has its eyes on a bigger prize: convincing other camera companies that this architecture is the way forward. "We're not just in Yokohama for CP+," says Kumar, who spoke to us at the Camera Intelligence booth, where the company is showing off Caira to attendees at the show.
"We're having deeper conversations with some of the camera brands to see if that resonates"
"We have more than a couple of meetings. We believe that the way in which we have gone about building our camera is more akin to how cameras in the future should be built. Because we use the image sensors from traditional mirrorless cameras, but we're using the processor from a smartphone. And that smartphone processor unlocks a whole bunch of things," he says. "So we're having deeper conversations with some of the camera brands to see if that resonates."
It's not just the features like AI-powered voice control, focusing, or exposure, he suggests. The company also wants to be able to implement other AI features, such as large language or image generation models; the current Caira app lets you edit images using Google's Nano Banana model, but it currently has to run in the cloud, rather than locally, and requires a subscription.
It's not all AI, though. "I think it's the full stack," Kumar says. "It's everything from the decisions we made on the electronics, the operating system layer that we've built, which is called Camera Intelligence OS, the image signal processing pipeline that we've built, which is entirely software-based, and can be updated over the air. And then it's also the user interface with the app and the large language model. It's that entire stack that we think is interesting."
 |
In that respect, the Caira is almost a means to an end. "That's the product that we sell at the moment to prove that architecture and infrastructure can actually be robust. And we need to ship these units because it allows us to stress-test the architecture and the units. And it allows us to do those software and hardware improvements. But it's more to it than just a camera that attaches to your phone."
In that vein, Kumar tells us that the Caira is on schedule to start shipping to the 611 people who backed it on Kickstarter last year by the end of March. He also says they triple-tested a batch of 25 from the production line in their office in London to make sure they were ready to go. We have our hands on a pre-production model and have been doing some testing with it, so keep your eyes peeled for further coverage.
Tied to Apple
To start, Kumar says the Caira's target audience is smartphone shooters who want better image quality. "Smartphones, for better or for worse, have automated large parts of the three A's and everything else for almost a decade. And we always tell our users, 'You need to learn basic photographic theory because it will help you take better photos, take better videos, et cetera.' But we're approaching it from the point of view of: 'OK, how can we automate a lot of this complexity so that the UX matches up with the UX that our target customer base is familiar with?' Not something that's like a minefield for them to navigate."
 |
The app lets you get at a level of manual control, letting you manually set aperture, shutter speed, and ISO (though there's currently no affordance for exposure compensation). "We're starting from a point of view of: 'let's automate as much as possible and then give you control,' rather than 'let's give you full control and then automate after.' Because our customers don't really want that."
There's also a hardware benefit; instead of having to build everything themselves at great expense, Camera Intelligence gets to benefit from the phone's display and connectivity. "I think the main reason was just a lack of resources," says Kumar. "So we were building the electronics. We're building the OS layer. We're building the Image Signal Processor. We're also building the actual physical body. So we thought, okay, why don't we just use the iPhone as the interface, as the screen?"
 |
Donovon also points out that the displays on iPhones, which are up to 6.9" and can reach up to 3000 nits of brightness, far outclass those found on traditional camera models. That brightness, along with readily available software libraries, allowed them to easily add gain-map HDR to the camera, with results you can see as soon as you've taken the picture. The feature is similar to what we've seen with the Hasselblad X2D II; while the Sigma BF also pulls a similar trick with its files, you can't actually appreciate the effect with the screen built into the camera.
Kumar notes that Apple has put a lot of resources into ecosystem features like the App Store, APIs and AirDrop. "So it made sense for us to attach to the phone for our V1 and V2."
For us, our entire product relies on the UX being good.
Of course, that implies that there's a chance the company will try to produce a fully-featured standalone camera in the future, but Kumar says it'd take more than just money to make it happen. "I think we can only promise bringing a product like that to fruition when we have more resources. Not just in terms of money in the bank, but team and the right skill sets. Because if you get that wrong... We've seen a few camera brands release Android-based cameras with standalone screens in the last five years. And they didn't quite execute the concept as well as it could have been done, but those camera companies can absorb that because for them it's just experimenting. For us, our entire product relies on the UX being good."
