“A drive to embrace new technology and change,” answers data scientist and CEO of Alice Camera, Vishal Kumar, when queried about what sets Alice Camera apart from other camera companies. He found that existing mirrorless cameras were too cumbersome and decided that a shift was needed in the camera industry. This stirred up multiple ideas in his head, which eventually led to the creation of the Alice Camera. We caught up with Vishal and his team for an exclusive interview about the upcoming camera and an insight into the AI algorithm it uses to produce pictures.
If you’re a millennial like me or older, chances are that your family had a compact camera even if there was an SLR in the household. It was the go-to camera for parties and trips around the country. You didn’t have to think much while using it; most of the time, the inbuilt AF and metering system got a decent picture. Somehow, when the digital camera revolution came about, digital compact cameras didn’t last very long. I wouldn’t say that’s because DSLRs quickly became more affordable; it’s just that the smartphone camera became more convenient. Here was a device that did more than just calls and messaging – it was the perfect pocket-sized camera to carry with you everywhere. And with the average person taking photos primarily for social media purposes without any real intent to print them, the quality they provided proved to be more than enough.
Camera specs and features drive smartphone sales more than anything else these days. It’s what phone manufacturers use first these days when promoting the features of their smartphones. These days, some smartphone cameras have a higher resolution sensor than the average DSLR or mirrorless camera. A handful of them even enlisted the service of leading camera brands like Leica and Hasselblad to develop technology for their smartphone cameras. And the most prominent casualty as a result of the rise of the smartphone was the compact camera. When your smartphone camera can do everything it does, and sometimes even better, what was the need to carry around a dedicated camera in your pocket or bag?
Maybe three years ago, I was at a photo walk when someone introduced me to a Huawei smartphone. It wasn’t the specs of the camera’s sensors and lenses that amazed me; it was the inbuilt scene recognition AI it had. Depending on where you pointed the lens, the phone’s AI could recognize the subject and enhance the color and details accordingly. Skies turned bluer and sand warmer if you aimed it at the beach. Food on your plate popped in color and looked more appetizing. Even when you took a selfie, you had the option of smoothening your skin. I can’t conclusively say if Huawei was the first to introduce AI into their smartphone photography, but they certainly made me realize one thing – camera manufacturers needed to innovate faster.
The compact camera is almost a thing of the past now because of the convenience of the smartphone. Yes, digital compact cameras still have larger sensors, but the benefits of these sensors are being matched with AI. Phones can now selectively and accurately blur the background behind a subject using AI, and they are getting better and better at this with each new release. I realized soon after seeing that Huawei demo that if camera brands needed to compete with this level of innovation, the only way to do it was to introduce AI features into their cameras. The way I imagined it could be done was by connecting the camera to a data network (possibly via a smartphone) to be connected to an AI neural network. Otherwise, to the average consumer, a DSLR or mirrorless camera requires more effort to get a similar result to what AI-enabled smartphone cameras can already do today.
Gadgets like the Arsenal 1 and 2 have attempted to make some headway in this regard, but Vishal and his team hope to shake things around with their Alice Camera. They believe that the way forward to getting better images for the average consumer is to harness the powers of computational photography. They acknowledge that professional cameras and lenses will always provide superior images to those coming from smartphones. But by adding AI techniques to these systems, they are looking to add ways to increase their image quality by leaps and bounds. They’re not the first to try and do something like this. Cameras like the Zeiss ZX1 and open platform Olympus AIR A01 had close smartphone integration but failed in the seamless execution of this idea. They don’t see camera brands taking this initiative soon, so the team at Alice Camera decided to drive the change forward by themselves.
The Phoblographer: Please tell us about yourself and how you got into photography.
Vishal Kumar – CEO, Alice Camera: I’m a cultural data scientist and content creator with over 30,000 followers on social media. My speciality is to apply data science to arts and culture and I’m a public speaker in creative AI. I was producing videos as a content creator and educator and found the mirrorless cameras on the market extremely cumbersome for producing videos. I’m mainly interested in cameras for producing videos.
Liam Donovan – CTO, Alice Camera: I got into black and white film photography at university, and learned how to develop film and wet print there. I found I enjoyed the darkroom side of things more than actually taking photos. I loved the creativity and physicality of the process, experimenting with techniques like dodging & burning, solarisation, and so on. I studied electronic engineering and did a Ph.D. in technology for media and the arts, and it’s a real pleasure to be able to combine my old hobby with my work on the Alice Camera.
Vik Kumar – COO, Alice Camera: I got deeper into photography as I started to travel more over the last 5 years, walking and exploring different cultures and nature provide eye-gazing opportunities to remember. On the go, I carry around a compact 14-140mm MFT lens and it’s great to stop, focus, and frame a shot whilst out. Filmmaking and directing are certainly avenues I’d like to explore mastering further over the years. I studied a degree in Physics, so the capture of light, the capture of a specific moment in time and space via photography is fascinating to me. As a keen science and tech enthusiast, having grown up opening and pulling products apart, studying electronics, natural sciences, and mathematics it’s great to be a part of a team developing a new camera system.
The Phoblographer: For the benefit of our readers, how would you describe Alice Camera to someone who’d never heard of it before?
Vishal Kumar: The Alice Camera is a Micro Four Thirds camera that attaches to the back of your smartphone. An app on your phone connects with Alice Camera via WiFi and gives you all the benefits of a modern user experience, user interface, and the connectivity of your smartphone. The Alice Camera’s unique selling point is that it has two dedicated AI processors and a GPU inside the camera which is used to run our novel and proprietary computational photography pipeline.
The Phoblographer: When did you decide that it was finally time that a new camera was needed to shake up the market. What was that turning point which led to the concept of Alice?
Liam Donovan: For me, the turning point was when I first got to use one of the latest few generations of smartphones that demonstrate really impressive computational photography capabilities. I was sitting around a campfire trying to take a photo of some friends of mine using a DSLR when it hit me – I just couldn’t get the DSLR to take a decent picture, I couldn’t get the white balance right, I couldn’t get the exposure right, and neither could the camera’s auto systems. I could’ve gone and got a tripod, and I could have concentrated a bit more on the settings and got a decent exposure if I had really tried, but I was enjoying sitting around the campfire with my friends, so I gave up. Out of curiosity more than expectation I got my shiny new smartphone out and tried to take a shot with it, and the exposure was instantly perfect. It genuinely shocked me because I had not expected smartphones would ever be capable of producing better photos straight-out-of-the-camera than a DSLR, let alone in challenging, dark, high-dynamic-range conditions, and it made me marvel at the thought of combining that computational photography capability with a professional optical system
The Phoblographer: We’ve seen AI used to assist smartphone camera tech for some years now. Existing products like the Arsenal ‘camera assistant’ have been doing similar things as well. How much smarter is Alice’s AI computational technology?
Vishal Kumar: Alice is built more like a smartphone than a traditional camera. It has been designed specifically to enable advanced AI-powered computational photography algorithms to be run on images collected by its micro four-thirds sensor. It contains an advanced Qualcomm Snapdragon processor similar to those in many Android smartphones with enormous image-processing capacity, alongside a Google Edge TPU, a dedicated accelerator for AI algorithms similar to those found in some Google Pixel phones. It is a radically different design to a typical mirrorless or DSLR camera, with a greater focus on flexibility and computational power.
AI is a transformational technology for photography, it enables many things that were previously impossible and automates things that previously required skilled human intervention. It allows professional-quality post-processing to be applied to video in real-time for live streaming, and it allows a camera to capture photos with higher resolution and dynamic range than its optical system is capable of. It is causing fundamental changes in how content is created, shared, and consumed, and Alice is designed to harness it. We have already collected over 5 million images from open datasets to train our AI models. Now we’re building our own proprietary image database to further train and enhance our novel end-to-end AI pipeline for professional imaging.
The Phoblographer: Phone manufacturers are also gradually increasing the sizes of phone camera sensors with Sharp and Sony introducing 1” sensors last year. What was the limiting factor that led to Alice starting out with m43 sensors as opposed to something like APS-C?
Vishal Kumar: The decision to go with the MFT format was an active one, not one we were forced into. As the most compact professional system, it sits at a sweet spot for us, giving us enormously better image quality than any smartphone but allowing us to retain a level of compactness and affordability that you wouldn’t get with larger sensors. The system does have some limitations compared to APS-C or Full-Frame, but we plan to address many of those with computational photography.
The most disruptive thing AI is doing to the camera market is breaking the long-held certainty that a bigger sensor inevitably means a better camera. Physics says that the more light a camera can capture from a scene, the better the image quality it can produce. It is true that a bigger sensor captures more light from a single capture than a smaller sensor, but a smaller sensor that can capture many frames in quick succession and combine them using AI to eliminate artifacts, can capture more light again, and so can produce better image quality. This is the core insight that has allowed smartphone cameras to improve so drastically despite their tiny optics, and we intend to exploit it with Alice.
It’s also worth noting that the 1-inch Sony smartphone is actually only able to use ~60% of its sensor’s area due to limitations in the lenses that can be crammed into the smartphone form factor. It will take a significant advance in sensor or lens tech before smartphones are able to fully utilize 1-inch sensors.
The Phoblographer: Tell us a bit about the team at Alice. Who are the key players and what do they bring to the product?
Vishal Kumar: Dr. Liam Donovan, is our CTO and he has over 10 years of experience as an electrical engineer. He did a Ph.D. in Media and Arts Technology from Queen Mary University London and an MEng from the University of Oxford, as well as an Enterprise Fellowship at the Royal Academy of Engineering specifically around commercializing AI embedded hardware systems for creatives. He’s responsible for overall product definition, development, and integration of core components and image signal processing software.
I, Vishal Kumar am our CEO and lead the brand, marketing, and strategy. I is recognized as a cultural data scientist and public speaker in creative AI and content creator with 30,000 followers on social media. He worked as one of the first data scientists at Sotheby’s and advises the European Commission, the UK Government, and the Mayor of London on data and AI policy for the cultural and creative industries. He has two masters – an MRes Data Science and Visualisation from UCL and a MA in Art Business from Sotheby’s Institute of Art – and a BSc in Applied Economics from the London School of Economics.
Vikas Kumar is our COO, and he leads on project management, operations & commercialisation. He is responsible for sourcing components and maintaining partner relationships, as well as for intellectual property management and financial operations. He previously worked at JP Morgan in the Sales & Marketing team for over five years and is experienced in finance, customer acquisition, and relationship management. He holds a BSc in Physics from UCL and is a practicing photographer.
We now have a team of seven people. Two talented Computational Photography Engineers, Shelly, and Ollie, are mainly responsible for our AI and computational photography algorithm development and deployment. You can see our Conversations in Computational Photography series featuring Shelly’s and Ollie’s work or our Meet the Team series (Shelly and Ollie) to learn more. Lavinia is our Innovation and Community Manager and liaises with customers, fielding feature requests and undertaking research, and Laura recently joined to manage our blog and newsletter and to coordinate on our internal team management. We’ve also worked with Aetha Design on the core product design of Alice. Aetha was founded by two senior ex-Dyson engineers and it’s been a real pleasure to work with Tom and the team.
The Phoblographer: Alice requires smartphone integration. Does this potentially limit its functionality based on the phone used with it? Why or why not?
Vishal Kumar: We only use the smartphone as a display and control interface, the image processing pipeline and computational photography algorithms run on the processor inside the Alice Camera. There is essentially no difference in functionality when using Alice with different phones, even relatively old ones.
The Phoblographer: If not, what were the limitations towards housing the necessary technology inside the Alice Camera itself, and avoiding the need for a smartphone?
Vishal Kumar: The fact is that smartphone makers build better screens, touch interfaces, and UI software stacks than we can – they are multi-billion-dollar companies who have been competing on precisely those features for many years. They’re better at it than the big camera companies too. But the experience of using a latest-gen smartphone screen as the viewfinder for a professional interchangeable-lens optical system is quite striking, and being able to upload instantly to the cloud or social media is pretty handy. Pairing the camera so closely with the smartphone unlocks a bunch of user experience improvements and allows it to fit in more seamlessly with the modern content creator’s workflow.
The Phoblographer: What’s the one feature in Alice that will set it apart from existing cameras and make it a must-have for photographers and videographers?
Vishal Kumar: The obvious answer is our AI-powered computational photography, which for me at least is the most exciting aspect. But then there’s also the connectivity and user experience improvements brought by the smartphone integration, the accessibility of the software running on the camera, and the open-access developer community we plan to build around it, the potential for the camera to be used as a platform for third-party applications. If I had to boil it down to one thing, I think it would be a drive to embrace new technology and change.
The Phoblographer: You’ve opted to go open source with this. Was this to allow buyers the chance to customise better, or to invite more collaborators for the betterment of the system?
Vishal Kumar: Our plan is to be “open-access” and release substantial elements of the source code and image and video processing pipelines running on the camera, with a few necessary exceptions. By doing this we hope to expose the internal workings of the camera and offer our users unprecedented scope for customization, and as you say, allow us to make the camera better for everyone by inviting people to collaborate and contribute. We surveyed over 1,000 people from our community and asked them whether this is a feature that they actually want from their camera and 92% of people overwhelmingly said they want more control over their creative process. As a team, we’re interested in the intersection between open source software and people’s creative pursuits. When you give people access to the software running on the devices they are using for their creative practice, exciting things can happen. In the future, we’re looking to develop infrastructure to make Alice’s software more accessible to non-expert programmers, and build a real open source community around the camera.
The Phoblographer: The Alice AI colour enhancement is able to apply LUTs in real time, at variable levels of intensity, which could prove to be highly useful to content creators. Please give us an insight into this algorithm.
Vishal Kumar: The idea behind our colour enhancement algorithm is to bring the skills of a professional photo retoucher onto the camera, and make it so fast that it can run in real-time on video, even if you are live streaming. Instead of doing colour editing afterwards, creators will be able to do it on Alice in real-time using AI, speeding up their workflow so that they can focus on other things. To do this we have built a set of tools similar in principle to those you would find in typical photo editing software, but specifically designed to be controlled by an AI. We have then taught the AI how to use those tools by showing it a large number of before/after pairs of images processed by a skilled human, and the AI is able to learn how to replicate the changes that a human would make, even on new images it hasn’t seen before. We have designed the tools very carefully to maximise the expressive power of the algorithm, allowing it to make specific local modifications to certain areas depending on the context along with more general global changes, but we have also carefully restricted the algorithm from being able to do anything that could be considered untruthful or unethical, like hallucinating details, warping reality or changing a subject’s body shape (see our article explaining our Four Rules of Computational Photography). We will be providing our own stock version of the algorithm with Alice which will have been trained to a particular style that we have chosen, but we also want to allow people to re-train the algorithm with their own photos in their own style.
The Phoblographer: With such processing happening in real time, on a frame to frame level, what sort of battery life are we looking at?
Vishal Kumar: One of the trade-offs of computational photography is that it does require more power, but we are using a very efficient and modern Qualcomm Snapdragon processor and we have a large 5000mAh capacity battery comparable to what you would find in a flagship smartphone. The camera can also be charged in the field using the same portable power bricks you can use to charge your phone.
The Phoblographer: Is the ongoing global chip shortage proving to be a hurdle for manufacturing? How many units are you expecting to sell by the end of this year? What’s the forecast like for the next handful of years?
Vishal Kumar: The design of Alice is extremely minimal when compared to a traditional mirrorless DSLR camera. We’ve stripped out a lot of complexity; there aren’t too many physical buttons and there are only a handful of core electronic components. The global chip shortage has had a wide impact on the whole market and we did witness delays of six months. However, we have been in continuous contact with our suppliers and have managed our supply chain for 2022. We’ve already pre-sold cameras to customers in 2021 but the specific forecast numbers are not yet public information. What we can say is that our community is growing rapidly and reached over 25,000 followers in 2021. This goes to show that there are lots of people out there interested in applying computational photography to professional optics and we want to provide them as much value as we can. The initial interest and reception of our technology has been overwhelmingly positive, and we are developing with long-term scale in mind.
The Phoblographer: You’re up against some established brands with decades of expertise in imaging sciences. How confident are you that Alice won’t meet the fate Lytro or the L16 cameras met?
Vishal Kumar: Both Lytro and the L16 relied on some pretty advanced experimental technology to fulfil the basic function of capturing images, light-field technology for Lytro, and multi-camera fusion for the L16. By contrast Alice uses the same basic image capturing process as any smartphone or mirrorless camera, the magic comes from the combination of a professional optical system, cutting-edge computational photography and mobile app software engineering, all of which are individually well-proven technologies. It’s not quite a fundamental leap in physics, but we believe our approach is an improvement to the status quo, especially for individuals who create content regularly and make a living from it. The fundamental supply and demand dynamics in today’s market are very different from a decade ago… even five years ago. In 2022, there are now at least 50 million individuals worldwide who make a living from being a creator, according to data from Signal Fire, and this group is growing at 15% annually, according to data from Re:Create.
There are also hundreds of millions of hobbyists and aspiring content creators, and the pandemic has accelerated this trend. Many creators who produce content for YouTube, Instagram, TikTok and Twitch and who get brand sponsorship for their content tend to use interchangeable lens cameras because of their superior quality compared to a smartphone. Creators are demanding more from their electronic devices, but the current mirrorless camera solutions on the market today are extremely restrictive and are miles behind in terms of modern UX/UI, internet connectivity, app functionality, and computational processing. It’s usually the companies who have been around for decades that are the ones who are most reluctant to overhaul their core product lines and undergo radical innovation. That’s because they are tied into supply chain commitments, have an established customer base, and also because of their company culture. We sense an opportunity to offer something new and fresh grounded in a lot of customer research and we have the necessary skills and resources to build this product for creators. We’ve just raised a combined $500,000 pre-seed fundraise to turbocharge our go-to-market and accelerate the development of our technology.
All images (unless specified otherwise) supplied by Vishal Kumar. Used with permission. Visit Alice Camera’s website and their Twitter, Instagram and Youtube pages for more information about their technology.