Listen now on Amazon Polly
The big news in photobooks these days is not printing, binding or substrates, but artificial intelligence (AI). Google has backed into the business through its Google Photos service, using AI to gather pictures and suggest a photobook. The book itself is pretty basic – limited to a single photo per page – but for many consumers, it will suffice.
To compete with Google, photobook brands are going to have to adapt by offering their own AI-powered functionality. That’s no mean feat but according to muvee CEO Terence Swee, his company is up to the task.
We caught up with the flamboyant, globe-trotting CEO to ask him some details about the new SDK offering, now with more information at www.photobook.ai.
Q: muvee pioneered on-device automatic video editing before there was a cloud. Does the new Albumstory AI Automatic Photobook Maker SDK take advantage of this?
Swee: Absolutely! We are huge champions of all the amazing computing power in your pocket. We have today, 100X more computational power than the first Space Shuttle, and when you are on the plane back from Paris, you don’t have the Cloud! Indeed, our first design objective was: “No Internet”.
So everything is done natively on-device. The only internet API call we make is a Google Location API to interpret coordinates for the book title (“Summer in Japan”) instead of “Summer in N44. 56 E78. 63” but it’s nice to have. Without location, we simply call it “Highlights from Last Summer!”
Everything is GDPR friendly as all analyses data of your media are on your phone, and never sent to the cloud. Upon deleting any particular photo, it disappears with the photo or upon uninstallation of the app.
In the era of iCloud, Apple Photos and Google Photos, we believe it’s strategically important for our customers (photobook printers and marketers) to “get into the camera”; get to those pictures before the cloud does. Once your user syncs with the cloud, its game over.
Q: What kind of expertise does muvee have to offer photo books? Your company is not known for its printing expertise.
Swee: muvee invented automatic video editing in 2002, garnered more than 26 patents during that time revolving around media analyses, summarization, storytelling etc. We accidentally became the world’s first video app company before the word “app” was coined when we worked with Nokia to deliver a video editor on the world’s first phone that recorded video (Nokia 7610, circa 2005.)
We have incredibly deep experience with summarizing hours of video on a phone. Our deep expertise is in delivering video solutions to handset makers like Motorola, Samsung, LG, HTC, and more. Analyzing 3,000 photos in your phone to find the best 50 photos from last weekend’s in Disneyworld trip is a walk in the park for us. After all, 3,000 photos are just 100 seconds of video! We write code that is “close to the metal”, and due to our work with handset makers, we often get the “inside track” on particular APIs to use for certain chipsets and GPUs. How some memory systems on some phones are shared between CPU and GPU, and how some are not (hence incurring time writing back data when processed by different chips). I could go on…
Q: What will the user experience be like?
Swee: The best time to make your vacation book is on the flight back from your trip! That’s when you look at all your photos and reminisce, and that’s when you should make that book. When you land and get home, the app connects to Wi-Fi and uploads the hi-res page layouts to be printed. The user can pay with an ApplePay fingerprint, and voila. The book can be at your doorstep four days later. That’s our “ideal” scenario. The SDK has multiple levels of APIs. So, for customers without an experienced dev team, they would use our high-level APIs and we take care of most of the nitty gritty underneath. For customers who wants a different workflow, they can use our SDK to assist users in choosing photos (like grouping photos of food, removing documents and whiteboards, removing duplicates, grouping persons and places), or in curation (choosing the best 50 out of a time range of hundreds), or in simply suggesting a “No Crop Zone” so that you never chop someone’s face or the Empire State Building in half…and everything in between.(Albumstory) is a key product and direction for us, and something we expect to make significant strides and innovation in the next few years, where we will continue to invent, to maintain and to improve. - Terence Swee, muvee Click To Tweet
Q: What will be the rollout?
Swee: We have already delivered Version 1.0 of the iOS and Android SDKs to several customers and we are going to be seeing releases from the first batch of customers over the Summer. We are getting ready to release V2.0 of the mobile SDKs which includes our proprietary Face Recognition SDK which is able to group the same persons together, make fairly uncanny guesses at the social graph between faces, and even able to track your child’s growth (especially in the first 5-7 years where their facial features change significantly, which is a challenge for most Face Recognition SDKs out there…we know, we evaluated over 5 SDKs in the past 5 years before finally cracking the problem on our own). The Mac, Windows and Online SDKs are releasing in August. As you rightly pointed out, we are not a printing company and have no wish to be one. We are a technology company.
The print industry has loads of excellent companies who make great print products and understands the needs of their users. We see our role in this industry as filling the gap with the computer vision mad scientists, the AI gurus and mobile engineering chops to help them compete with the Apple Photos and Google Photos out there. What we have developed came from more than 18 years of related experience, and even then, it took us more than two years to build the first workable version of Albumstory. Face recognition took even longer (and we still think it can be better). This is a key product and direction for us, and something we expect to make significant strides and innovation in the next few years, where we will continue to invent, to maintain and to improve. I foresee in a year’s time, we will only take half the time to process the same album as we are doing today (even on the same four-year-old phone)! It’s Moore’s Law all over again.
Q: How will customers use your SDK and what’s the response so far?
Swee: We are well-equipped to either toss across an SDK and API document if customers have a strong mobile software team to develop an app, or we can white label our own app for customers and integrate their e-commerce, print submissions, user login management and so on. Or we can also custom develop an app for customers who have their own ideas on workflow (or applying it to other products like yearbooks, school sports teams etc.). After all, we have developed apps for Sony, Alcatel, LG and the likes…that’s where we come from.
I think what we have created is the holy grail of photobook apps: an engine that is residing in your “camera”, and making a photobook when you don’t even know you want one! So the interest level, as you can imagine, has been overwhelming. As we are not a big company, we now find that we have to be mindful of which companies we work with each quarter to maintain the level of support they expect from us to integrate such deep and powerful experiences.