If you’re familiar with Fazan Buzdar, who until recently was VP of Product Management at Box, you’re probably familiar with the entrepreneur at a popular digital workspace platform in newsrooms (including this one). You might associate it with a certain Convo. But with an electronics background, Buzdar has long had a fascination with video and visual effects.
“I am a lifelong video and photography enthusiast, and have been making my own videos for years, but over the last few decades, video production has remained largely manual, especially in time-consuming tasks such as video editing. and we noticed little innovation,” Buzder said in an email. . “On the other hand, I noticed that the iPhone camera and sensor technology has improved step-wise over the last few years, and the image quality is now almost on par with his DSLR.”
So while at Box, Buzdar says he decided to combine the growing popularity of media with AI and machine learning innovations to improve the video capture and editing experience.Buzdar Tapped Adil Abbasa video engineer, and during his time at Twitter contributed to the infrastructure that powers the site’s live streaming capabilities. serge khan, Fahd Yakub And also Box executives Michelle Oh explores the frontiers of technology-enabled video production.
The result is a roll. New app for iOS offers Bokeh, multi-cam shots, motion graphics, and perhaps most intriguing to me: “AI-simulated” sliders, dollys and jibs.
Image credit: roll
“Our mission is to transform the world of high quality video production and set the new standard for video content creation,” continued Buzder. “Creating a great video requires a huge up-front investment in equipment, equipment, learning how to use that equipment, editing software, etc. We eliminate all of that.”
Roll targets the “prosumer” market (think influencers and podcasters, but also companies that create their own marketing materials) and consists of two products: the Roll iPhone app and the web app. . The iPhone app captures and records videos and automatically uploads them to Roll’s cloud for storage and processing. The web app, on the other hand, allows one person or her team of content creators to preview, access, share, download and edit footage.
Of course, video apps are around 10 cents. So what’s the role different? Buzdar says— like Remote video interviews, video podcasts, customer testimonials. Buzdar argues that while Zoom, Microsoft Teams and Google Meet meet some needs, they were not designed for “high quality” video production.
Roll also employs many real-time effects and (ostensibly) offers a wider range of post-production options than most video capture apps. For example, Roll records in the HEVC standard, which gives about double the bitrate and higher image quality for the same file size.Androll Up to two camera shots (a wide-angle shot and a close-up shot) can be recorded and processed simultaneously, effectively allowing users to create videos with a “multi-camera” perspective.
Role editing interface. Image credit: roll
Admittedly, multicam isn’t particularly unique. Roll is not the first app to offer it.but Buzdar says the magic is in the post-processing. Roll leverages generative AI to recreate rooms in his 3D space, allowing content creators to move a videogame-like virtual camera and simulate movements like panning left and right with a dolly or crane. .
“Today, generative AI is too often associated with creating fake content out of thin air.” Buzdar said. “that is Not our philosophy. Doesn’t generate fake pixels, people or scenes. We use generative AI purely as a productivity tool and want to democratize access to higher quality video production. ”
Buzdar explained that Roll’s AI is trained to use data to measure depth and shape and understand 3D depth within a scene, regardless of who is sitting in the room. Roll began training the algorithm using an open-source his data set commonly used for benchmarking in academia, but has since recorded over 22,000 video calls internally to create its own extensive database. is created.
The results aren’t half bad – at least in the demo footage Buzdar showed me. Some of Roll’s AI-generated bread plunges into the uncanny valley. This is the result of background objects being distorted unnaturally. As the virtual camera rotates. But in a short scene, the AI effects are compelling enough to make an eye-catching addition to the boring Remote His interview.
“We’ve done quite a bit of research on this, but I haven’t seen anyone using AI in the same way we do, combining sensor data from iPhones with large AI models in the cloud,” says Buzder. said Mr. “Our technology provides the basic functionality to simulate any visual effect you want.”
Call recording by roll. Image credit: roll
Any The visual effects seem a bit over the top. But Roll has other, more realistic algorithmic tricks. When Roll records a video, it collects metadata for use later in the video production process, such as recording and lighting conditions, camera-to-subject distance, subject’s face and body position, and more. Metadata is used to automatically calibrate your phone’s camera and sensors, as well as provide feedback and instructions on composition and lighting.
Like some other “AI-enabled” mobile video editors on the market, Roll also utilizes metadata to create fully-realized multi-camera reels within the editing cloud. No manual editing required. (Users can also change and adjust camera angles, add camera movements and visual effects, if desired.) will be able to publish directly to your social media. and aspect ratio.
“Today, it takes a lot of hardware and software to fully complete a video production.” Buzdar said. “Every time a video or audio file hops from one software to another, it loses its context and is simply passed off as a “dumb” file. We have completely rebuilt the entire video production “stack” from the ground up. Beyond traditional software boundaries, we applied AI to deliver a revolutionary capture-to-publish experience that vertically integrates and automates the entire remote video production workflow. ”
So how does Roll plan to make money? The company has so far raised money from traditional VC sources. Buzdar did not say exactly where he was. But from a revenue-generating standpoint, Buzder expects Roll to eventually grow to serve the needs of corporate organizations, especially in-house corporate marketing and video teams that pay some fee for Roll’s services. ing.
“Video production is ripe for cloud disruption.” Buzdar said. “Characteristics such as large file sizes, complex processing, and the need for multi-person edit and review cycles can greatly benefit from cloud computing such as scalable storage, AI, computing, and real-time sharing and collaboration. has been the best candidate. ”
Indeed it is true. Time will tell if rolls will be disruptors.
