Superpose Wants AI to Teach You How to Pose, Not Edit Your Photos
Melody Chu and Jing Liu, two former TikTok employees, have launched Superpose, an iPhone camera app that uses AI to suggest how to pose for photos. TechCrunch reported the launch on September 15, 2026.
The flow starts with a normal photo. Take a selfie or photograph a friend, and the app generates four possible poses to try next. The output is instruction, not retouching. It tells the subject how to stand, how to angle the body, or where to move, rather than editing pixels after the shot.
A second feature, called match pose, handles lining up the shot. It places the suggested pose over the live camera view, so the person in front of the lens can line up head, arms, and body with the guide. The loop is short. Shoot, review four options, align, shoot again.
Chu brings product experience from large consumer platforms. She worked in product roles at Meta, Nextdoor, Roblox, TikTok, and Slack. Liu brings imaging depth. She was a founding engineer at a 3D face-scanning startup and later worked on image and video models, the AI systems that create and interpret photos and video, at TikTok.
Superpose launched in July and has passed 22,000 downloads. Users have generated more than 190,000 poses to date.
Looking at those early numbers, use looks repeated rather than one-off. With 190,000 poses from 22,000 installs, the average install is producing multiple generations.
Users get five free generations, or sets of AI pose suggestions, per day. Extra packs cost $2.99 for five generations and $9.99 for 20 generations. Light daily tries stay free under that limit, while longer sessions, such as a portrait shoot or a group outing where more variations help, cost extra.
The company has raised $2.2 million from Khosla Ventures, Meitu, and OVTR Ventures. Its site, superposelabs.ai, describes the product as transforming anyone into a skilled photographer through real-time AI guidance. Superpose
Looking at what this means for consumer imaging tools, the choice to coach the subject instead of rewriting pixels stands out. Much recent AI photography has centered on post-capture editing, background replacement, or synthetic beautification. Superpose leaves the captured image alone and steps in earlier, at direction. For builders, that moves the hard work to pose conditioning, or steering the model with a target posture, identity preservation across suggestions, and low-latency overlay tracking that stays steady as hands shake and light shifts.
In my view, the pricing deserves attention. Selling packs per generation ties price directly to inference cost, the computing cost of running the AI model, which is straightforward but leaves little room for error. Five free per day is enough to form a habit. The test will be whether people treat four suggested poses as throwaway ideas or as something worth keeping at about $0.50 to $0.60 each. I watched my own children adopt portrait tools over the years, and the pattern held. They stuck with what made taking and sharing photos faster and more social, not what made one image technically cleaner.
The broader context here is encouraging. Camera skill has never been evenly shared. Having a capable friend with an eye for framing often made the difference between an awkward photo and a keeper. If an on-device coach can narrow that gap without pushing users toward fully synthetic portraits, more people get usable, shareable images from hardware they already own. That is a modest goal, and that modesty is its strength.
