Insta360's Spatial Capture technology rolls out in the US on September 28 for X6, X5 and X4 Air models, turning 360 video into explorable 3D scenes.
Insta360 is making the most of its semi-open shot at the US market due to the absence of rival DJI with the opening of its first North American flagship store in Times Square. At the same time it has officially unveiled Spatial Capture for the Insta360 X6, a feature that transforms 360 video into explorable 3D scenes. This rolls out in the US on September 28, with support also extending backwards to X5 and X4 Air.
For those that need a quick primer on the tech, 3D Gaussian Splatting (3DGS) uses images or video to recreate real-world spaces as 3D scenes. Getting decent results has typically meant capturing a space from multiple angles, then using specialized software to process the footage. Given that the X6’s two cameras capture every direction at once to create an image, that means that any footage captured by the device can be converted into an explorable 3D scene without repeated repositioning.
It’s a simple workflow. Users record a 360 video of the space they want to capture with the X6 up to two minutes long (longer recordings can be trimmed before processing). In the Insta360 app, they then select their footage on the Spatial Capture page and tap "Generate." This takes some processing grunt and is not something set up for instant gratification. Processing takes place in the cloud and typically takes one to two hours (though happily users can close the app while they wait and get a notification when the scene is ready). Users are also restricted to generating four 3DGS scenes for free each month. No word yet on the price of scene generation beyond this usage cap.
The finished scene can be explored using touch controls, a virtual joystick or the device's gyroscope. Users can add effects and create custom camera paths, then export the result as a video or share an interactive link that lets others explore the scene without downloading an app or file.
Here’s an example of what is produced, captured in what looks to us to be Stratford-upon-Avon, courtesy of Hugh Hou.
It’s interesting technology and feels quite similar in tone to the Dual Pixel 3D imaging introduced with the Canon EOS R8 Mark II. This is a Windows app that takes the Dual Pixel data generated for the AF in compatible EOS R cameras and uses it to create 3D images and models, though judging from Canon’s page about the technology, so far it seems more aimed at individual objects than entire scenes.