Uploading a Capture

Create a Capture by uploading the images or a video, which will be later processed into a 3D Gaussian Splatting (3DGS) model. This documentation covers guidance around the upload process, limitations on the supported source files and long-term file management.

Create and upload your Capture

A Capture is a named collection of source files that Blurry can process into one or more 3DGS models. Each Capture can contain either a set of images or a single video.

You can add, remove, or replace individual files while preparing your Capture. Once everything is ready, continue to Processing a Capture to choose your training parameters and generate a new Scene.

Supported source files

Blurry currently uses Teleport by Varjo as its processing backend, so Teleport’s input requirements also apply to Captures in Blurry.

TypeMinRecommendedMaxFormats
Photo200 images500 images3,000 imagesPNG, JPG or HEIC
Video2 minutes5 minutes25 minutesMP4 or MOV

A Capture can’t mix images and video. If the number, format, or combination of files isn’t supported, Blurry displays a validation error so you can correct the selection before processing.

Source file locking

You can change the source files until the Capture's first processing run is started. After processing begins, Blurry locks the source files to keep every model generated from that Capture tied to the same input data.

Reusing the Capture

Blurry stores your Capture files securely in the cloud. These files essentially become part of your reality capture library, the same way as Scenes. The same source files can be reused for multiple processing runs. For example, you can process them with different splat counts, resolutions, or spherical harmonics settings without uploading the files again.