Importing Images

Postshot supports numerous image and video file formats as inputs. This includes widely used file formats like JPEG and MP4, RAW image formats from most camera vendors as well as lossless and HDR video files.

The easiest way to import images is to drag a video file or a folder of image files into Postshot. See Importing Images or Videos for more details. For such an image-only import, Postshot will compute the camera poses and a sparse set of points from the imported images. These are used in turn to start the radiance field training process.

You can also use other apps to compute the camera poses and sparse points first, then import them together with the images. Postshot will then skip ahead to the radiance field training process.

This page describes the combinations of files you can import to generate radiance fields in Postshot.

To import one or several videos or sets of images, drag and drop some of them into the main Postshot window. After the import window pops up you can drag and drop additional files into it to add to the import files.

You can then adjust the training parameters (see Training Configuration for details) or just use the defaults to begin reconstructing the scene. Once the training process has started, images cannot be added or removed anymore.

Importing Images, Camera Poses and Sparse Points

When importing image files, you can optionally also import camera poses and points that have been extracted from the images using external tools. If the import file list contains supported camera poses, the Camera Poses parameter in the Training Configuration will switch to 'Import' and the Image Selection parameter will switch to 'Use All Images'.

When training with imported camera poses and points, Postshot will skip the image selection and camera tracking stages and start training the radiance field immediately.

Postshot supports the following file formats for camera poses and points:

COLMAP *.bin or *.txt files

The three files cameras.bin, images.bin and points3D.bin or their *.txt equivalents must be included. These files include the intrinsic camera parameters, extrinsic camera parameters and sparse points, respectively.

Bundler *.out and *.ply files

Since *.out files only include camera parameters, the sparse points must be included in a separate *.ply file.

Also note that cameras in Bundler format are always undistorted. The images you import along with these camera parameters must be undistorted as well. The tool that created the Bundler file should also have exported undistorted versions of the input images. If the original, distorted images are imported instead, they will be inconsistent with the camera poses and splat training will most likely create unusable results.

Reality Capture *.csv and *.ply files

To export camera poses from Reality Capture for use in Postshot, select the ALIGNMENT tab, then the Registration option from the Export sub-tab. In the Export Registration file dialog set Save as type to Internal/External camera parameters. This will save the poses to a .csv file. In the following window Exporter Settings, keep all settings at their defaults.

To export the points, select the Point Cloud option from the Export sub-tab. In the Export Point Cloud file dialog set Save as type to Sparse point cloud as Polygon File Format (*.ply). In the following window Exporter Settings keep all settings at their defaults except for the following: