Download: VidSync 1.821 (universal, macOS 14.6 or later), also on the Download page.
Small update. There is one fix that matters only if you digitize with more than two cameras at a time, along with several corrections to distortion editing and to how reprojection error is reported.
Assigning clicks to points with three or more cameras
Clicking in a video during a measurement used to record that click as belonging to the lowest-numbered point of the active event that did not yet have screen coordinates in the camera you’re clicking. With two videos, that rule works very intuitively: you click a fish’s head and tail for a length measurement in one clip, then the other, and they’re recorded in that order, as they should be.
It could go wrong with 3+ clips (calibrated together) as follows. Click the fish’s head and tail in camera A. But suppose you have two other cameras and neither of them has a complete view of the fish: camera B sees only the head, camera C only the tail. You click the head in camera B, it matches up to the first point without problem. Now you go to click the tail in camera C, but under the old rules, VidSync interpreted that as you clicking a third point for the head (because camera C didn’t have a head point yet) instead of a second point for the tail as you intended.
There is now a program-side setting in Preferences under Point Measurement called “max sight lines per point,” which controls how many clicks each point want to accept. The default value of 2 will make projects with 3+ videos feel more like the intuitive 2-video workflow. You should only set it to 3 if you actually want to triangulate a point position from 3+ lines in a video like this instead of 2. I’m not aware of anybody actually wanting to do that.
The basic key to digitizing multi-point events (like length measurements or other distances) smoothly is to always digitize the points in a consistent order: head-tail in one video, head-tail in other videos, or alternatively, head-head and then tail-tail. Never head-tail and then tail-head. The correct process was broken for sets of 3+ videos, and it should now work pretty normally.
Option-click to name the point. If for some reason you’re working in a 3+ video project and you want your click to define a point other than the default under the settings option described above, you can hold down the Option key. The click will then be recorded on whichever point is selected in the points table, ignoring both the numbering order and the limit.
Connecting lines (1.821)
VidSync has always drawn “connecting lines” between event points (e.g. length and speed measurements) as a straight line on screen between the two measured points. As a part of the revisions to make projects with more than two camera views work more intuitively, I changed how those lines are drawn: now they actually show the projection onto the 2D screen of the 3D line that connects the two points in 3D space. In plain English: connecting lines will now sometimes follow a slight curve between the two points (depending on your lens distortion) instead of a straight line on the screen. This is more geometrically honest, and it looks cool.
This is a display-only change which does not touch scientific results at all. Any old projects you open will automatically display their lines using the updated method.
I made this change as part of a broader revision to connecting lines for that 3+ camera situation, which sometimes called for connecting lines to be drawn between points when they aren’t both on-screen in the same camera. The correct view would be a single point with the connecting line running off the edge of the screen. This never worked before. Now it does.
Also, the event type setting for whether connecting lines show on all clips or only the master clip was reinterpreted for this 3+ video case only: it now interprets “master clip” differently (with regard to that one setting only) to try to replicate the same qualitative intent: labeling the one best view of the line rather than all of them.
Other changes
- Deleting a distortion point no longer throws the magnified preview down to the video’s bottom-left corner. The preview now stays where you were working.
- Nudging a chessboard corner that lies on two crossing plumblines now moves both stored copies of that corner. Previously each had to be found and corrected separately, and correcting only one left two contradictory observations of the same corner in the distortion fit.
- Switching iterative triangulation on or off now recalculates the existing 3-D coordinates. Before, a project could hold a mixture of both methods’ output while every export claimed a single method.
- Reprojection error is now reported for each point when iterative triangulation is off. That column was previously empty in this case.