< Previous | Contents | Manuals Home | Boris FX | Next >
Advanced Uses and Limitations
Since the single-frame alignment system is pretty simple to understand and use, you might be tempted to use it all the time, to use it to align regular full 3-D camera tracking shots as well. And in fact, as its use on tripod-mode shots suggests, we have made it usable on regular moving camera and even moving-object shots, which are an even more tempting use.
But even though it works fine, it probably is not going to turn out the way you expect, or be a usable routine alternative to tracker constraints for 3-D shots.
First, there’s the accuracy issue. A regular 3-D moving-camera shot is based on hundreds of trackers over hundreds of frames, yielding many hundreds of thousands of data points. By contrast, a line alignment is based on maybe ten lines, hand-placed into one frame. There is no way whatsoever for the line-based alignment to be as accurate as the tracker solutions. This is not a bug, or an issue to be corrected next week.
Garbage in, garbage out.
Consequently, after your line-based alignment, the camera will be at one location relative to the origin, but the trackers will be in a different (more correct) position relative to the camera, so…. The trackers will not be located at the origin as you might expect. Since the trackers are the things that are locked properly to the image, if you place objects as you expect into the alignment-determined coordinate system, they will not stick in the image—unless you tweak the inserted object’s position to make them match better to the trackers, not the aligned coordinate system.
Second, there is the size issue. When you set up the size of the alignment coordinate, it will position the camera properly. But it will have nothing to say about the size of the cloud of trackers. You can have the scene aligned nicely for a 6-foot tall actor, but the cloud of trackers is unaffected, and still corresponds to 30 foot giants. To have any hope of success using alignment with 3-D solves, you must still be sure to have at least a distance constraint on the trackers. This is even more the case with moving-object shots, where the independent sizing of the camera and object must be considered, as well as that of the alignment lines.
The whole reason that the alignment system works easily for tripod and lock-off shots is that there is no size and no depth information, so the issue is moot for those shots.
To summarize, the single-frame alignment subsystem is capable of operating on moving-camera and moving-object shots, but this is useful only for experts, and probably is not even a good idea for them. If you send us a your scene file at tech support looking for help on that, we are going to tell you not to do it, to use tracker constraints instead, end of story.
But, you should find the alignment subsystem very useful indeed for your tripod- mode and lock-off shots!
©2024 Boris FX, Inc. — UNOFFICIAL — Converted from original PDF.