I have recently had films scanned on the RetroScan Universal as part of a wider comparison of 8mm film scanners. I asked for the perforations to be included so that I could check the quality of registration and determine the causes of frame movement. Here’s what I found for Standard 8mm and Super 8mm. Video files mentioned below can be downloaded from: http://www.mediafire...(RetroScan).zip
The easiest way to see the frame movements is to open the files in Quicktime and step through, frame-by-frame.
Frame Jumps caused by camera
One of the films (R01a, Standard 8, 1947) has a cyclic, horizontal movement of the frame. Every six frames the image moves sideways relative to the perforation, then gradually moves back over the next five frames. There is only a minor impact on frame registration.
Frame Jumps caused by scanner
At splices, the RetroScan typically causes a vertical jump of between 5% and 15% of frame height, mostly at the lower end of that range. The jump seems to be caused by the splice tape covering part of the perforation, thus confusing the perforation detector. Examples are R01b, R01c, R02a, R02b, R04a, R04h.
For Standard 8, the jumps occur 7 frames before the splice; for Super 8mm, 10 frames before the splice. The jumps last until the splice itself is scanned, then the frame reverts to the normal position.
I suppose this means that, for the RetroScan, there is a single perforation sensor at a fixed distance behind the capture point.
At Scene Changes
A small number of jumps are caused by scene changes, where the incoming scene is brighter than the previous scene. The perforation detector seems to be confusing part of the bright scene with the perforation. The jump is typically 5% to 10% of frame height and lasts 1-10 frames.
Examples are: R04b, R04c, R04e, R04j
A zoom, in or out, can cause a jump, probably because of a brightness change associated with the zoom. i.e. if the zoom causes a white area to become larger (R04g), the extra brightness upsets the perforation detection. But a zoom into a dark area (R04L) can also upset the perforation detection. Typical jumps were 5% and lasted 10 frames.
Examples: R04d, R04i, R04j, R04k, R04L
If a bright area is near the perforations, it can cause erratic frame jumps. Examples are R04g, R04k
Occasionally there are erratic jumps that I can’t explain (R04f).
Dirt in Perforation
Dirt partly blocking the perforation can cause a jump (R04m).
1. All frame jumps were vertical, not horizontal.
2. Jumps can be caused in several ways: zooms into bright or dark areas, scene changes into a bright scene, dirt in a perforation, or bright or dark areas around the perforations.
3. For scans which are cropped inside the image area, frame jumps will cause problems because they will cause loss of part of the image.
4. Frame jumps will be less of a problem if the scan area is expanded by 10-15% of frame height because no part of the image will be lost during the jump. The jumped frames may be out of registration with the others, but at least they are intact.
Do high-end machines – Scan Stations, Kinettas, Xenas and the like – have any of the above problems?
Given that most home movies tend to have camera wobble and would benefit from a small amount of stabilization in a video editor – stabilization that would also bring the jumped frames into alignment – how important, really, is accurate frame registration for home movies?