Frequently Asked Questions
Table of Contents:
- Question 1: Which other components (python or otherwise) are required to run AfterScan?
- Question 2: What is the effect of the fill modes (no fill, dumb, fake) listed at the bottom of the Frame post-processing area?
- Question 3: After encoding a big reel (diameter > 7.5cm), I see some fragments are not correctly stabilized/cropped. What can I do to solve this?
- Question 4: Starting with AfterScan 1.9.0, the stabilization is not working as before. Is there any way to have the same behavior as before 1.9.0?
Q: Which other components (python or otherwise) are required to run AfterScan?
A: In order to run AfterScan you need to install, at least (there might be others), the following components (these are the package names in Linux, in Windows or MacOS might be different):
- python3-tk
- python3-pil.imagetk
- python3-numpy
- python3-matplotlib
- python3-opencv
- python3-psutil
- ffmpeg
Q: What is the effect of the fill modes (no fill, dumb, fake) listed at the bottom of the Frame post-processing area?
A: When scanning a film, no matter how careful we are, there are always some frames that are badly captured: The frame is not completely inside the captured image, as we can see in this sample:
- No frame fill: This one if the simplest option, and the only choice in early versions of AfterScan. The missing fragment is simply left as-is, black
- Dumb fill: In this case the missing part of the frame is simply filled with the colors of the last pixel row before the missing part. Not very nice, but better than black
- Fake fill: This option is normally the best choice, despite it's name. The missing part of the image is filled with the fragment in the same position on the next/previous frame, which appears in the same capture. Most of the time this will cheat the viewer (despite being 'fake') as two images in sequence are quite similar, most of the time. In some cases (change of scene in the misaligned capture, small cropping selection) will not work, but worth to try.
Q: After encoding a big reel (diameter > 7.5cm), I see some fragments are not correctly stabilized/cropped. What can I do to solve this?
A: Big reels are usually the result of joining several individual films, shot at different times, maybe with different cameras. In these cases (specially if different cameras were used) the position of the sprocket holes relative to the image frame will most likely be slightly different. Since that sprocket hole is the reference used to stabilize the collection of frames, if the relative position is not the same for all the stabilization and cropping will be affected. To prevent this problem, the solution is to encode each film separately by unchecking 'Encode all frames' and selecting the start/end frames for each fragment) so that you can have specific cropping rectangles for each. That should avoid the problem
Q: Starting with AfterScan 1.9.0, the stabilization is not working as before. Is there any way to have the same behavior as before 1.9.0?
A: Yes, it is possible. 'Custom hole template' button, briefly removed before 1.9.22, is now back. The tentative to have templates fully automated was working well in some cases but not in all (*). Therefore the new option is kept as default option, but custom templates are stil the recommended option to have a good stabilization adapted to each film. However, if you still experience troubles in this area, one possible reason why it might not work is that the film being treated was scanned not leaving enough space for the sprocket holes. If you cannot clearly see the sprocket holes in the frames you have scanned, chances are AfterScan won't see them either, causing the stabilization to fail. Below you have two examples of valid and invalid captures:
Valid capture: Sprocket hole visible (hole width should not be less than this)
Invalid capture: Sprocket hole barely visible