Our four reconstruction methods are:
• Spatial
• Temporal With Static Scene
• Temporal With Moving Scene
• Clean Plate.
The spatial method takes the background information from adjacent pixels in the current frame, and the clean plate method takes the information from a separate clean plate input. The other methods are temporal and attempt to get background information from frames either side of the current frame. Where this information is not available, for example, because the wire covers part of the same region in one or more of the other frames, the spatial method will be used for the repair.
The spatial method is fastest. It uses a slope-dependent filter that interpolates across the wire at the most likely angle, given the image behind the wire. It works well when the wire is over backgrounds such as sky, clouds, or smoke, and can also cope with some backgrounds where there is a noticeable gradient, such as the edge of a roof, building, or car. If this fails and the wire is moving relative to the background, you should try one of the temporal methods. These look at frames before and after the current one to find likely pixels with which to repair the region and use the spatial method in areas where there are none available.
Where traditional clean plates are possible or give better results than using F_WireRemoval to repair the wire on each frame, you can supply a clean plate as the fourth input to the plug-in. It will then automatically match-move it to repair each frame. If the overall luminance of the background is different to that of the clean plate or varies during the sequence, turn on Luminance Correct. The luminance of the background pixels used for the reconstruction will then be adjusted before the repair is made. Of course, various FurnaceCore tools can be used to create a clean plate, including an F_WireRemoval pass on a single frame where the repair has been successful.