Inverse Telecine is the process where video editing tools reverse
telecine process. Basically inverse telecine
(or IVTC as it is also called) brings back movie's original
framerate from
NTSC's 29.97fps to 24fps.
Inverse Telecine involves more than shifting fields from one frame to another. Depending on the quality of the video, there may be significant differences in the two fields that make up a reconstructed frame, as well as between different copies of the same (duplicated) field.
Field Matching
Field matching is the first step of the IVTC process. In order to avoid as much post processing as possible (and some will almost certainly be required) it's important that pairs of fields that make the best matches are combined. This isn't generally too complicated for frames with no motion, but motion can cause the edges of objects to shift slightly from one field to the next. Once fields are matched, you have an approximation of the original 24fps film, with additional frames inserted from the duplicate fields.
Decimation
Since the frames containing extra fields should now occur in a regular pattern, we remove them from the video, which is often referred to as decimation. After decimation we still don't have progressive frames though. We have pairs of interlaced fields, which will likely not line up perfectly. Because the encoder treated them as two separate pictures from two different moments in time, some corrections must be made to weave them together into a single picture.
Interlace Artifacts
Artifacts are imperfections that occur as the result of the encoding or decoding process, compared to noise, which is caused by interference from an outside source. Encoding film sources interlaced with a telecine pattern introduces combing artifacts. Combing refers to the jagged lines that result from the same object in two fields of the same frame not lining up correctly. This is fine for actual interlaced video, where each field is from a separate moment in time, but for film we want frames to represent individual pictures. This is a much simpler process than deinterlacing, because no decisions have to be made about whether there's motion between two fields in the same frame, but can still result in low quality if the differences between fields are too great.
