Behind the Video Scenes
A video represents a series of still images that are displayed in a rapid succession. If the image display is not quick enough the resulting picture will be shaky. A smooth video playback on TV requires roughly 50 images per second. But let’s take a closer look.
A TV frame consists of 625 lines. The 576 of them are visible as they bear the information about the image, with 49 being supporting without any image content. To obtain a video signal the transmitted image gets decomposed into lines. There are 2 ways to do that: interlaced and progressive scanning.
Interlaced Scanning
With the interlaced scanning the lines get displayed on the screen out of sequence, e.g. not 1st, 2nd, 3rd, etc. They appear in the following order: at first odd-numbered lines and then even ones or vice versa. A frame gets divided into 2 half-frames which are called fields. Each field consists of odd numbered (1st, 3rd, 5th, etc.) and even numbered (2nd, 4th, 6th, etc.) lines. If a field has only odd-numbered lines, it’s an odd or upper field. A field containing only even-numbered lines is an even or lower field. The fields can appear in a different order.
Progressive Scanning
Progressive scanning allows to transmit each line of the frame one after another, e.g. 1st, 2nd, 3rd, etc.
Interlaced vs Progressive Video
- A progressive video signal requires almost double bandwidth for transmission in contrast to the interlaced video.
- Progressive videos don’t have visual defects that are often called a "comb". Such drawbacks are usually associated with interlacing.
- A progressive video image doesn’t need anti-aliasing to eliminate flickering and comb. So the artifacts won’t appear.
- An interlaced video can’t be scaled to a higher resolution unlike a progressive one.
- An interlaced frame has 2 fields, so it can’t be saved as a separate image. Progressive scan allows this.
Here are some deinterlace methods or filters that are widely used today to remove the interlace defects.
Disabled
No deinterlacing is provided. Everything is left as it is. It’s used for PsF content.
Blending
It’s somewhat a mixture of odd and even lines. With blending the lines are not displayed at different times.
Bob
This filter doubles lines and displays each part as a full frame. It keeps the temporal resolution that the interlaced video had.
Discarding
With discarding method only one of the half-frames is exposed, the other one is left behind. The interlaced resolutions get cut in halves. This filter is great for slow computers. It also gives interlaced videos a movie-like look with judder.
Linear
It’s a combination of Bob and linear interpolation. Each line is not doubled but is made as the average of the previous and next coming lines.
Mean
This filter illustrates a half-frame that was made as the average of 2 original half-frames.
X
It creates a full image by using the odd lines from the odd half-frame. The even lines appear with the help of a complicated algorithm that takes data from both half-frames.
Yadif (v1.1.0+)
The filter makes a full image of the odd lines taken from the odd half-frame. It creates the even lines via a complicated algorithm that uses temporal as well as spatial interpolation.
Yadif (2x) (v1.1.0+)
It’s a mix of Bob and Yadif interpolation.
Phosphor (v1.2.0+)
Phosphor displays the latest 2 half-frames. The old one fades out. This fade effect can be easily configured.
IVTC (v1.2.0+)
IVTC is an abbreviation for inverse telecine. The filter deletes telecine from NTSC telecined video in real time. As a result, the progressive signal is restored without losses.
