Some video sources output an interlaced video signal. Each frame is split into odd and even horizontal scan lines called fields which are sampled at different time instances. This effectively results in doubling the frame rate without increasing bandwidth requirements. To eliminate “combing” artifacts which are produced by interlaced frame acquisition, the deinterlacing enhancement should be used.
|
Low
|
Low is the fastest and the least accurate. With this setting, deinterlacing is performed by discarding the odd field and upscaling the even field to fit the video’s vertical resolution. The processing uses a bicubic interpolation algorithm to upscale the even field. |
|
Medium
|
Medium is more computationally demanding and produces better results. With this setting, deinterlacing is performed by discarding the odd field and upscaling the even field to fit the video’s vertical resolution. The processing uses a more advanced algorithm, known as kernel regression, to upscale the even field. The edge-preserving properties of the kernel regression algorithm is able to handle slant edges better than the bicubic interpolation used by the Low setting. |
|
High
|
High uses both odd and even fields to perform deinterlacing. If used in combination with Resolution enhancement, this setting instructs Multi-Frame Super-Resolution processing to use both odd and even fields. |
Tips:
- Low and Medium settings may yield better results for deinterlacing videos showing a scene with non-rigid objects with uneven lines and textures, like people and animals.
- Using High setting with Resolution enhancement may result in a significant drop in processing bandwidth