# Introduction

Monitors of digital playback devices such as computers or smartphones display videos in _Progressive_ mode, which means each frame is displayed as one pixel row after the other.

Older analog devices such as CRTs display videos in _Interlaced_ mode. This means, each frame is divided into two _fields_. One field contains the odd-numbered pixel rows (1, 3, 5, ...) and the other field contains the even-numbered pixel rows (2, 4, 6, ...). When it was introduced 70 years ago (primarily for television video formats like NTSC and PAL), Interlacing reduced flicker and provided well distributed brightness on the screen. With today's digital screens this is no longer necessary.

Many cameras and equipment still support the interlaced format and thus, many source files contain interlaced video (e.g., if your source files originate from a broadcast playout system, there is a high chance that they may still be interlaced).

Interlaced files are labeled with an "i" for the frame rate, e.g.

When encoding interlaced video into Progressive mode, the encoder must construct each frame from two fields.

Typically however, the fields have been recorded separately, which means that two consecutive fields show the image with a time difference of double the frame rate (e.g. 1/50 second). Especially when the video contains fast movement, this results in undesirable artifacts known as "the comb effect" where the odd and even pixel rows are clearly visible in the frame. The result is a blurry video quality.

Thus, to remove the comb effect and improve video quality, _Deinterlacing_ must be applied to the video before encoding. The Bitmovin Encoder includes a Deinterlacing filter which does exactly that.

**Please note:** _If possible, check with your content provider if they can provide the sources as Progressive mode. They will be of better quality than deinterlaced even with the best filter._

# The Deinterlacing Filter

The Deinterlacing filter's documentation can be found here: [Create Deinterlace Filter](🔗)

A Java SDK example that uses the Deinterlacing filter can be found here: <https://github.com/bitmovin/bitmovin-api-sdk-examples/blob/master/java/src/main/java/Filters.java>

The example is also available in other languages.

# Using the Deinterlacing Filter

We assume that you worked with the Bitmovin examples before and know how to configure an encoding. If you are new to Bitmovin encoding, please refer to our tutorials such as [Get started with the Bitmovin API](🔗) and closely check the examples we provide in your preferred language: [SDKs](🔗).

In the _Filters.java_ example, after creating the video codec configuration

and after creating the video stream

you need to apply the filters. You do this by creating a list of filters, adding all filters you need to this list, and applying the filter list to the video stream.

The example shows the mechanism in the _createStreamFilterList()_ function. In that function, a list of _StreamFilter_ objects is created. Then, each filter is registered in a _StreamFilter_ object and added to the list. Note that you can determine the order of the filters with the _position_ parameter. At the end you create the filters and apply them to the stream and the encoding.

# Configuring the Deinterlacing Filter

The most interesting function of the previous section is _createDeinterlaceFilter()_ as it contains the configuration of the filter.

In the example however, this function is as simple as it can be:

Here, a default Deinterlace filter is created.

If you want to configure the Deinterlace filter differently than default, you can do it by altering the function:

The `(...)` describes the place where you can set optional configurations.

The fields you can configure are:

• mode

• frameSelectionMode

• parity

• AutoEnable

Following the description of the fields in detail:

**mode**

**frameSelectionMode**

**parity**

This sets the picture field parity assumed for the input interlaced video. As said above, in interlaced video streams a frame comprises of two fields. One includes the odd pixel rows (1, 3, 5,...). This is called the _top field_. The other includes the even pixel rows (2, 4, 6...) This is called the _bottom field_. When reconstructing a frame from a stream of alternating fields (top, bottom, top, bottom...), it is important to understand whether a frame starts with a top field or a bottom field.

**autoEnable**

This value controls the application of the filter per se. With this comes in certain cases an alteration of some configured parameters.

The possible values of this field are:

• ALWAYS\_ON (default)

• METADATA\_BASED

• META\_DATA\_AND\_CONTENT\_BASED

This table shows the consequences of each value

**Advice** for using _autoEnable_

• If you know exactly what your input is, and you know how to set _mode_, _parity_ and other settings, then set _autoEnable_ to `ALWAYS_ON` (default).

• If you don't, or want to allow for variable input files, then it's safest to use `META_DATA_AND_CONTENT_BASED`. This may overwrite other settings if the source file analysis indicates a contradiction.