In Audio and Video Call or Interactive Live Streaming, video quality is compressed for network bandwidth limitations and losses in encoding and decoding. CommsEase provides an AI Super-Resolution (SR) algorithm that can restore low-quality videos and image details, and improve the image quality of video calls.

Background

In a real-time video call, limited by the network bandwidth, a video is sometimes transmitted in low resolution. The quality of decoded and encoded image is compressed. Therefore, the receiver experiences a poor video quality, which effects the quality of the video call. The self-developed AI SR algorithm of CommsEase is improved after many data testing. To improve lower video resolution, we reconstruct high-resolution images based on the original low-resolution images to restore the image details. In video calls and interactive live streaming scenarios, the AI SR feature can restore low-quality videos in real time and enhance the video call experience.

Visual display:

Considerations

• Before using the AI SR algorithm, contact CommsEase technical support to enable the AI SR feature.
• The AI SR feature only applies to following types of video streams.
• The first 360p video stream received by the local client.
• The bigstream captured by a camera. The AI super resolution feature does not currently support restoration and reconstruction of low streams and screen sharing in the substream.

Procedure

Call enableSuperResolution to enable or disable the AI super resolution feature.

After the local side joins the room and enables the AI super resolution through enableSuperResolution, the SDK will perform AI super resolution for the first received video stream to improve the image quality and restore the image details. After the user leaves the room, the SDK will select the next eligible video stream for AI SR feature.

Sample code:

NERtcEx.getInstance().enableSuperResolution(enableFlag);