Frame Blending

We have all by now seen the press coverage of the recent events in Boston, USA. The speed to which video was obtained, viewed and then subsequently released to assist in identification, was obviously as a result of many hours of teamwork. The other point, that is often overlooked, is that it quickly subdued the social media investigators, who were unwittingly pointing the finger at innocent people.

Anyway, this is not the place for discussing many of these issues.

The issue that I thought was worth mentioning is one that a number of video technical experts have already commented on. The frame blending seen within the media file created from the surveillance footage. Take a look here….

The blending, also referred to as ghosting, is an often seen result of transcoding video from a low frame rate to a higher one.

In the Boston footage example, we have a low frame rate surveillance video. The three camera views have different rates. It would appear that the three views of interest have been cut together and then transcoded to one video file. It may have been at this point where conversion has added in the blended frames. Rather than the static frame being duplicated for x number of frames, it blends each frame into the next across x number of frames.

The transcoding is often an automated process. The reasons for transcoding are many, but in the CCTV World, it is often to enable proprietary video to be universally playable. If the options included in your transcoder do not enable you to identify how it deals with the frame rate conversion, you may need to pre-process prior to any transcoding.

To explain this further….

Footage lasting 3 seconds that had been recorded at 6 frames per second. That’s 18 individual images.

I need to place this onto PAL Video DVD which is standardised at 25 Frames per second. The transcoder looks at the duration of the video and then adds in frames to reach 75 total frames. As mpeg2 is a motion predicted form of compression, the transcoder detects changes in the images and perceives this as movement. It uses a pattern of apparent motion known as Optical Flow.

The optical flow vector of a moving object in a video sequence.

Our problem though is that there wasn’t any movement – The images are static. The transcoder doesn’t know that though and as a result, it blends the two static images together. This then results in a video with the blending / ghosting.

Most higher end transcoders and NLE’s will have an option of how to deal with the low frame rate and any conversion required.

In Virtualdub, you can manage the frames prior to using a transcoder.

Vdub-frame rate

Instead of changing the frame rate, which will then play our example video back very fast, I convert the frame rate, prior to using a transcoder. This then adds in duplicate frames to make up the 75 total frames for the duration.

When it comes to transcoding, there is no frame adding required and as such no frame blending / ghosting will occur.

I hope this helps to identify the problem and it is also worth testing and verifying any software used regularly to identify what methods are best to avoid this issue.

By Spreadys Posted in EEPIP

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s