HikVision FPS vs i-Frame vs Disk Station performance

[myrotfl 0]

Still adjusting parameters on my Hikvision DS-2CD2142FWD-I cameras (2) and Synology DS-213j Surveillance station.

 

I think I have some pretty good settings so far, but just not sure about the FPS vs I-Frame settings.

 

I tried setting my FPS and I-Frame settings to the same (6 FPS and 6 for I-Frame interval) and my Diskstation (DS-213J) started to peak on CPU usage hovering around 90% (or more) most of the time.

 

I then left my FPS settings at 6 and changed my I-Frame settings to 12, and the Disk Station CPU seems to be in the 50-70% range most of the time, sometimes a little lower.

 

Not sure I completely understand the I-Frame settings, but from what I understand a higher I-Frame setting means that the camera sends an I-Frame less often than a lower I-Frame setting.

 

 

Posting here because no response on Synology forum, maybe not many there familiar with I-Frame setting.

 

Any problems I need t be aware of with my FPS and I-Frame settings?

 

Any help much appreciated.

 

Andy

[myrotfl 0]

I don’t like to answer my own questions, but in this case I feel it may benefit others.

 

Before I posted here about I-frames vs FPS I had been scouring the Internet to try to find something that explained it to me. After I posted here I continued to look.

 

I finally found something that I believe is an answer to my question. I found a page at:

 

Understanding H.264 Encoding Parameters - I, P and B-frames

 

http://www.streaminglearningcenter.com/articles/producing-h264-video-for-flash-an-overview.html?page=4

 

It explains that I-frames: Also known as key frames. I-frames are completely self-referential and don't use information from any other frames. These are the largest frames of the three, and the highest-quality, but the least efficient from a compression perspective.

 

P-frames: P-frames are "predicted" frames. When producing a P-frame, the encoder can look backwards to previous I or P-frames for redundant picture information. P-frames are more efficient than I-frames,

 

So from this I gather that I-frames contain the most information and that they are not very compressible as they contain all the information about the scene.

 

The majority of frames are P-frames and contain only the information that has changes since the last frame.

 

And that you don not require a lot of I-frames to get a good video.

 

I have set my FPS to 6, and my I-frame interval to 18. That means that I get one new I-frame every 3 seconds. I have noticed that doing this has produced a significant decrease in the load on my DS-213j.

 

For those interested I suggest you read the page referenced here as I believe it can help to optimize the camera settings to help get the best video with the lease CPU and network usage.

 

IHTH

 

Andy