Subscriber Discussion

What Are The Pros And Cons Of Client-Side Vs Camera-Side Dewarping Of Fisheye Cameras?

U
Undisclosed #1
Jun 14, 2018

Was reading the new Hikvision fisheye test and I don't understand the tradeoffs between client side and camera side dewarping. Anyone can help.

U
Undisclosed #2
Jun 14, 2018

Takes the processing load of the client/server side. Dewarping at edge makes it much more feasible to display multiple, many fisheyes dewarped at same time. 

MM
Michael Miller
Jun 14, 2018

Camera-side dewarping is like a PTZ camera. You only record you're what looking at. 

Client-side deparping you are recording the whole image so you can dewarp the image live and recorded.  Plus multiple people can be dewarping different parts of the image at the same time. IMHO this is much better than camera-side.

(3)
(1)
(1)
U
Undisclosed #2
Jun 14, 2018

Depends on the integration between VMS and camera. If you configure the VMS to use a digitally cropped region, this may be the case.

(1)
(1)
MM
Michael Miller
Jun 14, 2018

Unless you are pulling multiple streams how else would this work?  Also, how would you dewarp the recorded video if your using camera side dewarping? 

U
Undisclosed #2
Jun 18, 2018

You are streaming a full resolution stream to the VMS for recording and one of the digitally cropped, camera side dewarped streams from the camera for live viewing. 

Playback would be client side.

I wish VMS / camera integrations were more transparent about how things are done, but they all have to market it with crap like HDSM, dynamic scaling, etc. and make it cryptic. Are things being transcoded or using multiple streams?

(1)
MM
Michael Miller
Jun 18, 2018

So your saying the camera would send 2 streams to the server.  One full image and the dewarped.   What is dewarping the recorded full image stream? 

Are you asking me if HDSM transcodes video?

 

U
Undisclosed #2
Jun 18, 2018

One stream to the recording server of full resolution warped image. 

Second stream to the client as live view, dewarped from camera. 

Playback would be dewarped at the client from the recordings, which is fine because you aren't typically looking at an array of cameras, less probable all fisheyes, during playback.

 

I am not asking you about HDSM, just using it as an example. HDSM works similar to the way I just described fisheyes, with exception to HDSM recording both profiles. HDSM does not transcode.

Avatar
Ethan Ace
Jun 14, 2018

Client Side

For client side dewarping, there are two key advantages:

  • Using virtual PTZ controls on the client, users can virtually pan/tilt the view to see activity in all areas, on both live and recorded video. This allows a single stream to cover many angles, instead of using multiple camera side dewarp streams to view the same area.
  • This also allows multiple users to view different areas at one time, like Michael said. This isn't typically possible with camera side dewarping, unless multiple dewarp regions are used, which adds to bandwidth and complexity. 
  • This single stream is often lower bandwidth than multiple dewarped streams, with many cameras now offering smart codecs on warped overview streams.

The key disadvantage is integration complexity. Client side dewarping often requires the VMS to integrate a proprietary dewarping SDK for each manufacturer. For smaller camera manufacturers, this may be a challenge as VMSes are less likely to integrate less frequently used models.

Camera Side

For camera side dewarping, pros:

  • Simplicity is the main advantage to camera side dewarping. For some people, they want to view the whole scene, but not train users on how to control the dewarping. So they can use a panorama/dual panorama or dewarped views and the camera appears as any other would, without needing to use virtual PTZ controls.
  • Additionally, viewing fixed camera side dewarp streams makes sure operators are looking at what you plan them to be viewing, instead of potentially missing things because they are panning around in an overview stream. 
  • Finally, camera side dewarping reduces and may eliminate the VMS integration required for client side dewarping (though integrating multiple streams and/or PTZ controls may still be necessary).

The cons of camera side dewarping are:

  • Only what's viewed is recorded, so if a dewarped stream is looking in the opposite direction of activity, nothing wil be captured, whereas users could use virtual PTZ controls on recorded video with client side dewarping.
  • Bandwidth is generally higher, because you are recording multiple dewarped streams instead of a single overview stream. These streams may also not allow smart codecs to be used (as is the case in the Hikvision 63C2 we just posted).

 

 

 

(2)
UM
Undisclosed Manufacturer #3
Jun 26, 2018

You may also want to see if the camera is using Immervision lens with their SDK.  Immervision uses different settings for different panamorphic lenses and the client has to detect which lens is in use to dewarp properly.  This is true of Hanwha Wisenet 5 panamorphs.

(1)
New discussion

Ask questions and get answers to your physical security questions from IPVM team members and fellow subscribers.

Newest discussions