Camera stream/recordings darker than other apps


#1

I have 3 camera in my setup and with each camera, the picture is considerably darker than on the feed from the camera on its own web console or when using a third party app.

For example, I have a DLink 5222L which looks over my back garden. On the camera’s web console, it is able to make out a fence in the dark around 30 meters away. The iOS app called “IP Cam Viewer” is also able to make out the fence 30 meters away in the dark, however, Netcam studio can possibly only see half of that distance.

I have tried tweaking settings in NCS which make no difference. All three camera are the same.

Ill upload some pictures.


#2

From the Cameras Web Console


#3

From “IPCam Viewer” app


#4

From NCS


#5

This is quite an issue as we had our shed broken into last night and the robbers scaled the fence at the bottom of the garden which was obviously not dectected by NCS as it is unable to make it out.

I am quite concerned that I am relying on this for security but it cannot be relied upon when it matters. If I was recording from the cameras on to an ad card into the cameras, based on the difference in the the above picture it would have picked up the motion and recorded


#6

NCS use only the raw signal all the time and do not process it when it comes the brightness etc. So if the pictures above are from the same monitor I have no answer to you for the moment. I am sending this to our developers.
@Steve
-Henrik


#7

When passed further to the clients (web/mobile) then the image is recompressed in JPEG or in MPEG-4, this is the only part which could alter slightly the image or lose a bit of details but it’s anyway not the case with the input / what is being used for the motion detection, only the output may be altered a bit because of this.


#8

Ok, so what do I need to do/change to get the output from NCS to be the same as that of the camera itself? My compression settings are already at 100% quality. See screenshot


#9

Also, you say recordings/motion detection wont be affected as this will be due to compression between NCS and the client device, however, the recordings are just as bad. On the recordings, I am unable to see the fence at the bottom of the garden and if I go to the bottom of the garden and move around, it does not trigger the motion detection.


#10

Hi there!
I have been thinking a little bit more about this and this will actually be quite interesting to sort some thing out and also to detect uninvited objects at your fence 30 meters away. So here it goes:

  1. Here is a flow chart from camera to recorded video on the HD.
    Camera(video signal coded in MJPEG or h.264 transmitted in avi format) – > – NCS(motion detector) ------ avi-format — codec(recoded into mp4) ----- recording stored in mp4 format on HD.
    The recordings are re-compressed into mp4 and not in original avi. Since mp4 takes much less space than the avi something must be lost and it is the details and small differences. The dynamics in the video is lost and in your case the number of levels in the gray area have decreased.

  2. The camera can deliver 2 signals; MJPEG and h.264. MJPEG is a high quality signal which use a lot of bandwidth. h.264 is compressed and therefore details are lost especially if the difference between the pixels are small. So check that all the programs use the same signal, MJPEG or h.264. In this situation I would say that it can make a difference.

  3. To detect uninvited guests at the fence 30 meters away.
    I would say that this is not an easy situation for any system. It is very dark and the intruders have probably black clothes, do not use flashlight and move rather fast. Camera delivers IR to 8 meters. This is nothing for the automatic calibration system in the motion detector. In the configuration of the motion detection the Threshold is by default set to 0 which is automatic. I would say that you can experiment with Motion Algorithm and Threshold. Start with Threshold since that is easy. If you have a tablet open the web client or even better using TeamViewer you can do this in real time when moving around in the dark at the fence.

In the web browser use this

Now you can see in real-time the areas in red where the system detects motion. If you have Teamviewer you can change the Threshold level and see what happens. Also test another Algorithm.

I do not say that this will work since the situation is not easy for any system. What these systems need is light and I would recommend to put a second IR source maybe on the shed that illuminates the fence.

-Henrik


#11

Hi Hendrik,

Thanks for your information. Changing the streams to MJPEG fixed the issue, however, I have also replaced these cameras for better quality ones that are mounted outside for better IR performance.

Thanks,