Hikvision - Noise/Quality Problem - $50 Reward!

I have my cabling guy coming in the morning to finish putting all the cameras up. Once they are all up I will post a bunch of videos to see what you guys think. He is going to check all the cables and troubleshoot with me.
 
I should say something about Nelly's Security. This is one of the best companies I have ever dealt with. They will do anything to make sure their customers are happy and I could not be more pleased with how they have treated me and the support they have provided.
 
NoloC said:
Well one more "opinion" can't hurt so I'll chime in. I think you're on to something there Aster in that the 12vdc and the 48vdc would need to be handled differently as you point out. Maybe someone on here knows what is done? The 48 might go through a dc to dc conversion like an inverter?

One note on your comment, oscilloscopes don't measure current or microamps. They display voltage on the Y-Axis vs time on the X-Axis. Voltage not current. Would you agree?

But it is the best theory yet!
Click to expand...

Of course the oscilloscope does not measure or display current, every electronics engineer knows that. However if you pass the supply through a very small resistor and measure the voltage drop on that resistor, then you can measure AND display the current drawn from the camera. HIK cameras use PoE Mode B with +48-56V at pins 4 and 5 of the Ethernet plug and the -V at the pins 7 and 8. The communication is also at 100Mbps level. The 12V supply from the pig tail does not go through any special conditioning electronics. It is the cleaner power supply compared with the PoE method. That's why some people do not see the pixels pulsating when the camera is fed with 12V.

On the other hand, the PoE supply goes through the isolation transformers of the Ethernet intrenal in the camera. Since these transformers are very small in dimensions, they have very thin coil windings that create voltage drop when current is drawn from the camera. This voltage drop has a form of spikes which must be filtered in order not to reach the rest of the electronics. This can be done with a very sophisticated electronic power conditioning circuit that adds cost to the camera. The alterative solution would be to have a better quality ethernet isolation transformer. That would also add cost to the camera. Another solution would software image proccesing. Even that has its own cost an complexity and in fact it tries to cover the problem, not solve it. So both proper solutions add cost.

Remember that in order of higest cost the most expensive items in electronics are power supplies, transformers, and capacitors (that age quickly).

In a nutshell, quality costs and quality can not always be defined in high level protocol specifications.
 
Please lower the bitrate and see if more pulsating pixels arrive. Tell me the response.
I am 99% sure that the behaviour you see is not pulsating. Its just encoder placing I frame over P frames - this is normal in H264 because this is how H264 works.

Stream is divided into I frames and P frames (and sometimes Bframes also). Structure is as this: I P P P P P I P P P P I P P P P

In simple explanation (really) - I frame is IDR frame (IDR - Instantenous Decoder Refresh). I frame is a complete ONE frame like JPEG image.
in 25 fps and keyframe interval set to 25 you will recive one I frame every 24 FPS (24+1) - This 24 fps are just DIFFERENCES between movement in time.
After 24 FPS when next I Frame arrives - it will clear the buffer of decoder and place on complete frame in front of you and again over this IFrame - place next comming P frames (differences). This is the moment you see pulsating. Because the encoded compelte frame differs than the already buffered video (it has to, it's nature of lossy compression).

Now imagine you set bitrate of 2Mbps on FULL HD resoltiuion. Now get a snapshot of one frame. Place it on you image editing software and lower the quantization level to acquire 256 kb image size (2Mpbs = 256 kbytes of data). You will se quality degrading. and know that this 256 kb of data has to fit in one second of video. So 256 kbytes will have on I FRAME (one JPEG image) and 24 P frames (differences). It has to fit somehow. So encoder degrades quality.

Normally you wont see this hapening (high bitrate and good light conditions). But when sensor is small and produces a lot of noise (yes, noise like in simple DSC) you will see pixels pulsating. Its just noise on sensor which encoder sees as a object movent in time. If so, then it lowers IFRAME quality to fit also P frames in this second of stream.

Thats why bigger sensor means better light gathering and better quality (pixels on sensors are bigger so can catch more light). But bigger sensors i mean sensor size like 1/2.8 inch or 1/2.5 inch. Anything like >2Mpix and sensor below 1/2.8 (1/3 etc) is a complete useless. Because it will only work in good light and very good light. Forget about low light in such small sensors. Thats why buying 4Mpix Hik with sensor size 1/3 inch is not good idea.

Check Sony VB635 which has only 2Mpix but sensor size 1/2 inch in low light - you will se the difference.

Also thats why cheap cams implement tons of functions called DNR/3DNR/2DNR and other shit - it's all about blurring those noise from crappy sensor in low light.
 
Last edited by a moderator:
  • Like
Reactions: aster1x, jordanb and Q™
Speed666 said:
Please lower the bitrate and see if more pulsating pixels arrive. Tell me the response.
I am 99% sure that the behaviour you see is not pulsating. Its just encoder placing I frame over P frames - this is normal in H264 because this is how H264 works.

Stream is divided into I frames and P frames (and sometimes Bframes also). Structure is as this: I P P P P P I P P P P I P P P P

In simple explanation (really) - I frame is IDR frame (IDR - Instantenous Decoder Refresh). I frame is a complete ONE frame like JPEG image.
in 25 fps and keyframe interval set to 25 you will recive one I frame every 24 FPS (24+1) - This 24 fps are just DIFFERENCES between movement in time.
After 24 FPS when next I Frame arrives - it will clear the buffer of decoder and place on complete frame in front of you and again over this IFrame - place next comming P frames (differences). This is the moment you see pulsating. Because the encoded compelte frame differs than the already buffered video (it has to, it's nature of lossy compression).

Now imagine you set bitrate of 2Mbps on FULL HD resoltiuion. Now get a snapshot of one frame. Place it on you image editing software and lower the quantization level to acquire 256 kb image size (2Mpbs = 256 kbytes of data). You will se quality degrading. and know that this 256 kb of data has to fit in one second of video. So 256 kbytes will have on I FRAME (one JPEG image) and 24 P frames (differences). It has to fit somehow. So encoder degrades quality.

Normally you wont see this hapening (high bitrate and good light conditions). But when sensor is small and produces a lot of noise (yes, noise like in simple DSC) you will see pixels pulsating. Its just noise on sensor which encoder sees as a object movent in time. If so, then it lowers IFRAME quality to fit also P frames in this second of stream.

Thats why bigger sensor means better light gathering and better quality (pixels on sensors are bigger so can catch more light). But bigger sensors i mean sensor size like 1/2.8 inch or 1/2.5 inch. Anything like >2Mpix and sensor below 1/2.8 (1/3 etc) is a complete useless. Because it will only work in good light and very good light. Forget about low light in such small sensors. Thats why buying 4Mpix Hik with sensor size 1/3 inch is not good idea.

Check Sony VB635 which has only 2Mpix but sensor size 1/2 inch in low light - you will se the difference.

Also thats why cheap cams implement tons of functions called DNR/3DNR/2DNR and other shit - it's all about blurring those noise from crappy sensor in low light.
Click to expand...

I suggested and explained this to the OP a few pages ago. But he said it made no difference whatsoever and he has already solved the problem. Thanks for the detailed explanation though.
 
aster1x said:
Of course the oscilloscope does not measure or display current, every electronics engineer knows that. However if you pass the supply through a very small resistor and measure the voltage drop on that resistor, then you can measure AND display the current drawn from the camera.
Click to expand...

Weak.
 
Update: All the cameras were finished being installed all day yesterday into the night. I had the cabling company and my alarm company who both do camera installs here and they both agreed that something is wrong, but the install took a long time and they really couldn't troubleshoot (13 cameras). Today I am fixing camera angles etc. I can tell you all the cameras are doing the same thing, some worse than others. I have a few cameras that are even doing a transparent effect on people ("The Predator Effect"), you can only see their outline. I bought a new UPS as Fenderman suggested and I have it charging now to test later in the day. I will post another update later today.
 
I think the ghosting effect is caused by using too high noise reduction and/or WDR at night.
 
spixel said:
Which cam?
Click to expand...

It was happening on the Hikvision bullets. I did a reset to default and that fixed that probelm. The UPS is taking a long time to charge so I wont be able to test it until tomorrow. Also, the Dahua PTZs are a nightmare on the Hikvision NVR. The PTZ controls really don't work so the cameras are facing nothing (wall/ground) at the moment. I had to set the ip manually for them to work with the NVR but now I can't access them (192.168.254.x) via the browser port for PTZ functionality. Seems like I will need to reset them tomorrow by taking off the dome. Has anyone tried running Dahua cameras into a Hikvision NVR and if so did you run them using the NVR PoE like I did or did you use a PoE injector and then add them to the NVR separately?
 
Its been several days. I hope you all had a happy Thanksgiving. Here are some things I have done.

1. Replaced NVR - no difference
2. Tried plugging the NVR into a UPS - no difference

The two Dahua PTZ Domes I plugged into POE injectors and they no longer have the problem, but all the Hikvision cameras still have the problem. There is something else that is happening at night that maybe you guys can help me with. I have some cameras in Day Mode at night because they have tons of light and it looks better in that mode, others I have in Auto mode. All the bullet Hikvision cameras are flickering the IR on and off every so often. It seems to me like it is changing between modes or something like that. Im not sure why the day cameras would be doing that if they are set to day. Anyone know how to fix this? I have the IR sensitivity set to 0 on the Auto Mode cameras and IR filter I have raised up to 50. This is only happening on the Hikvision Bullets, the Dahua Domes and Hikvision Dome does not do this.

Also 2 of the Bullet cameras are stuck in 50hz, each time I change them to 60 it reboots and it stays at 50.
 
Last edited by a moderator:
DaveFL said:
Its been several days. I hope you all had a happy Thanksgiving. Here are some things I have done.

1. Replaced NVR - no difference
2. Tried plugging the NVR into a UPS - no difference

The two Dahua PTZ Domes I plugged into POE injectors and they no longer have the problem, but all the Hikvision cameras still have the problem. There is something else that is happening at night that maybe you guys can help me with. I have some cameras in Day Mode at night because they have tons of light and it looks better in that mode, others I have in Auto mode. All the bullet Hikvision cameras are flickering the IR on and off every so often. It seems to me like it is changing between modes or something like that. Im not sure why the day cameras would be doing that if they are set to day. Anyone know how to fix this? I have the IR sensitivity set to 0 on the Auto Mode cameras and IR filter I have raised up to 50. This is only happening on the Hikvision Bullets, the Dahua Domes and Hikvision Dome does not do this.

Also 2 of the Bullet cameras are stuck in 50hz, each time I change them to 60 it reboots and it stays at 50.
Click to expand...

Have you tried updating the hikvisions firmware?
 
Uh. Oh.
 
You must log in or register to reply here.
Top Bottom