Thanks for the discussion on frame rate. I tend to set ours pretty low at 10 fps, as I would rather use the bandwidth to get higher quality images than full motion video. I have a hard time imagining a security event that couldnt be captured with 10 frames in a second, but I sure have seen many where poor resolution rendered the video event useless.
I agree. At my last job after years of using 6 FPS, a couple yeras ago we upped it to a whopping 8FPS and it still worked fine (excepting special applications).
These days though I think more integrators are forced to use 15 FPS or higher because as one commenter alluded too, the customer probably expects it or thinks something is wrong or not configured right.
When I started we discussed “seconds per frame” when dealing with tapes and multiplexers.
I used to demo high action cartoons recorded at 15 and 30. It was almost impossible to see the difference in mpeg4.
Gaming would be a slight of hand exception.
I would also tell people when bandwidth and storage costs were more of a consideration to identify “how many frames will the person be in the scene, how many shots at capturing usable info do you want?”
Lets take a parking lot, how many seconds to cross the scene? 10? Indeed, 5 fps will give you 50 images.
Lets take a hallway from one side to another? 1 second? To get the same amount of opportunity you would need almost 50 fps. While 10 would work, 15 and maybe 30 would seem practical.
Using h.265+ (proprietary codec) it seems there isn’t a lot of savings between 15 and 30, but noticeable between 10 and 15 from my experience in a low and medium activity scene.
Of course, the dynamics of the scene including lighting are also a part of the decision process.
Unless....you have techs that just turn on the recorder, set a password and walk away but that never happens.
perhaps you missed the word static in my description.
in a static scene I am assuming that there are 0 bits in P frames (as no pixels are changing) while key frames would contain all the bits. so multiplying key frames by 30 would then increase storage by ~30x.
I definitely missed that, however a scene will never be 100% static with zero P frames. The encoder will always find some noise, etc. that it needs to transmit. The encoder still sends P frames, even if they have a lot of motion vectors or references to the initial I frame.
We have started to require 15fps on all recorded video with interior cameras having dynamic frame rate enabled and exterior cameras having it disabled. We require dynamic compression and dynamic GoP enabled on everything. We have found exterior cameras with their long FoV do not adapt the frame rate quickly enough or at all for objects in the far part of the FoV. We also require recording be 100% of the time and do not place bandwidth caps. We have found that this combination maintains a high level of video quality with a relatively low impact on storage due to how efficient smart codecs are.
Here is a screenshot showing that exact configuration with nearly 200 cameras using H.264. They delete video at 31 days but there is old bookmarked video which is why the archiving span is so long. They essentially use 32TB to maintain video on 200 cameras.
What kind of bandwidth average are you getting with those settings? I see the network traffic in at 46.7 Mbit/s but thinking that is at the moment the screenshot was grabbed and not a daily/hourly average?
The true 24-hour bitrate average is about 25 kb/s just dividing between all the cameras. Between the building closing for night and opening in the morning, there is over 12 hours where they use 1 kb/s. I would say 85% of the cameras are 2mp and the rest are 5mp. During high traffic times I see them hit upwards of 20 mb/s for short durations and the exterior cameras get close to 30 mb/s when we get heavy snow. Because there is no cap we let them hit what they need to maintain quality video.