What is the optimal bit rate?
To decide what the flow can be, it is necessary to understand how it works, at least in the first approximation.
In the H.264 codec (MPEG-4 Part 10), the construction of the video image is as follows: the camera creates a reference frame (I-frame) and, based on it (therefore, it is called a frame of reference), subtract the still parts of the picture from the frame – a P- frame. The third of this second frame is then subtracted and a modified P-frame is also created. Thus, a series of P-frames is created, which only carry changes between two adjacent frames. Since in the process of “subtracting” frames, errors may occur leading to artifacts, then after a certain number of frames the scheme is repeated, a reference frame is sent again, and then a series is sent of frames with changes across the network. The resulting keyframes and modified P-frames are sent over the LAN to recreate a series of full frames on the display device. It does this by “appending” or “superimposing” P-frames to a reference frame and sending the resulting series of frames to the monitor. This is how codecs work with interframe compression. Of course, in practice everything is more complicated by an order of magnitude (there are also B-frames, which, based on the analysis of a large set of frames, predict where a particular object will move in the field of view of the camera, which allows to significantly reduce the flow when observing the translational movement, for example, that -to the mechanisms).
With a slight change in the observed scene, the changes between adjacent frames will also be minimal, respectively, and the traffic on the network will be minimal. The reverse is also true: if the scene changes significantly, the changes between frames will be significant, consequently affecting the network load.
Do not forget about the peculiarities of codecs with interframe compression, they introduce quite a noticeable delay between the moment of the event itself and the moment it is displayed on the screen. This is due to the work of the codec itself, as the camera needs to receive both frames, then subtract one from the other, send it to the image output device, and overlay it on the reference frame, which takes time and therefore introduces a delay.
There is no point in talking about optimal resolutions without referring to the observed scene, since, for example, when observing an empty corridor at a quality of 1080p / 25 fps, the transmission can be approximately 1 Mbps, and when observing a station of meter at rush hour, an image in the 720p / 25 fps format may take up to 20-30 Mbps
