People sometimes ask and vendors sometimes list a 'frame size' for an IP camera. For example, the frame size is 20KB. So if you have 30fps, then the bitrate would be 4.8Mb/s (20KB x 30fps x 8 bits / byte).
Frame sizes for modern H.264 are highly misleading and I recommend not to use them.
Why?
The frame size is a function of a number of different factors (even beside the obvious one of resolution / pixel count). They include:
- Scene complexity - a lit stairwell is going to have a much higher 'frame size' than an intersection
- Quantization / compression level - depending on the compression chosen, the frame size can vary significantly. Indeed, manufacturer default level vary widely (see: IP Camera Manufacturer Compression Comparison).
- Number of frames per second / I frame interval - even if the scene and compression levels are the same, simply dropping the frame will increase the 'frame size' as inter-frame compression savings are limited at lower frame rates. For example, a camera might be 3Mb/s at 30fps (a frame size of 100Kb) but be 300Kb/s at 1fps (a frame size of 300Kb). See Testing Bandwidth vs Frame Rate.
The only application where frame size would be fixed is if one used constant bit rate, which is a bad idea and should be avoided (see:CBR vs VBR).