I have read statements like this a number of times (simplified):

Camera X has a bit rate of 50 MBit and if you record in 60 FPS that is really equivalent to 25 MBit at 30 FPS and that means the quality is bad.

This sounds logical because the argument is that having fewer bytes for each picture results in lower quality and thus the video at 60 FPS is at the same quality as a 25 MBit at 30 FPS. However, this is only true for intra-only codecs such as Prores, DNxHD, AVCI or IMX. Let me explain why.

The key factor that makes compression so much more efficient for inter-frame codecs (also commonly called “long GOP”, although that’s a bit of a misleading name) is that they do something called motion estimation. That means they analyze the content of the input signal and try to represent the motion of certain blocks of pixels between frames by something called vectors and those take up very little space. Then the difference between a picture approximated by moving around those blocks from the source signal along those vectors and the actual image is computed and then compressed as an image. The trick here is, that the better the approximation is, this difference image will not contain a high density of information and will compress really well.

The slower things are moving in the picture, the easier it is to estimate the motion and thus compress well. So by filming the same subject in 60 FPS instead of 30 FPS you are effectively cutting the speed of motion in half. The higher the frame rate is, the better the motion estimation will work for a given motive and then the formula used above no longer works.

So what is the right number? The answer is, you cannot say, because this whole effect depends on a number of pretty complex factors. One is the actual codec implementation, e.g. one h.264 encoder is not the same at all as another h.264 encoder as far as quality at the same bitrate is concerned. Making encoders is almost an art form.  Another thing is that it highly depends on the motive (for each individual codec implementation!). All that means that you can not solve this using a formula. The sad truth is, you have to test.

The bottom line is, don’t trust anyone making this statement to dismiss a camera until you’ve seen actual results.

Hope this helps