`>>Is it possible to use an NIR sensitive camera to take the image and then use the red and the NIR as two separate copies of the same image and merge them instead?
This is only possible if one of the channels (RGB) receives no visible light and therefore records only NIR light. So a filter must block all light of one color and transmit NIR. The recommended filter is one that blocks all blue so the blue channel captures only NIR light.
This solution has a serious flaw because the other two channels will also record NIR light along with the visible light that is normally recorded. It is not straightforward to determine what proportion of the light recorded in those channels is visible vs. NIR.
This problem of contamination is avoided by using two cameras. It is easy to place a filter in front of an NIR sensitive camera so only NIR light is captured (in all three channels). A separate normal camera can capture a normal photo in which the red channel will record primarily red light. Using two cameras requires registering (aligning) two photos, but this can be done very effectively with free software. The results can be excellent if parallax is minimized by keeping the two lenses close together and far from the subject (or moving the lenses so they alternately occupy the same position).
With either single or dual-camera systems, calibration is required to convert the brightness values (0-255) into an estimate of radiance (energy). Brightness values recorded by cameras differ from radiance because cameras are not radiometers and also because camera sensors are not equally sensitive to visible and NIR light.