The light source is a pulsed array of Red Green and Blue LEDs, it can pulse incredibly quickly so acts like a 'flash' or strobe and will freeze any film movement. This allows for a continuous speed scan, but the rig also has the ability to stop the frame in place, pulse each LED array in turn to a 2K mono sensor. This gives you 3 images, one red, one blue, one green.

If you use a colour sensor, your colour resolution is reduced by the bayer mask, and you get 50% of the resolution in the green channel and only 25% each for Blue and Green. Some of this is recovered via clever algorithms, but you never get the full resolution that you can with using a mono sensor. Each channel is captured in either 12bit or 16bit per pixel, and a LUT is applied to ensure the full dynamic range of the film is captured, as the toe and shoulder of the film needs more 'steps' to be captured fully.

The 3 mono images are then combined into a single colour image.

For important scans each channel is captured 3 times and averaged to remove sensor noise, so there are actually 9 images per captured frame.

If speed is of the essence, then a 2K colour sensor is used and the film is run at 24fps and captured in real time at 48bit colour. This still gives an astoundingly good result, and is a good choice if the film is already graded and you just want an accurate transfer. The mono setup is really for when you want the headroom to do corrections in post.