Good presentation. But he is a little mixed up on the term CMOS. Yes, all the other sensors being produced now except for a few remaining CCD specialty designs, are made with CMOS transistors, CMOS being the construction process for the transistors, and they are being called CMOS sensors. But the Global Shutter can also be made with the same CMOS transistors, but with additional circuitry around them to capture the charge on each pixel simultaneously and to hold that charge value until the camera can grab that information.
Yes, the global sensor captures all pixels in parallel. But the camera can't read all that data in parallel, so it reads it in successive reads until all the data has been presented to the camera. This happens very rapidly, but this part is not parallel. And this can still be designed with CMOS transistors. And it has to happen very quickly for the camera to capture at 120 fps. In fact, at 120 fps, the sensor has to be ready to capture the next image in 8.33 ms. So, capture in parallel, and hand off all that data within 8.33 ms which has to also include the time to reset the sensor to make it ready for another capture. And also, to be included is the fact that the camera has to be able to process all that data to be ready for the next capture. Of course, they get around this by buffering up the information where a buffer is filled and once full, it can't collect any more until space is made available. Which makes me wonder what fps rate can the camera run at that the buffer empties at the same speed it is filling with so that the buffer is never filled to capacity?
I personally have no pressing need for a camera this fast. But I do appreciate what a game changer this is, and read and watch everything that is made available for this new camera. Eventually, we will get the actual spec's, and down the road farther, the costs will come down, and the design perfected to correct the current deficiencies. But it sure is fun to see the progress.
A parallel story to this is non-volatile memory. I remember in the early 80's, we used EPROMs and I remember they came in 2K x 8, and then 4K x 8, and bigger and bigger as time went on. They had that little window on the top of the package to erase them under UV light! Then came Flash memory, which at the time we used EPROM, it was thought that maybe Flash memory was too difficult to manufacture! But they figured it out, and it went through the same growing process. I was involved with both NOR and NAND type memories. NOR was smaller, but could be randomly accessed and was popular for BIOS memory in PCs. NAND was read a page at a time, first ones were only 512 bytes, and was popular in most other designs. The first ones were slow and didn't hold much data. But they got much faster and got much bigger as time went on. Having been retired for 7 years now, I can only guess what they are now! The growth of memory is one of the developments that unlocked the growth of everything that used the memory.