that looks promising. From the data sheet: “this product is designed for use in consumer use camcorder”.
So it seems that this is the first Raspberry Pi camera chip without a CRA-compensation by microlenses. That would be fantastic, as the CRA-compensation (a usual technique for short fixed focal length setups like mobile phones) always led to color drifts and desaturation issues if you replace the standard lens. These issues are impossible to compensate fully in post production. Nevertheless, quite a few people based their movie scanner on Raspberry Pi cameras and got resonable results (I got only acceptable results with the v1-camera, which has a lesser CRA-compensation than the v2-camera.)
If you match one of the v1- or v2-cameras with a Schneider Componon-S 50 mm, with the lens placed about 100 mm from the chip and 100 mm from the movie frame, you will image the usual Super-8 format perfectly, even with a little headroom for sprocket detection and the like. This chip has double the size of the v2-camera, so the optical setup (distance chip-lens + distance lens-frame) will need to be adapted. It would actually shift the Componon-S slightly out of the working range this lens was designed for. Should still work however. Otherwise, the lens could always be reverse mounted to solve this issue.
One will have to be see how the software support of the new v3-camera will work. I guess only time will tell. One standard approach of utilizing these Raspberrry Pi cameras, the picamera-library (Python-based), hasn’t seen much update activity recently. Using the standard software supplied with the Raspberrry Pi (raspistill or -video) is way too lame for most film scanning applications.
From my experience with the v1- and v2-cameras, the Raspberry Pi processing pipeline does employ a lot of automatic processing. It is not easy to circumvent these automatic processing steps - and they can easily introduce flicker and the like in your scanning results if not switched off. I never succeeded in turn off all processing deemed necessary by the Raspberry Pi foundation or the chip manufacturer for proper use. For example, the v1- and v2-cameras employ noise reduction algorithms which actually work not only on the digital noise of the image sensor, but also on the noise introduced by the film grain. Well, depending on your taste, that might even be considered a valuable feature. However, a chip designed for “consumer use camcorders” is certainly a different beast than a machine vision camera and might feature some processing steps which you do not want to have in a scanner application. I will certainly have a look at the new camera chip.