
I was speculating, we are all here together trying to answer OP’s question. As far as I remember, interlacing H.264 used to lead to weird really long processing times. Encoding took a long while as a tradeoff for way faster decoding. Which resulted on the speed of data writes being violently throttled when image processing took too long, not a problem when encoding disk to disk, like burning a video into Blu-Ray for example. But it was a problem when recording from a camera, which risked running out of working memory while frames were still coming in hot from the CMOS. So, engineers resorted to compressing on the storage rather than on RAM, as to free up processing resources, this was called real-time compressing (I know it is a counter intuitive name but software devs aren’t the best at naming things). They wrote pre-compression data directly to storage first, this freed up resources on the processor and RAM side so frames weren’t dropped. Then later, the video would be slowly compressed and the result written, again, in-situ on the final file format. In exchange, this process locked the video processing to the formatting of the physical storage media (stuff like format, block size, addresses and caches).
I have no basis to affirm whether this is the case on this specific camera. Maybe it is or maybe it isn’t. My best guess is that, either way, Sony engineers just said “fuck it, force the user to reformat, no one is ever gonna change that one setting other than the first time they use the camera”, and just moved on.
Certainly for more modern cameras—with faster processors, better compression algorithms like HEVC, variable rolling shutter speeds with anti-flicker (even global shutters which are all the rage now) and free from the archaic compatibility issues of NTSC/PAL video signal processors—this is no longer a problem.







That would make sense. If the image processor is manhandling the storage driver while recording video, and also only supports or expects FAT32. Then forcing an SD card reformat every time the camera initializes is a heavy handed but effective way to ensure video is not lost.