Correct
@Jjrss, thanks.
And
@Mrbrock generally the ability to retrofit is not a binary possible/impossible issue but a matter of cost. I just got done running a new display cable from the computer to the steering wheel. It wasn't the simplest job many because of access to the connector and the DIY circumstances, but it was orders of magnitude below impossible.
The biggest issue in retrofitting new cameras would be if they were to be mounted through body openings that presently don't exist. That is a big reason why the headlights idea is good IMO. Aside from being in the right place for the intended function, it solves the problems of retrofitting to existing body dedign and physical wiring access to the interior.
I have already proposed, in other threads, various possibilities of video merging using preprocessor modules, to address the question of adding or upgrading camera feeds within the current HW3 suite, and also and harness routing in case that would be pursued. This us to deal with the main issue that the HW3 FSD computer has only the 8 canera ports and it's very unlikely those are equipped for demultiplexing multiple HD video feeds.
I'm not aware of the specific limitations of the present hardware platform regarding communications bus bandwidth etc. Future platforms are likely to have very wideband buses; newer generations of CANbus are very capable and some cars are doing 100G ethernet etc.
In saying all this, I'm certainly not indicating that I believe Tesla will do it or that it's a piece of cake. But to say it's impossible is clearly incorrect. In this circumstance, the "Not happening" prediction is the safest but not the most imaginative. I've known many people who live on pessimistic naysaying, and indeed they are right at least 80% percent of the time. I care about what could be done given the will, not about Having Been Right when we check the internet comment history next year.