Did some experimenting over the weekend. For those of you who guessed the auto-dim sensor was IR activated, you were correct. In separate trials I applied predominantly white light, UV light, and IR light...IR had the strongest effect by far. As stated previously, this would explain why some tints of the same visible light transmission rating can exacerbate the problem more than others, as IR transmission rates surely vary widely among different films. Comparing the two photos you can see just how bright the backlighting can get in the day time when the sensor is saturated with IR.
Driving the IR circuit with stepped PWM, I was able to achieve 12 distinct levels of instrument cluster brightness. The brightest setting being significantly brighter than I've ever seen in this car during the day time, even on a bright sunny day
before my windows were tinted.
The next step is miniaturizing everything and designing an unobtrusive case for it, then optimizing it to run off a small LiPo pack.
(And yes, I am fully aware how utterly ridiculous it is that I am resorting to this on a brand-new $50k car. But based on the feedback from my dealer so far, I have little hope this is going to be fixed via coding any time soon.)
