I keep coming back to physics first principles:
Google: Dimensions of Model S:
196"L x 77"W = 15,092 sq in = 97,367 cm^2 = 9.73 m^2
Solar insolation
Average Insolation in Los Angeles in the year is 5.4kWh/m^2/day.
5.4kWh/m^2/day * 9.73m^2 = 52.58kWh/day.
Which Solar Panels Are Most Efficient?
2.5 years ago: maximum efficiency of panels was at 36.7%
in the lab.
Google: Solar Concentrator
"A
solar concentrator uses lenses, called Fresnel lenses, which take a large area of sunlight and direct it towards a specific spot by bending the rays of light and focusing them. Some people use the same principle when they use a magnifying lens to focus the Sun's rays on a pile of kindling or paper to start fires."
Display Enhancement Solutions - Protection and Light Control Films: Display Enhancement Films: 3M United States
This is what Tesla Roof's 3M engineers already are able to do.
Holographic optical element - Wikipedia
I heard of these abilities in 1984. By now, they must be good at this. Similar ideas as previous link above.
This fully transparent solar cell could make every window and screen a power source (updated) - ExtremeTech
some other examples of glass being used for solar collection ideas
So, the fact that max efficient panels were at 36.7% a couple of years ago only in lab settings tells me that this is something that will not yet have been commercialized, but that the sets of capabilities I referred to regarding lensing, directing of solar energy into collectors, etc., will allow this both on opaque and on transparent surfaces.
So, let's assume that the target is to bring what was possible in the lab a couple of years ago on opaque surfaces to reality in production on mixed opaque and transparent surfaces within a decade, which is a grandiose assumption but allows me to use simpler math to boot, which is that covering the Model S's exterior surfaces 100% with these materials (all the way around, on top, all glass, all body panels, with exception of wheels and some crevices and equipment where it's just not cost effective, but that is in doubt too, since even cameras can be covered with transparent glass) would yield something in the ballpark of 36.7% efficiency
somehow.
52.58kWh/day * 36.7% = 19.3kWh/day
Charging a Tesla Model S Might Be Costing More Than You Think
Charging efficiency of Model S is 81.5% (using current standard installations).
Charging efficiency of DC-DC power electronics is probably as good as AC-DC.
19.3kWh/day * 81.5% = 15.72kWh/day.
Lifetime Average Wh/mi
avg 363Wh/mile
15.72kWh/day/363Wh/miles=
43.32miles/day
So, you drive your car to work, leave your car parked out in the sun while at work, then drive your car home. If you have a 60kWh Model S, then one quarter of your battery can be charged per day this way. If your commute to work is 15 miles and back from work is 15 miles, that's 30 miles/day. Let's assume you also go on errands, but not every day. So, on the days you don't go on errands, your 60kWh car battery will store some excess energy from the sun that you can use on other days for longer errands travel or for less sunny days.
I imagine having solar on cars as a way to increase general energy flexibility in transportation, and as an offset to energy use and land use. But, in the above exact calculation, that person with only a 15 mile commute to work would actually no longer need any additional energy for their commute during summer. Whenever they plugged in at home, it would unnecessarily use shore power (home solar, utility power) to give insurance that they will have more SOC after work or to make up for extra trips; they certainly don't need it to get to work or get home in regular commute. During winter, the driver would have to plug in more often.
When do I think this will happen? In the poll, I said 6 years, but that's optimistic. I actually think it's more like 3 - 15 years of constant R&D. Whenever R&D is put on pause, add that pause time to that estimate.
When do I think the costs will drop to reasonable? After all the above has been completed and more time passes during which more R&D continues, something that could be done by selling less cost efficient systems to the types of people who used to buy Roadster, Model S and Model X. So, I'd say about another 3 - 15 years from above, so about 6 - 30 years from today (plus pause time) before this is cost reasonable. That's within our lifetimes, if pause time is assumed to be low (and, at the extreme optimism end of the scale, is before Trump is out of office). Put another way, if no pauses are taken, by the time we have a colony on Mars.