Are there any realistic stats/charts on what kind of bandwidth is required to do what kind of video/stills over the most popular RF, even the ESP8266) chips? What about LoRa?
In addition, I have yet to come across code/projects that create a wireless camera & local receiver. Can I take regular NTSC video and encode it to serial? I have a pile of all the various RF modules and a bunch of EPS8266 variations, but am holding back on buying a camera (or using one of my 'normal' cameras) because I don't see the subject matter on it online.
My use case is for a friend's long driveway, where she can't even see her mailbox. She knows someone is going through it, but can't get any evidence of it. Sounds like a great real life excuse to build something! 
So I envision something along the lines of a cadmium sulfide/light sensor in the mailbox that triggers a nearby camera to either start video or taking stills. That camera then stores to SD, and periodically transmits back to the house. Distance is about 1,000', and from the telephone pole (on property) I'd mount a camera xcvr to, it is line of sight. I'll worry about a timedate stamp on the imagery later.
I assume that it isn't possible to do HD (1920x1080) in real time, say, recording 10fps for 1 minute, with serial, but is that realistic?
Context:
-There is no AC power at the mailbox, and the pole there (first of ones that take AC to the house) is set back about 20' from the mailbox-perfect.
-I've been looking for an excuse to try solar powering the remote camera and trigger, with a charging board, LiPo, ultra-aps, etc.
-I have WiFi cameras and traditional 'video' cameras to use. Simply keeping a WiFi camera on 24/7 would never work for power reasons (I assume).
-However, I COULD use one of the outdoor WiFi cameras with a directional/yagi antenna back to the house (I have a few), and maybe it could do the range if coupled with another yagi at the house...
-The trick would be providing enough remote power to let it run without swapping batteries. I'd have to trigger, time, and cutoff power to the camera/transceiver... If I go this route (plan b?), I still need to see i bootup (first-pic) time is before someone drives off.
-In a pinch, I could run wires from the mailbox to the nearby pole, where the light sensor could be wired in instead of making an additional wireless link...
So this is my project, and my random musings on it... She's recently divorced, and the ex- is making her life hell. I'm that geek with lots of parts looking for a practical use for it all, and I'm fascinated with the idea of self-sustained (powered) remote sensors.
Thanks in advance for your thoughts on approaching such a thing...