When you get more time, make some notes on distance and speed and the amount of off-sync you use for the trailing view. I plan on getting a drone at some point and want to try this out. I think 2-4 frames for 5-10 mph is about right on paper from what I've read, yeah the wind and if it's a smaller drone it will be harder to keep it steady. I'll have a look here in a bit on your test.
While I can use the mph it is a bit misleading since air speed and ground speed gets confusing when trying to interpret for 3D. So, the way I do it is move the drone laterally for a ground distance and time it in post. The speed will vary but when you have the time for say a 50 ft travel you know your ft per second ground speed. Then you can count your frame rate and know how many frames you shot. With all this I determined that my distance spread from one frame to the next was 1.33 inches of travel. This is 33.8 MM per frame. We know that good all around 3D for a 28mm lens is 65mm. So this would equate to 2 frames of slip between the two clips before pairing. Considering I was shooting a near object of 50ft distance and my far object down the street of 300 ft. The 3D calculator tells me my inter axial should be 100mm with a 28mm lens. So I set the spread between the two clips at 3 frames. You can download these 3D calculators in the app store. If you work with twin cameras and a wide variety of scenery, it comes in handy to calculate the IA for the lens.
Tom you must have superb vision for compression. The dji Mavic Pro was set for 4K 30p at 60Mbs. But as we've been discussing, I can't get 4K to pair. So I recompressed and resized the clip to 2K 30 fps at 25Mbs. Then rendered the clip as a 3D top/botton for 25Mbs compression. Until I figure out how to duplicate your method of pairing 4K30P I plan top reset the drone for 2K recording to eliminate the down conversion in post. But the way these cameras work, that means I will be using 1/4 the image sensor for the base file. Only a test run will determine which is better in the final output.
There is a bit of jerkiness in the lateral travel dolly of the drone. This was caused by a momentary gust of wind that caused the drone to not travel at the steady speed. That can be a problem I actually did 6 runs and none of them were perfect gust free. I'm hoping shooting at high frame rates say 120 fps and using slomo will resolve this for my desert scapes and mountains. Wind is always a problem anywhere.
BTW- unrelated to 3D drone shooting but I have made quite a number of modifications to my remote controller. Stock has lots of problems with sunlight and ability to see, plus the iphone 6+ battery gets used up real fast. I switched to an ipad air 2 with a sun shade and custom mount. A little more time to assemble but now I can see it in bright sunlight and the ipad battery lasts as long as the RC controller battery. My goal is to be able to get an hour of flight recording on my kit before having to recharge everything.
you won't get enough movement to offset the other view unless you're moving pretty fast, but so in this test I think it works fine, but if there's movement on the ground it might look weird.
Actually, the experiment I need to do is spin the drone about a center axis as opposed to a linear travel. This may work better for wide angle too. But it needs to be tested. I have done this with My NEX5n for 3D panorama and it works.
My 3D videos and more
Don Landis HT System: Projector Sony VPL VW665ES Players: Samsung UBD K8500 OPPO BD93 Sony BDP S6200 All Regions Player Denon AVR 4311ci, 7.1 JBL Professional series and Klipsch PS3, XBOX360, Dish VIP722K; 3D Edit Suite: Edius7.53, Vegas Pro v13, Power Director15, i7-950, LG 3D TV DM2752