Here's some examples of sweet videos filmed by the new wave of Vertical Studio users. Most of you are VS veterans 
but here's something to enjoy:
but here's something to enjoy:
Vertical Studio - Looking for Beta Testers
- Thread starter Neil Mathew
- Start date
- Joined
- Jan 8, 2016
- Messages
- 1,162
- Reaction score
- 291
- Age
- 60
I'm honestly not impressed by those videos as they really only show slow moving objects. I keep monitoring this thread but I'm probably not interested in buying until the app can auto track the target from a configurable distance and height. These features with Walls would really make this app stand out. Any plans for these asks?
Does it automatically lead your dog and stay a certain number of feet and height away?Do you think my French bull dog is fast & close enough?
Haaah! This is my ultimate favorite video so far.Do you think my French bull dog is fast & close enough?
There is no auto-follow on Vertical Studio. The goal is to allow creative control to fly your drone while the camera control is managed. We're releasing a new feature that lets you draw a flight path for the drone and then let the drone fly along that path while tracking something. It's not auto follow but could be used like that.
I think that there is some misunderstanding on your idea of what the app does automatically and what it assists you to do. Once locked onto a subject, the app controls the yaw and the gimbal angle to keep the object centered (or off-centered if you want so) but it does not pilot the craft automatically. You have to control speed and heigth (and even help with the yaw if the subject moves very quick). As it is now it is not a full automated tracking solution, but it assists you pretty well to keep an instance such as the dog on focus. It requires some piloting skills but it allows you to do pretty cool things
I would bet Neil would have a big hit on his hands if he could emulate what the P4 is doing. At any rate, I would like to test this out with a wake boarder behind a boat once it warms up in another 6 weeks in Florida. Neil, let me know if you want a beta tester for this. I would have no problems promoting the video of wakeboarders flying through the air if it works.
The way I see it, VS seems well ahead of DJI on the visual tracking technology. They just need to make some choices on how they interface that with craft controls, either through their SDK or through their own flight controller (AKA Autopilot style). For the time P4s becomes widespread and trouble-free, similar visual control modes will be ready from 3rd part developers and most of the features can be made compatible with the P3 (since they are not hardware-dependent). The challenge VS guys face is that while very powerful for some particular uses (and very sexy), visual tracking is just an autopilot tool that is always going to be very vulnerable and probably unsafe for everyday flying except in expert hands. In order to make sense commercially as an autonomous software solution it needs to be integrated with other tools (GPS and beacon-based). Right now this market is very well covered with comprehensive solutions such as Autopilot and Litchi, which are a couple of steps ahead. So the logical solution would be to merge VS technology with these leading automated flight control packages instead of the pain and trouble to develop new solutions by either of them to get all done from scratch, with the (comercial) pressure of DJI saying that they "already have it all"). By no way I have a deep knowledge of this particular technological market so these are just my 2 cents.
I agree in principle, but I believe DJI utilises all three front facing cameras.
I also support the notion of VS teaming up with another 3rd party.
Dji uses the main camera to visually track the object as the front cameras don't have a big enough viewing angle. If VS can drive the phantom to follow the subject then my money is theirs. Neil. Is there an issue with with driving the phantom to follow the object like dji?
I just got back from testing Autoflight's Airspace peer-to-peer Follow Me function and am disappointed. I was cruising around a parking lot going 5mph on a hover board and it could hardly keep me in the frame. That was a bust!
That is interesting, what was the quad heigth and distance re the beacon?. Was it an stable peer to peer network or could there be interference issues? I have only tested it once and very superficially because I do not have data plan in my ipad, and I did not notice any issues. But I suppose it could be hard to keep the track if the drone is very low and you are not moving straight, further if the peer-to-peer connection was bad.
46 feet in the air. Distance of 20 feet from the me. I played around with the altitude of the subject from 0-30 feet high and no matter what, it could mostly not keep me in the frame unless I was going very slow as in casual walking speed.
Hmmm maybe you should move that to to one of the AP threads, they will tell you if it is anomaly or expected behavior. I am curious and I'll try it this week, but a priori my guess is that those conditions are on the edge of what is feasible for GPS tracking precision at such distance. With roughly 5 m horizontal error in GPS position and at only 15-20m distance it makes sense that a moving subject goes out of frame at the minimum direction change. The beacon speed relative to the area framed at that distance is very high and even with a high latency I would not expect anything close to a visual tracking system.
- Joined
- Mar 24, 2015
- Messages
- 312
- Reaction score
- 138
This is exactly correct, except that the stated distance was only 20ft (6m)! Visual tracking algorithms will beat GPS in this case every time. Whether or not this is actually safe given the current state of the hardware (no collision avoidance on the P3, only forward collision avoidance on the P4, etc) is an entirely different question. As for normal / safer flights (i.e. distances further than 10m) GPS will win every time.
Hi Neil, weather in the UK still poor but whilst grounded had a little play with the app and in particular the walls feature which I find very interesting. On large areas it seems easy to place the limit with a wall but for a small area having zoomed in the wall is as thick as your finger making it difficult to be exacting with the placement. Is there any adjustment to the wall dimension when necessary to zoom in? Thanks Ian .
That would be excellent. I'm actually curious to see how VS would perform with Wakeboarders. I've noticed in the past that the wake causes some problems with tracking because of how unpredictable and foamy it is. Are you on Android ?
Similar threads
