LITCHI HELP: Will Litchi way-points and Go4 Intelligent Flight Mode play together?

[KDog]

Hi All,

Apologies if this has been discussed here already. I bought litchi about a week ago and I've been having a play around with it. I'm going to be on a beautiful, winding coastal road in about a week and there's a specific shot sequence I'm really hoping to get, but it's relatively complex and I'd say, either impossible or fraught with danger to try to manage it manually. Hence enlisting the help of some robots.

To provide a very brief overview of the shot, the plan is to track a car, first flying backwards until the car overtakes the drone, then flying forwards with a several instances of flying sideways. There are a number of direction and elevation changes to capture different framing and to fly over hills and go below the the elevation of the cliff-top road. It looks fantastic in my head.

I've been testing a number of techniques to help capture the shot. The first was Litchi waypoints with POIs to focus the gimbal. Since it's going to be very difficult to anticipate where the car is going to be at any moment, POIs aren't much help. Track mode would no doubt do a reasonable job of tracking the vehicle, but I'm not confident in managing the elevation and direction changes manually because the framing won't allow me to see what I'm flying towards at all times, and there's a high chance I'd either crash or fail to get the shot I'm after because I'm worried about crashing. The solution that I've arrived at so far is to use a litchi waypoint mission to manage flightpath and elevation while managing aircraft rotation and gimbal pitch manually. This has produced reasonable results but has taken a pretty high number of takes because there are still a number of ways to make small stuff-ups, and the shot will only achieve the intended effect (wow factor) if I can pull it off in one continuous shot without cutscenes. The slightest jerk of rotation or miss-timing of the gimbal can ruin the whole shot. I've got all my sensitivity settings dialled down as low as practical, but it only takes the slightest little error to ruin a shot like this IMHO.

So my question is this: If I started a litchi mission and then jumped back into Go4 and put it into track mode, could I get the best of both worlds, or would an intelligent flight mode in Go4 cancel a litchi mission? Or would I be unable to set an intelligent flight mode in Go4 while flying a litchi mission? Or does litchi cancel a mission if I'm not within the app?

Obviously I could test this, but it's a Sunday night here and I probably won't get to fly until next weekend, so just wanted to reach out and see if any litchi super-users have any thoughts around this. If the two won't play together, does anyone have any thoughts on a better way to nail this shot than what I'm currently working with (and that doesn't involve buying an Inspire 2!)?

Thanks in advance.



K.

 

[pixl]

I know nothing about Litchi, but from pretty extensive drone experience on an Inspire 1 Pro, I'd say this is not do-able smoothly on the Mavic, and even difficult on the Inspire with two operators. I think definitely you do need a drone with independant full rotation camera yaw. It's an inspired shot to try to do, and difficult for very experienced cinematographers. I'd love to see the result -- it's great to see people pushing the limits of what's possible! Even Emmanuel Lubezki would be impressed of you could pull this off.

 

[KDog]

I know nothing about Litchi, but from pretty extensive drone experience on an Inspire 1 Pro, I'd say this is not do-able smoothly on the Mavic, and even difficult on the Inspire with two operators. I think definitely you do need a drone with independant full rotation camera yaw. It's an inspired shot to try to do, and difficult for very experienced cinematographers. I'd love to see the result -- it's great to see people pushing the limits of what's possible! Even Emmanuel Lubezki would be impressed of you could pull this off.

I see. Well that's a bummer...

The idea here is to try to emulate the two operator setup of the Inspire, with Litchi taking the role of the pilot and me taking the role of the camera op. It's easy enough to hand control of the drone over to litchi with a bit of planning, so theoretically if I can be savvy enough as a camera op, then I should be able to pull it off. Ultimately what I'd like to try to achieve is to replace two human operators with two robot operators that would no doubt to a much better job, but not sure if the technology is there yet.

Anyway, as you say, it can't hurt to try. Better to think big and fail than to think small and succeed.

I'll post the results up here once I have them (or if nothing gets posted that's likely an indication of an epic fail!).

 

[pixl]

It may well be possible, but where I have doubts is in you getting the cinematic fluidity that you're hoping for -- particularly the move into and out of flying sideways -- which to retain that fluidity would likely have to be mostly one long pan, and at the same time descending and then ascending. I'm not sure if there are any helicopter pilots here (there are a couple on Inspire Pilots), but they might have some helpful input on such a maneuver.

Sidenote: It's refreshing to encounter someone capable of expressing themselves well here. Spelling and grammar blah blah blah.

 

[KDog]

It may well be possible, but where I have doubts is in you getting the cinematic fluidity that you're hoping for -- particularly the move into and out of flying sideways -- which to retain that fluidity would likely have to be mostly one long pan, and at the same time descending and then ascending. I'm not sure if there are any helicopter pilots here (there are a couple on Inspire Pilots), but they might have some helpful input on such a maneuver.

I'm planning the transitions to be at a bit more of a distance from the subject so that I've got a bit of framing buffer, but you're right, this fluidity will be the challenge. My test runs have come up quite well, but did take a number of attempts to get it to passable quality, and were also a little shorter. That said, I still managed to get some pretty fluid shots at 54km/h, but they were of static objects, so the dynamism introduced by the car will introduce another level of complexity. I think it just takes practice and persistence, and maybe a bunch of batteries.

Sidenote: It's refreshing to encounter someone capable of expressing themselves well here. Spelling and grammar blah blah blah.

Haha - you can thank my mother for that [emoji6]

 

[pixl]

the plan is to track a car, first flying backwards until the car overtakes the drone, then flying forwards

I'd be interested in hearing your transition plans for this move... camera straight down while the Mavic slowly turns 180º...then camera slowly to horizontal ?

 

[Path Finder]

I see. Well that's a bummer...

The idea here is to try to emulate the two operator setup of the Inspire, with Litchi taking the role of the pilot and me taking the role of the camera op. It's easy enough to hand control of the drone over to litchi with a bit of planning, so theoretically if I can be savvy enough as a camera op, then I should be able to pull it off. Ultimately what I'd like to try to achieve is to replace two human operators with two robot operators that would no doubt to a much better job, but not sure if the technology is there yet.

Anyway, as you say, it can't hurt to try. Better to think big and fail than to think small and succeed.

I'll post the results up here once I have them (or if nothing gets posted that's likely an indication of an epic fail!).

Great post, hope to learn more from it! (Sorry, please post the result either way!!!!)

 

[KDog]

I'd be interested in hearing your transition plans for this move... camera straight down while the Mavic slowly turns 180º...then camera slowly to horizontal ?

Well here's the thing - since I'm not 100% sure where on my route this transition will take place due to the unknown car speed relative to the camera, I'm going to have to suck it and see when shooting I think. As best as I can tell I've actually planned the transition to take place at a very low angle (horizontal to the car) and fairly wide, to give me this buffer. Litchi will take care of the flight direction, so all I'll theoretically need to do is manage a continuous pan.

Secondary problem that I'll like have flying it sight unseen is that the framing might actually not be any good from my planned litchi flight path. I've tried visualise framing, but I've not gone to the effort of calculating field of view along the route, so it's possible the framing won't be tight enough at certain transitions. If this is the case I'll just have to adjust the litchi mission (waypoint location and altitude) in the app in between takes.

 

[KDog]

Great post, hope to learn more from it! (Sorry, please post the result either way!!!!)

Thanks - I'll be sure to post results and I'll write up a lessons learnt and 'workflow' as well if I can get it right.

The only other problem is that we're just going to be in a hire car, so no fancy wheels I'm afraid. If I nail the shot, I'll hire an Aston Martin for next time [emoji6]

 

[pixl]

Man, it all sounds fun. I'd love to be around for this.... but sounds like you may be in Australia. I'll be a little closer in a few days time (Vietnam), but not close enough.

 

[KDog]

Man, it all sounds fun. I'd love to be around for this.... but sounds like you may be in Australia. I'll be a little closer in a few days time (Vietnam), but not close enough.

Yep, just another nearly 9 hour flight!

I'll be in up Far North Queensland which is where I'm planning this shot. There's heaps to film up there, so this is just one planned scene in a little film I'm planning, but by far the most complex.

We'll see how it goes...

 

[Hollow Dog]

I remember seeing at the top of the page on the how to use Litchi app that the DJI app should be shutdown or there could be problems.
I'm following this thread.

 

[pixl]

Australia is on my bucket list, so who knows? Someday. I'll probably only get as close as Bali on this trip.

 

[KDog]

I remember seeing at the top of the page on the how to use Litchi app that the DJI app should be shutdown or there could be problems.
I'm following this thread.

Doesn't the DJI app need to be running these days to remove flight restrictions around the no fly zone database, or does it just need to be opened at the start?

I'm actually glad that I've run Go4 in the background because I've already had litchi crap itself and become unresponsive and had to take control via Go4. Interesting consideration...

 

[pixl]

I'm planning the transitions to be at a bit more of a distance from the subject so that I've got a bit of framing buffer, but you're right, this fluidity will be the challenge. My test runs have come up quite well, but did take a number of attempts to get it to passable quality, and were also a little shorter. That said, I still managed to get some pretty fluid shots at 54km/h, but they were of static objects, so the dynamism introduced by the car will introduce another level of complexity. I think it just takes practice and persistence, and maybe a bunch of batteries.



Haha - you can thank my mother for that [emoji6]

BTW, you might considering registering at InspirePilots.com and asking the same question there. There are a fair number of professional cinematographers there, and they may be able to give you some pointers. Just a thought.

 

[KDog]

Well the sum total of advice that I got from the Inspire Pilots forum was to check that I'm allowed to fly at my location.

There also seems to be a whole lotta hate over there for us irresponsible kids with our toys...

Good times...

 

[pixl]

Well the sum total of advice that I got from the Inspire Pilots forum was to check that I'm allowed to fly at my location.

There also seems to be a whole lotta hate over there for us irresponsible kids with our toys...

Good times...

yikes, sorry about that. Yeah, there are a few blowhards over there, and a particularly bad one named Dennis, from Perth I believe. He's been banned a few times for being a jerk.

 

[KDog]

Immediate post shoot update:

Just finished up, and it was a little harder than I was hoping. Firstly, my litchi mission was too long and lost reception, so I had to do a quick edit on the phone and cut it right, right down.

Lesson 1: Always do a camera forward test flight - this saved from flying into the side of a hill which litchi would have done. Steep terrain and slightly inaccurate terrain data was the cause, but the low speed test run helped and I had my finger on the sports mode button, so all good.

Lesson 2: F&#k wind. Not much can be done about that, but it stopped me from doing the reverse run to double my filming. But bloody hell that little Mavic is an impressive little drone - in sports mode it handled a very strong wind without any trouble.

Lesson 4: Practice, practice, practice. No shortcuts, I'll get better as I do more.

Of around probably 10 runs I think I got possibly two runs without any major errors in them. Will have to review the footage to see how it comes out.

I've just got my laptop here, so will have to wait until I'm back home and have my proper editing suite with all my LUTs to get it looking ok before I post [emoji51]

 

[KDog]

Righto - sorry about the delay. Bloody work getting in the way of hobbies!

As promised, here is the result of my efforts. My number one takeaway from this is that this is single most technically difficult shot I've ever attempted, and at max speed, it really pushes the MP to the technical limits of what it can do. This isn't a perfect shot, but for my first attempt, I'm pretty happy with it, and I'm confident that I'll get better with practice. The videos here are just HD for ease of getting them up on YT quickly, but it was shot and edited in 4K. So, without further ado...

The first video shows the footage taken directly off the card.

The second video shows the output of post.

WORKFLOW:

1. Prior to arriving on location, I created a Litchi way-point mission with gimbal focus and heading set to manual. I also checked to ensure there were no flight restrictions in the area.

2. Upon arriving at location, I flew a test mission with the camera facing forwards at 22km/h. This saved me from crashing, because - as previously mentioned - even though I had the waypoints set to 'elevation from ground level', some very steep terrain messed with these settings and I would have flow straight into the side of a hill. I was also losing reception around a headland, so I ended up deleting around half of the waypoints. This reduced the technical skill required to a degree, but actually the shot I had envisaged wasn't possible.

3. With the route confirmed and the test flight successful, I dialled the cruising speed in Litchi up to 54km/h.

4. Camera settings for the video was 4K, 25fps, shutter speed 1/50 with a PolarPro ND8-PL (polarised) filter, D-Log, -1,-1,-1.

6. With litchi flying the mission, all I had to do was control the aircraft heading and gimbal pitch. It still took about about ten takes to get this best bit of footage.

That gets you to post-production. For my post I used FCPX.

7. Firstly, applied the free Ground Control D-Log to Rec.709 LUT, to get me into the publishable colour space.

8. Next, manual colour grading to get a bit more pop out of the footage. I was shooting on a cloudy day, so had relatively flat light and not overwhelming colour, so needed to bring the shadows down and the highlights up a bit, and saturate the mids and highs a little.

9. Applied the free Alex4D widescreen matte. Usually I would just apply Alex4D letterbox, which applies to the clip rather than the project, but to enable my next steps, I needed a matte that would go over the entire project.

10. Scaled to 150%. No matter how close you try to get, the car looks tiny in the original footage, so I needed to scale it up. I tried to create a 2K project which would allow me to scale without loss of quality, but something weird was going on with FCPX and the scaled footage was going through to the final, so I ditched that end used a 4K project. This scaling will result in loss of quality, but hopefully since 90% of content is watch on mobile devices, it won't be that noticeable.

11. Due to the scaling, and the original footage, the car was getting either very close to the edge of the frame or was going out of frame all together. Since I had plenty of footage to play with outside the frame, I simply used keyframes and the transform tool to manually keep the car as close to the centre of the frame as possible.

12. Finally, whacked some music underneath. Never underestimate the multiplying effect some good music has on your footage!!

And voila! That's my first attempt at a complex composite shot of a moving subject travelling at over 60km/h, from a drone travelling at around 54km/h with changes in subject direction and aircraft direction, altitude, heading and gimbal pitch!

Not perfect, but a great learning experience nonetheless.