Best way to use Litchi for orthomosaic collection?

[Greg_R]

I have been using ancient.land for missing creation, and then Litchi for execution. My results have been excellent, but my biggest complaint is the mission duration. I know that I can’t use curved turns, because the actual points are bypassed and photos will not trigger, and when using straight line paths, the software forces a stop at each point. This greatly extends the time need to complete a given mission.

I have tried other software (Map Pilot Pro) that takes photos while in motion, but have I had frequent issues getting connected with the app to even get started, and mission interuption for battery changes has been problematic as well.

I have had really good success with Litchi, and I understand that the thing I’m complaining about is not a “bug”, but rather the way the app is supposed to work. Is there a different process I can use to improve effiency, like perhaps deleting all intermediate points between row ends, and setting time-interval photos? Assuming that is an option, my next hurdle is how to calculate the interval to achieve the desired overlap percentage.

Thanks for any input you may have.

[wesbarris]

@Greg_R

My results have been excellent, but my biggest complaint is the mission duration.

The easiest way to speed up the mission is to increase the cruising speed (Mission Settings). I don’t know what speed you are currently using but this is an option.

Are you using any “Wait Actions” before taking a photo? If so, try removing them to see if it still works. I found that I do not need to uses “Waits” in my panorama missions.

@Greg_R

Is there a different process I can use to improve effiency, like perhaps deleting all intermediate points between row ends, and setting time-interval photos?

Yes, then select all waypoints and do a batch edit. You can specify the photo capture interval using either time or distance. Perhaps using distance might work for you.

[Greg_R]

[quote=“wesbarris”]
The easiest way to speed up the mission is to increase the cruising speed (Mission Settings). I don’t know what speed you are currently using but this is an option.[/quote]

I’ve been flying ~10mph, which I’ve seen recommended as the fastest recommended if you are taking photos without stopping. To get my desired 75% overlap, I only travel 103’ between photos, so I’m not sure the drone could even get too much faster, or how trying to get faster (and all of the associated stops) would affect overall battery life.

[quote=“wesbarris”]
Are you using any “Wait Actions” before taking a photo? If so, try removing them to see if it still works. I found that I do not need to uses “Waits” in my panorama missions.[/quote]

I am not. I only specify that the gimbal be at -9 degrees and “take a photo”

@wesbarris

select all waypoints and do a batch edit. You can specify the photo capture interval using either time or distance. Perhaps using distance might work for you.

I’ll have to try this out. The mission I’m trying to improve on right now has waypoints (photos) every 103’. If I delete almost all of the row, and have photos triggered at that distance interval, that could save a great deal of time. I think I’ll keep 2 points at each end to make sure my area edges get captured properly. Even if it stops at those handful of points, I should have a significant time gain.

[Sam_G]

@Greg_R

I think I’ll keep 2 points at each end to make sure my area edges get captured properly.

Thats my method for making it easier to adjust speed and photo interval.
Your speed is about right for my use too, about 10-12 mph (using Mavic 2 Pro).
You can also use the ‘Z Check’ adjustment in AncientLand to eliminate the points between the edges. That saves some time after you import the CSV file into Litchi. It will only set Waypoints at the ends of your travel lines.

[wesbarris]

@Greg_R

I’ve been flying ~10mph, which I’ve seen recommended as the fastest recommended if you are taking photos without stopping.

But you are stopping. At least that is what you said in your first post. You are using straight-line missions which forces a step at each waypoint.

@Greg_R

I only travel 103’ between photos, so I’m not sure the drone could even get too much faster

I think you can go faster. I set my speed higher when running photo missions. I set it to 20mph. I haven’t bothered to look at the exact speed though.

@Greg_R

I only specify that the gimbal be at -9 degrees and “take a photo”

If you replace the “Tilt Camera” action with an interpolated -9 degrees at each waypoint, that would reduce the time too.

[Greg_R]

@wesbarris

But you are stopping. At least that is what you said in your first post. You are using straight-line missions which forces a step at each waypoint.

Yes, I am, but I don’t want to. As I understand the functionality, If I don’t use straight line missions I won’t get ANY photos (because the waypoints…at least some…won’t actually be reached). The tip that Sam_G gave me is very helpful in that my planning tool can make the simplified mission plan, rather than creating a more comprehensive one that I have to then go and remove a bunch of waypoints from.

I’m not sure how well that would work for an area with a very high object in the center, because there are also now far fewer spots for the altitude to be adjusted via elevation data that each waypoint has, but for my current project, that shouldn’t be an issue.

@wesbarris

I think you can go faster. I set my speed higher when running photo missions. I set it to 20mph. I haven’t bothered to look at the exact speed though.

I did see that Litchi help says “If the requested interval exceeds the camera’s capability, the photos will be taken at the maximum possible speed.”, so maybe I’ll bump it up just a little bit and see what happens when in flight.

[wesbarris]

@Greg_R

I did see that Litchi help says “If the requested interval exceeds the camera’s capability, the photos will be taken at the maximum possible speed.”, so maybe I’ll bump it up just a little bit and see what happens when in flight.

In this case “maximum possible speed” is referring to how fast the drone can capture a photo and save it to the microSD card, not the speed that the drone is traveling.

[Sam_G]

@Greg_R

As I understand the functionality, If I don’t use straight line missions I won’t get ANY photos (because the waypoints…at least some…won’t actually be reached).

Thats true if you are using any Waypoint Actions. If you use Photo Capture Interval to take photos they will be taken regardless of the Straight or Curved Line setting. I do set my missions to Straight Lines on mapping missions.

@Greg_R

I’m not sure how well that would work for an area with a very high object in the center, because there are also now far fewer spots for the altitude to be adjusted via elevation data that each waypoint has, but for my current project, that shouldn’t be an issue.

Good point! If you have a situation like you described, you can easily add more waypoints in Litchi to adjust for those altitude changes.

[Greg_R]

Thank you both very much! I believe this is will achieve my goal using the tools I am most comfortable using. I did decide to add back just a few waypoints to the center of the longest runs (it’s an odd shaped area), mostly for just a little more flexibility in terms of interrupt/resume options.

If either of you would be willing to take a quick look and offer an opinion, the mission is here: Mission Hub - Litchi

The goal is to collect photos with a 75% overlap and then process them into a photomosaic for my client. While my deliverable, processed by MME will also contain orthmetric information, a precise level of measurement detail is not required, so I do not use ground control points or make any claims on the accuracy of the lat/lon data included in the result.

[Sam_G]

That is quite an interesting and odd shape! I hope it turns out well. Let us know.

[Greg_R]

The relevant event is our county fair. That actually takes place on land owned by a local sporting club and a farm. Some type of protected land status prevented the use of part of the sporting club as “parking”, so they expanded into an adjacent farm field until that status/prohibition expires in several years.

My client in an “Event Management” company that does parking/security/first aid etc. They prefer to use current imagery in their logists planning software, rather than relying on Google images that can be outdated in terms of existing buidling, etc.

They asked me to go ahead and include the currently disused areas, and I’m also trying to exclude as much wooded area as I can, as that also is superfluous. As much as it hurts my OCD, this odd shape allows me to cover the required area while being able to move as needed to maintain VLOS and also not push the limits of signal strength while over heavily wooded terrain.

[archer]

Why don’t you just fly the complete mission at 10 mph ( with curved turns ) and record it all
as a 4K video.
Then you can extract photos from single frames from the video file.
This will save a lot of flight time.

[Sam_G]

That would seem like a LOT more time to do. Scrolling thru the video to extract frames, trying to be sure they overlap. Then you still have the same amount of processing time of stitching them together.

[Greg_R]

That was my gut feeling too, but I’m admittedly very new to this space, and fortunate to have stumbled upon some side gigs that I believe are definitely with my capability, but I certainly want to be working as efficiently as possible.

If @archer knows of some tool that would facilitate what he suggests, I’d be willing to test it out. I don’t know of any program that can automatically take video and automatically “extract every 100th frame and convert to .jpg” or similar. Seems very processor intensive.

There is one service that will take video and convert directly to a 3d model (SkyBrowse), but that’s not the result that I need, nor can I afford the $300/month price, even if it did amazing 2d as well.

[Greg_R]

Just a quick follow up. I flew the above linked mission last night, and the drone performed as hoped. I did some comparisons to another job I had to compare specs. Altitude and overlap were slightly different, but these are the closest projects I have for comparison.

Older project: ~16 acres, 2.4 miles mission length, 304 photos collected. Drone stopped at every waypoint for photo. Mission average speed was 4.69 feet per second. The mission included one battery change.

Last night’s project: ~62 acres, 4.7 miles mission length, 282 photos collected. Drone took photos while in motion, stopping only at row ends. Mission average speed was 12.6 feet per second! This mission also included one battery change.

Thank you again for the input you all provided. This will make a very big improvement in my efficiency!

[Octavio]

Hello Greg, when calculating your overlap percentage, there are 3 important factors for the overlapping. They are Drone speed, Drone altitude and Photos taken in distance or time. Please keep in mind that your exposure should be as fast as possible.

[archer]

What is it exactly are you trying to do with the drone ?

[Greg_R]

See thread title. I am (was) attempting to improve the speed at which I collect photos used to create orthomosaics. The info already given in this thread has provided the info I need.

[Jeff_Gaudette]

I have seen a thread on the WebODM forum discussing frame extraction from video. Might be worth a look even though it looks like you have achieved what you set out to do.

Cheers,

Jeff