Extended Hyperlapse planning and workflow?
I've just started using Dronelink (Hobby-Elite) with my Mini2. I'm really interested in creating hyperlapses over periods of time, but I'm looking for some pointers on organising and filtering all the photo files.
For example, let's say I want to create a 5 second dronie type hyperlapse starting from close in to my house then ascending to 30m and pulling back 50m, and showing the seasons change over a year.
So a path component that flies the dronie path at 0.1m/s with a photo every 0.5m would take 17 mins to run and take 100 photos, and I'd need to run that mission every day for a year giving me 36,500 photos.
Mission Questions
- If I take off from the same paving slab each time, will the GPS error be consistent, and therefore will the positioning be consistent? I don't have GPS offset available in my Dronelink plan.
- Today was quite gusty, and the drone was bouncing around a bit with vertical altitude varying by +- 1.5m. This would matter close in I think. Would using the "achievement" parameters improve the photo consistency between flights?
- If I wanted to do a section of the hyperlapse as dusk approaches, how do I get the consistency across multiple days? Set a fixed shutter speed across the whole mission, and use ND filters and a light meter to only start the mission at similar light levels?
- Can I get the drone to move a defined distance and stop moving before it takes the photo? I've read that this seems to be available in the Map component, but not path?
- Is it best to just repeat the same mission with 100 photos each time, or to try and edit the mission to only capture what I need on each run?
- What else don't I know?!
Workflow
- A year would result in 36,500 photos. What software could help filter and pick out the best photos to make up the video? I've read about a lightroom plugin.
- I know I'll need to stabilise the footage. I have Lumafusion on the Ipad, and Resolve on the PC.
- What else do I need to consider?
Cheers
/
Comments
5 comments
Hi.
This is a very interesting project and you seem to be asking all the right questions. But...for clarity. Are you looking to create one video that starts at a season and then runs through the seasons during the length of the hyperlapse?
If so, it would seem that you are looking to pull just 100 photos out of that 36,500. That seems like a lot and you are throwing away a lot of those photos. Given that, it would (I think) make more sense to not run every day, but only on days when the weather is ideal. This would give nicer photos anyway, and less to select from.
Ultimately, which photos you pick are going to be dictated by the time that season is due to show up in the video. 100 photos across 12 months means you have 8 photos for each month, or 24 per season. You can ignore all of the photos that all out of that range, so that cuts things down a lot to start.
Per your questions:
GPS is only accurate to within about 16 feet. Without GPS offset in your plan, your best bet is to run the mission once. Land. Then run the mission again. Landing and running again gives you two shots at things being correct. It is surprising how far off things can be not just in location but also direction but, in my experience, this helps a lot.
No. But, if you limit flight to just good days anyway, that should help.
Not sure what you mean here. Exposure will try to be the same every time, even at dusk. Are you trying to introduce motion blur? Sorry if I'm not getting it.
Do you have inspection in your plan? That would do that.
You are only going to need some of them but, personally, I would just run the whole thing - it leaves less room for error.
Lining up the photos (or video if you go that way) is a bugger. I did this back in 2015 (see here https://vimeo.com/avios/zabriskieseasons) and, while I can do a better job now, lining up was a pain. It is easier to line up photos because, if you are smart, you will drop everything into Photoshop as layers, ask PS to align them, and then export each layer as a photo. That is what I did on a recent photo timelapse which has not been released yet but if you email me at barry@aviosaerialmedia.com I can send you a link to the sample.
Also, try to run at a time when the shadows are going to be similar. If they move all over the place it is very noticable.
Good luck with the project!!
Thank you Barry, Your posts and videos have helped me get past most of my daft questions!
Yes, just looking to pick 100 or so photos out of the collection. I am wondering if I'm over complicating this. Most examples I've seen seem to take a few seconds of video from each mission run, and blend them together over a relatively small number of runs.
Why would achievement not help? If the wind is blowing the drone off course, won't it wait until it is within tolerance before taking a photo?
I think I'll start flying some test missions to help iron out what will work with the drone.
I'll then need to try and optimise the workflow, and see what helper software is out there. I can imagine it will get difficult to pick the the correct photo out of each run.
I'll send you an email for that hyperlapse - thanks for the offer.
DroningAl
The achievement I think of as what to do once it has reached the start point. But if you are looking to take photos you need accuracy in each photo location.
You could adjust the motion error tolerance to smaller values. But...too small and I think you are going to find the mission aborting constantly.
Thanks Barry.
Just been experimenting with various features.
Is there a way to fly a path or orbit, while taking AEB photos every 0.5m? Will inspection do that?
DroningAl
Yes, Inspection will do that for you.
Please sign in to leave a comment.