"Mapping" with Sparky

Haggi · Aug 4, 2019

I'm trying to create a big flat image of the ground, and have yet to find the sweet spot of settings..
Has anybody created a workflow which will give a high detail image?
Any advices?
Altitude here is around 40m/130ft, 50 photos and approx 1.5cm/pixel..

Original file link (44mb)

Brojon · Aug 10, 2019

Have you looked at mapping apps?

Haggi · Aug 10, 2019

I landed on Drone Harmony app.
Its not a special 2d/3d mapping solution, but more of a on-site inspection service.
It goes well with my Spark, and is able to provide according to my needs.
Hasn't done extensive research and testing regarding the other mapping apps. Most of them has only/better support for IOS devices..
As an Android only user, I'm fairly happy with Drone Harmony!

Another post about this

chascoadmin · Nov 29, 2019

While you can map with it, Litchi is not a traditional mapping app, but I would highly recommend it for Sparky. It is one of if not the most versatile autonomous flight apps available.
Back to your original question, good orthos are typically based on the ground resolution or GSD that you desire. Really good maps are 1in/px and professional survey maps are 0.5in/px. It's hard to say with the spark camera, but I am guessing 150ft AGL with at least 70 frontlap and 65 sidelap should make a good map. Litchi would be more manual, but at 150ft with the 2-second interval shutter you would need to set your speed at about 10-15mph.

dudez · Nov 29, 2019

Here are my recent takes with Drone Harmony app -
- 2D: https://api.soar.earth/short/60a1o707r
- 3D rendered by Pix4D: https://skfb.ly/6OT6T

Basically applying tips from https://support.pix4d.com/hc/en-us/articles/115002463243-Image-acquisition

Bob Hospadaruk · Mar 13, 2020

Using Litchi, what is a good way to stitch together the resulting photos?

chascoadmin · Mar 13, 2020

Bob Hospadaruk said:
Using Litchi, what is a good way to stitch together the resulting photos?

The one major downfall to mapping with the Spark is the fact that DJI truncated the geotags to the nearest second so processing solutions that align and bundle adjust using the GNSS values cannot do so because every image within about 60ft of that second look like they are in the same place. Obviously a goof on their part as our cell phones tag more accurately than that.
I would suggest using Microsoft ICE and using the planar motion option instead of auto. Sometimes auto will negotiate how the image were taken, but we know how.