Are Mini 2 photos georeferenced?

[nicdain]

Hi everybody,
there is a question about Mini 2 photo files that I would like to ask: are they georeferenced? I mean, may I open a Mini 2 photo in a GIS application (i.e. qGIS oe ArcGIS) and have it positioned and oriented correctly according to its coordinates (I assume GCS WGS84)? Or should I manually georeference it in respect to known ground coordinates?
Thank you!
Niccolo'

 

[surfbird]

It's in the EXIF-data

 

[Howard70]

Hello Nicdain:

I‘m not sure, but I think that the EXIF location data mentioned by surfbird applies to the center of the image (assuming the recording camera was pointed straight down), or perhaps more accurately, the location of the drone when the image was recorded.

When I think of a ”georeferenced” image, I imagine an image where the location of each pixel can be interpolated such that the image can act as an aerial mapping photograph. That might be more complex to achieve with the EXIF data as it would require information about the spread of pixels across the landscape.

Howard

 

[flyrunner]

They are georeferenced on the central pixel for what I read. I think this is valid for all drones. I did some photogrammetry with the Mini 2 and the results were pretty good even if I had to take the photos manually.

 

[nicdain]

They are georeferenced on the central pixel for what I read. I think this is valid for all drones. I did some photogrammetry with the Mini 2 and the results were pretty good even if I had to take the photos manually.

that's what I would like to do! I hear that the SDK that allow the mini 2 to perform a pre-planned flight are not yet available. So you shoot every photo manually? And I suppose you evaluate the photo and strip overlap approximately...
What software do you use to create the photogrammetric model, orthophoto etc.?
Thank you!

 

[flyrunner]

that's what I would like to do! I hear that the SDK that allow the mini 2 to perform a pre-planned flight are not yet available. So you shoot every photo manually? And I suppose you evaluate the photo and strip overlap approximately...
What software do you use to create the photogrammetric model, orthophoto etc.?
Thank you!

Yes the SDK was not released yet so no mission apps work with Mini 2. I did the overlaps manually by using 2seg timer (this was I can concentrate on the flight direction). I uploaded to different platforms: DroneDeploy, Pix4D, Maps Made Easy and Carlson Photo Capture. By far the best results were with DroneDeploy. I uploaded all the same photos to the platforms and the results were pretty good.

I'm doing tests with survey to compare results and currently using the Mavic Air 2 with the same method.

 

[nicdain]

Yes the SDK was not released yet so no mission apps work with Mini 2. I did the overlaps manually by using 2seg timer (this was I can concentrate on the flight direction). I uploaded to different platforms: DroneDeploy, Pix4D, Maps Made Easy and Carlson Photo Capture. By far the best results were with DroneDeploy. I uploaded all the same photos to the platforms and the results were pretty good.

I'm doing tests with survey to compare results and currently using the Mavic Air 2 with the same method.

Thank you flyrunner for the valuable information!

 

[flyrunner]

Yes the SDK was not released yet so no mission apps work with Mini 2. I did the overlaps manually by using 2seg timer (this was I can concentrate on the flight direction). I uploaded to different platforms: DroneDeploy, Pix4D, Maps Made Easy and Carlson Photo Capture. By far the best results were with DroneDeploy. I uploaded all the same photos to the platforms and the results were pretty good.

I'm doing tests with survey to compare results and currently using the Mavic Air 2 with the same method.

Also I keep the drone in cinematic/tripod mode to keep the speed lower.

Be careful that with the Mini 2 in nadir/90º position the gimbal goes up some degrees every time you stop moving to change direction. This happened to me in all modes and only way to bypass it is to slow down very gradually.

 

[DanMan32]

How can individual pixels be georeferenced? What's in the EXIF is where the camera was when the photo was taken, not where it was pointing.

If you're pointing the gimbal straight down, then incidentally the EXIF would reference the center of the picture.
But if you tilt the camera up, then the EXIF would reference the lower part of the picture up until the scene no longer includes the ground the AC is over.

 

[flyrunner]

It

How can individual pixels be georeferenced? What's in the EXIF is where the camera was when the photo was taken, not where it was pointing.

If you're pointing the gimbal straight down, then incidentally the EXIF would reference the center of the picture.
But if you tilt the camera up, then the EXIF would reference the lower part of the picture up until the scene no longer includes the ground the AC is over.

It's called photogrammetry! Doesn't matter if the camera is pointed at 90° or 60°. The center pixel is the one referenced. The process of photogrammetry based on x,y,z + calculations from photo overlaps extrapolate that central coordenate to the rest of the pixels.

 

[Meta4]

They are georeferenced on the central pixel for what I read. I think this is valid for all drones.

It's called photogrammetry! Doesn't matter if the camera is pointed at 90° or 60°. The center pixel is the one referenced. The process of photogrammetry based on x,y,z + calculations from photo overlaps extrapolate that central coordenate to the rest of the pixels.

You are confusing two different things here.
The location in the image metadata is (and can only be) is the location of the drone when the image was taken.
The camera cannot calculate location data for things in the picture.

You can process the images using photogrammetry to create a georeferenced orthophoto map and use that to derive the location data for any point on the map.

 

[Rich QR]

How can individual pixels be georeferenced? What's in the EXIF is where the camera was when the photo was taken, not where it was pointing.

I used to work as a software developer for some GIS software. There are two concepts being mingled here. A "Geotagged" photo has EXIF data telling where the camera was located, at least by lat/long, and optionally by altitude. Camera angle, both altitude and azimuth, may also be included. At best, this says "The camera was here, and pointed in that direction".

A georeferenced photo has a mapping of pixels from the photo to world coordinates. At a minimum, it would mean the actual world coordinates of one pixel, plus an orientation, an x-scale, and a y-scale. But that only works well if there's a nice linear relationship between coordinates on the photo and coordinates in the world. So often, georeferencing would require a more complex sequence of pixel coordinates on the photo combined with their corresponding geographic coordinates on the earth. Some sort of "rubber sheet" algorithms is used to interpolate the coordinates of intermediate points.

Here's a description of what's involved in georeferencing a photo:
How to georeference images in ArcGIS | Map and Data Library

The georeferencing info is often stored in a file separate from the image itself.

Geotagging is easy, and most drones should be able to put their location into the exif data, along with the camera direction. I don't believe any drone can automatically produce georeferenced photos, though. In theory, over flat ground, they should be able to get pretty close. Hills or any other terrain variations captured at oblique angles throw a monkey wrench into the process, though.

 

[flyrunner]

Sorry, but the camera has no idea of the location of anything apart from where the drone is.
The location data in the image metadata is the location of the drone when the image was taken.

The exif contains gps information from the moment the photo was captured. That info i

You are confusing two different things here.
The location in the image metadata is (and can only be) is the location of the drone when the image was taken.
The camera cannot calculate location data for things in the picture.

You can process the images using photogrammetry to create a georeferenced orthophoto map and use that to derive the location data for any point on the map.

Probably didn't explained correctly or I may be completly wrong because I started to read about mapping with drones a couple of months ago. I meant to say that the exif info with GPS coordinates is associated by photogrammetry software to the central point of a photo ("central pixel") and then extrapolated to the rest of the pixels. Is this correct ?

 

[Meta4]

The exif contains gps information from the moment the photo was captured. That info i

Probably didn't explained correctly or I may be completly wrong because I started to read about mapping with drones a couple of months ago. I meant to say that the exif info with GPS coordinates is associated by photogrammetry software to the central point of a photo ("central pixel") and then extrapolated to the rest of the pixels. Is this correct ?

No .. the exif location data for the image file is the location where the drone was when it took the picture.
It has nothing to do with any pixel.
Photogrammetry is what you do to the overlapping images to create an orthophoto map.

 

[tvtess]

Yes the SDK was not released yet so no mission apps work with Mini 2. I did the overlaps manually by using 2seg timer (this was I can concentrate on the flight direction). I uploaded to different platforms: DroneDeploy, Pix4D, Maps Made Easy and Carlson Photo Capture. By far the best results were with DroneDeploy. I uploaded all the same photos to the platforms and the results were pretty good.

I'm doing tests with survey to compare results and currently using the Mavic Air 2 with the same method.

So if I take photos manually, I can upload them from my Mavic Mini 2 to DroneDeploy or Carlson Photo Capture to create 3D maps? I bought a Mini 2 for this purpose, I should have done more research.

 

[Meta4]

So if I take photos manually, I can upload them from my Mavic Mini 2 to DroneDeploy or Carlson Photo Capture to create 3D maps? I bought a Mini 2 for this purpose, I should have done more research.

You could but it's quite difficult to manually capture the images with enough overlap for anything more than a very small area.
If you want to do mapping projects, you really need one of the drones that's supported by DroneDeploy.
More info in this post from a few days ago:

Mapping

All I have available is a Mini 2 and I need to create a map of 40 acres. Because there isn't SDK support I'll have to do this manually. Any suggestions on how much overlap I should have for each picture? Software I should use to stitch them together? Any other tips that will help me succeed...

mavicpilots.com

 

[RFilipe]

Yes the SDK was not released yet so no mission apps work with Mini 2. I did the overlaps manually by using 2seg timer (this was I can concentrate on the flight direction). I uploaded to different platforms: DroneDeploy, Pix4D, Maps Made Easy and Carlson Photo Capture. By far the best results were with DroneDeploy. I uploaded all the same photos to the platforms and the results were pretty good.

I'm doing tests with survey to compare results and currently using the Mavic Air 2 with the same method.

Hi there,
I tried your tip about DroneDeploy...uau just amazing and fast. I use my drone just for fun, do you know any software that i can use for free? Thank you.

 

[flyrunner]

Hi there,
I tried your tip about DroneDeploy...uau just amazing and fast. I use my drone just for fun, do you know any software that i can use for free? Thank you.

MapsMadeEasy it's free and the results are good if it's not critical work
Cuurently I use DroneLink to automate the captures and then process the photos with MapsMadeEasy or one of the others.