Thermal - Best settings for Zenmuse XT and Drone Deploy


I’m using DJI Inspire 1 for a mapping mission with the objective to fly a farmers field and compile a thermal map of the soil temperature. This is to indicate moist and dry areas.

What are the best settings for the Zenmuse XT when mapping with Drone Deploy ?

If you’ve had success please share share any insights and tips and if possible the final thermal map results.


I’m using the higher resolution 640 x 512 at 30hz = 15fps
Images have embedded GPS information in metadata.
I’m flying my mission relatively slow to avoid motion blur.
Approximately 230 feet (this might have been too high)
Shot at 1:00pm - 2:00 pm so the soil has time to heat up during the day.

Any help would be greatly appreciated.


1 Like

I can offer no help or advise, but I would be interested in your results once done. Good Luck !

I tried this myself and Drone Deploy didn’t play well with the XT. None of the images had geotags so couldn’t stitch in Drone Deploy because a minimum of 5 need data. Also because the combination I have uses the 19mm lens there was no overlap. Very little settings can be changed for using the XT. Additionally because it’s not a camera and it’s just a sensor that gives pixel visualisation of what it detects the resolution is super super low additionally making stitching even harder. Good luck.

I’ll also be interested in reading about your results and your use of the Zenmuse XT camera with DroneDeploy in general. Keep us posted!

any progress in getting the XT to work?

I have gotten this to work, but I had to assemble my own solution and it required a lot of time on my part.

There is an issue I have had with dronedeploy that I hope they fix. I am using an xtr640 on a matrice 600 with an A3 flight controller. Dronedeploy does not allow me to select radiometric jpeg as the photo type, I either have to use “manual settings” to get radiometric data, or forego radiometric data and use “automatic”, which generates standard jpegs. To get radiometric photos, I have to set the camera configuration in the DJI Go app and then use dronedeploy in manual camera mode on the flight. Unfortunately, the DJI Go app is quirky and if the app has connectivity problems with the drone it reverts back to standard jpeg and radiometric data is lost. This is a serious flaw and almost rendered my thermal camera useless. I came up with a workaround that allows me to take qualitative thermal photos and stitch them together. The important thing is to use the “white hot” greyscale palette as your default palette in the DJI Go app. By doing so, you can emulate a non-colorized thermographic photo. It’s not as good as having the real thing, but it’s not bad either and can show temperature deltas really well once the greyscale orthographic map is colorized. I have the system generally working at this point. Last night, I was able to process a 388 photo flight into a thermalish orthomap in about 3.5 hours. A thumbnail with some very poor (default) colorization applied is below:


So, it is possible but it’s not easy. Just a note, you are flying way too late in the day for good thermal photos. You want to capture temperatures when the delta is highest. Early to mid morning is best after the sun has started warming the ground but before the ground temperatures have evened out. Lastly, the geotags are important in mapping. If the photos do not have geotags, then you won’t really be able to create good maps. Make sure that the photos from the Inspire are properly geotagged.

Hi Will, I would very much like to talk to you regarding your experience with drone deploy, and the XT. I am working on waterfowl surveys with a large NGO and we are using simmilar technology. We are running into the same problems and I have come to a similar conclusion as yourself. Please email me at if you would like to collaborate and maybe help each other out. Thanks Roald

Very interesting. I have not yet been able to get good thermal mapping in DroneDeploy. Currently we use Pix4D and the tiff format, but if there is a way to get good data entered into DD I would try that. I am assuming that JPEG R is the preferred format in DD over tiff, because the tiff files aren’t recognized. Can someone validate this?


We currently do not support tiff formatted images. We are working to support radiometric data but at this point, we are only supporting .jpg imagery. I have had limited success with the 640 sensor using a 9mm lens. This combination can provide a much higher resolution image than the 336 sensor. This feature is still in beta and we are looking to build this into the product soon. Sadly, I do not have a firm ETA for the release. I apologize for the inconvenience.


Thank you for the insight Zach. That is very unfortunate to hear…
Hopefully your team is able to accomplish this soon!
I currently see good results using a 640 with a 13mm lens (I purchased that model to have a range of versatility)
I am using pix4D real currently with decent results.

One thing I am hoping that DroneDeploy can add soon is a built in calculation for overlap for high structures (roofs). When mapping mid and high rise structures, even 90% overlap is not enough to compensate for the added altitude and sometimes the imagery won’t stitch together properly. Just shy of literally taking off from the rooftop,there is no other way to do his other than manual shots.
Is this something you guys are working on?

Hi @LTF,

That sounds like a great feature and I can definitely see that helping out a lot of other users. I’ll forward that on to the appropriate team but other than that we don’t have any official news or announcements to make at this time.


Thanks Christina! Yes, I think it would help many, especially those that have a good understanding of how altitude adversely affects overlap.

I agree with you! It’s been noted. :slight_smile:

1 Like

@LTF The Overlap Optimizer does not work for you?

Initially I had some trouble with it. I’ll try this again. I think my main issue with this was trying to get an overlap of 90% at the roof surface. The optimizer automatically adjusts for optimal RGB stitching, but doesn’t factor in a high overlap for specialized datasets such as thermal which require much more pixel overlap due to low resolution to begin with. I will experiment with this more to see what happens, but as of now the option to literally type in the height of the tallest structure and THEN be able to set the overlap at that level would make this perfect. Especially in instances where we need more details of the structures protruding from the roof.
So if the app just “reset” the focal point to the surface of the structure that is being captured, it would make everything easy. Currently the only way to do this without worry is to start the mission from the roof itself.

Yep, I understand. Until DD has time to get this done, you may want to give Map Pilot a try if you are using an ios device. They have this exact feature and it is called Ground Offset.

Last time i looked Map pilot dose not support saving in R-JPEG image format. Also map pilot will not tolerate any other flight app running in the background, really weird stuff starts happening. Just an FYI.