Hello,
we collected a few thermal images (r-jpeg) from a M210 RTK V2 and a XT2 camera, just for testing purposes (using UgCS auto mission).
We noticed that ContextCapture Master does not see the FLIR radiometric jpegs, we cannot make an ortho using IR files.
Also, the camera misses pose metadata, resulting in issues for AT.
Any advice to work with thermal jpegs in CC?
Thanks.
Hm.
New here, but that is about my question too only now using a Flir Duo Pro R 640. I am even suprised CCmaster doesn't seem to know how to read the specs from EXIF or to know about this camera from its database, and I'm afraid I don't see any answers to this thread :0
I've processed a couple of projects from data created by the DJI XT2. There are couple of points to note here:
1. The thermal camera has a much narrower field of view compared to the RGB camera (14mm focal length compared to 8mm respectively). Therefore in order for the thermal AT to work effectively, you need to set up your flight planning as if it was for a 14mm focal length to get the correct overlap and sidelap on the thermal images. This does mean that the corresponding RGB imagery will be significantly overlapping - but it does mean that you will get the AT solution to work on the thermal imagery. Note that thermal (in spite of the longer focal length) will have a lower imagery resolution in GSD terms when compared to the RGB images.
2. CC Update 15 most certainly has 2 camera models for the XT2 in the camera database - choose the correct one for your image dimensions. However, my recommendation is that you create your own calibrated camera model for both image types on the XT2 for your specific cameras. This will speed up your AT and provide more reliable results, especially if positional accuracy is important. The procedure for doing this is clearly outlined in the Drone Capture Guide found in this location: C:\Program Files\Bentley\ContextCapture\doc .
3. I have not had any issues loading or seeing the thermal jpegs in CC - if this continues to be an issue, please log a service ticket after having installed the latest version of CC. Likewise, CC can certainly read the EXIF data for positioning info.
4. Having addressed the above then the processing stages tend to be fairly simple. Firstly you need to note that thermal (in spite of the longer focal length) will have a lower imagery resolution in GSD terms when compared to the RGB images., So if the resultant model surface geometry is important to you, then my recommendation is that you include the RGB imagery in your processing as it can supply the reference model for the thermal imagery in rendering. If you process the thermal data on its own, then your 3D model will be fairly coarse in terms of geometric accuracy.
5. In CC, load both sets of imagery data. If some of the EXIF data for the thermal imagery is missing then the AT process can calculate the camera positions for the missing POS data. I find that it is worth including tie points between the RGB images and the Thermal images across the scene to ensure good alignment between the two image sets.
6. Once the AT solution is complete then you can then set up the reconstruction and productions as normal, remembering to select the Thermal option when confirming the Texture Colour Source.
Hopefully this helps.
There are a number of different camera and lens options out there, so the CC camera database isn't exhaustive, but there are two options to solve this. Log a service request to get your specific camera added to the database (which will benefit all CC users ). Alternatively you can just create your own so that you can continue with your project work, as delays can cause frustrations.
At a bare minimum you will need to know the sensor size in mm and the focal length of the lens that you will be using to add your own camera model to your CC database. The good news is once done, it will persist even through running updates on the CC versions, and you can customise the name of the camera to make it easy to find. Once you have the basic model information set up, then you can run a calibration AT (see my response to the earlier question on Thermal cameras above) to create a more detailed camera model that includes computations for lens distortions and save that as an update to the original camera model you created.
The specific sensor details for your camera can be found here: https://flir.custhelp.com/app/answers/detail/a_id/3395/~/duo-pro-r%2C-vue-pro-%26-vue-pro-r-sensor-and-lens-information/session/L2F2LzEvdGltZS8xNTkxOTIzNjE4L2dlbi8xNTkxOTIzNjE4L3NpZC9mVUY4VEo2QUZEWDBkUyU3RU9VT21yTVpINXNsNzgxYnhwUEtVQXJhbm5WTXc5QlY1NjRydSU3RTc0MUp1RWtKTnRvMm9JaTBuaXVzQnNNenM0WkZnRndMM2JiQmdzWmwwWW5lRlNJQWNKWU1sJTdFa1N4b3A2bWZtRlpTYUElMjElMjE=
Then follow the outlined processes above and you should get a thermal rendering on your 3D model.