In this first trial the result is a bit confusing: the model is up side down. it’s textured but no orthomosaic even in the quality report.
I think I did everything ‘by the book’ as I did earlier with samples.
Camera is not a gopro but extra wide high definition camera, geo-referencing has been made manually from GPS data without elevation data on the exif of each picture.
Thank you in advance for you help
Regards
Dominique
Hi Dominique,
this is interesting. Could you put the quality report and a screenshot of the raycloud (that shows the upside-down effect) in a dropbox link and share it with us.
I am sure that Pix4D will also like to see the report
Cheers
Marc
I have the exact same problem! Just asked the same question but my images are geolocated (EXIF tags), and I used a VRS-RTK system for my GCPs.
Marc, please take a look at this OneDrive folder with the quality report and screenshot. I suspect it is similar to Dominique’s issue.
Hi Marc,
I crosschecked the Geo-referencing and it was completely wrong: The picture at the far east was referenced as the far west (without Indians)! Then I made a test run without any referencing and got the right model (without dimensioning). A third test run with only 5 manually Geo-referenced pictures gave me a correct result.
For Geo-positioning I used the flight data from a drone: I made a synchronization between camera and drone clocks and with Geosetter manually selected the time data of the camera, found the same moment in the drone flight report and imported manually the position into the exif of the picture…
Not the best way to save time!
Regards
Dominique
@Dominique: So the problem is solved! Good news! You should consider a system that automatically registers coordinates to your images.
@Jacob: Did you firstly process the 1st step without the GCPs? Was the reconstruction fine before adding the GCPs?
Cheers
Marc
Well I’m afraid I launched the computing late in the evening in order to get a result next morning… first step… I just jumped over 
I succeeded with an other test run (other pictures) with few Geo-referenced points plus manual stitching and am now running a third test… I hope I will have enough results on Monday, last free test day (I’m going to work during weekend.
Regards
Dominique
Marc,
I first ran Step 1 with no GCPs imported. This produced a low-resolution orthomosaic. I then imported the GCPs, marked 2 of them as well as a manual tie point and re-processed. This also produced an orthomosaic. Since I forgot to mark all GCPs, I then did so and re-processed. The last processing produced no orthomosaic.