Weird Problem with GPX files when using Registrator Viewer

[Gman]

I have a Viofo A119 V3 dashcam and am using RV 6.0.0.8 (seems to be without the advert problems). I have 2 (equivalent) scripts that I use to extract GPS data based on work done by Sergei here. One is a Perl script that works perfectly, the other is a Java script which has the problem. Both scripts produce what, as far as I can see, are identical files with the exception that lat, lon values are to 5 decimal places in the Java script and 6 decimal places in the Perl script. Both scripts create .gpx files with the same name as the video file and save them in the same video file directory (ie I have 5 video files and 5 gpx files all in the one directory). When I select the 5 video files in RV without any attendant .gpx files I get the full 10 minute timeline (each file is 2 minutes), the video, a placeholder Leaflet/OpenStreetMap and no speed location data - which is all as expected. When I run the Perl script to generate the .gpx files I get the same 10 minute timeline but showing the speed graph plus a route in Leaflet and an arrow that moves in time with the video. However, if I generate the same (as far as I can see) .gpx files using the Java script the videos are concatenated together and play through as before but the timeline is for 2 minutes only (ie the duration of a single file) with each speed segment overlaying each other with what looks like vertical connecting lines (see image).

It is as if RV is processing the files in parallel rather then series and both programs do run very quickly (even through a 441Mb video file!) and so the .gpx files have the same date - but then they do for both versions. Another odd thing about the faulty version is that when RV gets to the end of the first 2 minute segment it flips back to the start again as does the arrow on the map (although the video carries on regardless). A single video clip works well in both versions, it is only when multiple files are being viewed that it is a problem.

Does anyone have any idea of how RV stitches video and .gpx files together and runs them as a sequence? (Or any idea at all of what might be wrong.)

 

[jokiin]

not sure but you maybe should be using 6.0.0.22 as that was the last build ever done, there's a patched version available that fixes some of the problems with expired api's etc

 

[Gman]

Thanks for the reply. I have already tried 6.0.0.22 but without the registry update and so it wasn't showing any maps. I'll give it a go again but the weird thing is that 6.0.0.8 works perfectly fine with the Perl generated .gpx files but not with those generated using Java. As far as I can see the files are identical so I'm thinking that it must be the way the different languages create the files but if someone knew how they were stitched together that might give a pointer to what is going wrong. I'll retry 6.0.0.22 though.

 

[jokiin]

do the registry update or it will have issues, may not relate to your particular issue but best to have it working correctly as a starting point

 

[Gman]

OK I've downloaded @DeLorean's version and done the registry update but it is still exactly the same, the Perl version works, the Java version doesn't. I will say, though, that the later RV version does get rid of the error windows .

The problem is not a train smash because I can still use the Perl version but the Java one was just a bit of an improvement.

Do you know whether RV reads the .gpx data into memory or does it step through each file in turn? Judging from the odd timeline I would say that when the video bundle is clicked it runs through all files to create the initial timeline and then processes each .gpx file in turn but, with the faulty files the .gpx filename isn't changed when it starts the next segment. I'm just guessing, though - throwing some thoughts in the air for others to latch on to.

 

[jokiin]

not sure, hard to find out info about how RV does things since the developer died, he worked on it alone so nobody else really knew the inner workings

you might also check the thread here where @lacibaci does something similar for his camera using exiftool https://dashcamtalk.com/forum/threads/extract-data-from-garmin-66w.40407/

 

[Gman]

Thanks for the link but sadly it doesn't really help. My problem is not with getting the data out, it is with the way (I think) that the file holding it is created.

 

[jokiin]

don't know, most people just use Dashcam Viewer now

 

[Gman]

Dashcam Viewer doesn't work for me. There is a noticeable pause between videos (probably something like 2 or 3 black frames) and I don't feel like paying for something that doesn't work (particularly when there is a nice solution that does work).

I've just tried creating the .gpx files using Perl and then modifying the Java script so that it just clears the existing files rather than deleting them. Then as each video file is processed it just appends to the existing file. Ie my plan was to use the file container created in perl and add the data using Java. Unfortunately that didn't work but it might point to the fact that there is some data issue with the Java version rather than a file structure one which is worth exploring.

Oh well, onwards and ...

 

[Gman]

SOLVED IT!! - and it turns out that it was a coding error after all

For anyone coming across this later, the problem was the construction of the DateTime group. The correct (Perl) format is:- 2019-09-27T09:35:03Z whereas the incorrect (Java) format that I had was:- 2019-09-27T:09:35:03Z. Note the additional colon after the T, that is what was wrong.

What confused me was that both versions worked when only a single video file was selected but the Java version failed when multiple videos were selected. In my mind that pointed to a problem of concatenation (which was true enough), but of the files rather than the data within the files.