VIRB Edit G-Metrix Gradient not matching Edge 1000 fit-file data

I?ve made a video in VIRB Edit (v5.1.3) with clips from my VIRB Ultra 30 and imported G-Metrix data from an Edge 1000 fit-file on my computer. I also synchronized each clip. All the meters seem accurate except for the gradient values.
When I compare the data from the fit-file with the G-Metrix meters in VIRB Edit it looks like VIRB Edit does some processing witch gives inaccurate results.
I used GoldenCheetah to read the Edge 1000 fit-file and matched the distance and height fields with the distance and height G-Metrix meters in my video. Although the height/distance data at different times match up, the gradient values don?t.
In general the G-Metrix Gradient is much lower than the (original) data in the fit-file.

I hope this can be fixed soon so I can finish my video and show my friends how steep those mountain passes in Italy and Austria, I?ve climbed by bike, really are.

0 schinder over 7 years ago

I wouldn't trust the Edge, either, since it's just estimating from the altimeter and the distance. My 1000 seems to lag a bit (not unexpected given how it's measuring), and I don't really trust the numbers it displays. Unless you've gone out and measured the incline with the proper equipment, then you've no idea how steep they "really" are. For popular mountain passes, though, I'll bet you can find out pretty easily with a web search, since somebody (whoever maintains them) probably has measured them.

0 LEAVS over 7 years ago

The point is that I want to display the actual gradient at the point where I?m riding during my video.
The gradient data in the Edge 1000 fit file looks OK.
The G-Metrix Gradient value based on that fit-file shows much lower values.
I think the processing of the fit-file gradient data by VIRB Edit has a bug.
I would like that bug to be fixed.

I was hoping that others would confirm they experience the same problem, or ? could tell me what I?m doing wrong.

0 schinder over 7 years ago

So I checked today's fit file from my 1000, and there's nothing that looks like the gradient in it. (My fit file reader is old, though, so may not be showing me everything.) So Golden Cheetah must be computing the gradient from the fit (the source is available, so it'd be easy enough to find out if you were determined enough), and VIRB Edit must be doing the same thing, only differently.

If you look through the archives of this group, you'll see that the grade is something that's complained about often. In the past I've seen many problems, like the gradient oscillating as I climb (usually from zero to something bigger than correct) These forums are unusual in that the developers actually read them (in most other forums Garmin people aren't active), and I think they've indicated that they're still working on it. So hang on to your files, because things may improve.

I looked at whatever videos I have hanging around in VIRB Edit. The only on I have that uses a fit file from my Edge 1000 can be found at https://youtu.be/HgKhFBn_aUw This was produced by the then current OS X VIRB Edit. The camera is a VIRB Elite, and the gradient looks reasonable, but since it's a time lapse, that may be masking problems. The only other video I have in VIRB Edit at the moment is from my VIRB 360, and it's much too big to throw onto YouTube. The grade again looks reasonable, but a 360 has a barometric altimeter, an accelerometer, and a gyroscope, so it may well be able to measure gradient directly. I don't seem to have any GMetrix files from the 360 (or they're well hidden), so I can't look to see what's in them. I do have a few clips from the 360 on YouTube (https://youtu.be/onOimyRCYaU for one), and the gradient looks reasonable. (Whether they're right or not is another story.)

I agree, though, that if VIRB Edit is doing smoothing, it should be possible for us to request simple unsmoothed gradient calculations like the 1000 itself does.

0 LEAVS over 7 years ago

Thanks for your reply schinder.

So I checked today's fit file from my 1000, and there's nothing that looks like the gradient in it. (My fit file reader is old, though, so may not be showing me everything.) So Golden Cheetah must be computing the gradient from the fit (the source is available, so it'd be easy enough to find out if you were determined enough), and VIRB Edit must be doing the same thing, only differently.

You are right. Both programs do their own math to calculate gradient/slope. But the values of VIRB Edit seem wrong where other programs calculations of my activity are closer to the "mountain profiles". Besides Golden Cheetah I also compared the data in VeloViewer and Garmin BaseCamp.

If you look through the archives of this group, you'll see that the grade is something that's complained about often. In the past I've seen many problems, like the gradient oscillating as I climb (usually from zero to something bigger than correct) These forums are unusual in that the developers actually read them (in most other forums Garmin people aren't active), and I think they've indicated that they're still working on it. So hang on to your files, because things may improve.

The grade values from my VIRB Ultra 30 are all over the place (even over 120%) but the Edge 1000 (calculated) values are much better. I hope Garmin will come with an update soon.

I looked at whatever videos I have hanging around in VIRB Edit. The only on I have that uses a fit file from my Edge 1000 can be found at https://youtu.be/HgKhFBn_aUw This was produced by the then current OS X VIRB Edit. The camera is a VIRB Elite, and the gradient looks reasonable, but since it's a time lapse, that may be masking problems. The only other video I have in VIRB Edit at the moment is from my VIRB 360, and it's much too big to throw onto YouTube. The grade again looks reasonable, but a 360 has a barometric altimeter, an accelerometer, and a gyroscope, so it may well be able to measure gradient directly. I don't seem to have any GMetrix files from the 360 (or they're well hidden), so I can't look to see what's in them. I do have a few clips from the 360 on YouTube (https://youtu.be/onOimyRCYaU for one), and the gradient looks reasonable. (Whether they're right or not is another story.)

Thanks for your time and looking in too this. I want to post my own examples but haven't had the time to put it all together. When I have I'll post it here.

I agree, though, that if VIRB Edit is doing smoothing, it should be possible for us to request simple unsmoothed gradient calculations like the 1000 itself does.