09-21-2020 01:38 AM - edited 09-21-2020 01:39 AM
I have briefly been testing the telemetry capabilities of the HERO9. The first difference I see with the HERO8 is the addition of two metadata streams: MRV Frame Skip and LRV Frame Skip.
On the GPS front, my unit seems slightly worse than my Hero 8, in terms of accuracy and reliabiliy. The weather was overcast so maybe that played a role, but I've had many instances where the GPS path went nuts. It seems to acquire a lock (bright white GPS icon) much more easily than the 8, they might have changed some thresholds there.
Has anyone had different results? It might just be my unit, which also has the widespread touchscreen unresponsiveness issue, so I might need to have it replaced.
09-21-2020 11:18 AM - edited 09-29-2020 11:53 AM
Ok. I just went for a drive to test the H9 GPS against the H8. Things are not looking good, at least for my H9 unit. I really hope I got a defective one, or that this is solveable via firmware.
Conditions were pretty bad: overcast and inside a city. I had the cameras mounted on the gas tank of my motorcycle, so I was probably covering some of the satellites with my body. Still, the H8 produced pretty decent results while the 9 is completely unusable. And, even worse, it reports good fix and precision values even when the location is dozens of kilometers away.
Samples with worse than 2D fix or 625 Precision have been filtered out.
Clip one (4k 25P linear hypersmooth boost on both cameras, H9 has horizon levelling on)
Hero8 (some errors but close)
Hero9 (miles away)
Clip 2 (same settings)
Hero9 (just registered the initial location)
Clip 3 (same settings)
Hero9 (at least it's in the same neighborhood. Close but bad. Elevation freaked out)
In some of my tests, the H9 seems to get the first position right and then go nuts. So I thought maybe the processor was unable to process the horizontal levelling, hypersmooth, high resolution and the GPS data on top of that. I decided to use modest settings to give the processor a rest.
Clip 4 (1080 25P Wide Hypersmooth off)
Hero 9 (first and last positions seem ok. The rest is a visit to the ocean)
Clip 5 (same settings)
Hero 8 (good)
Hero9 (good first 100 meters, then nuts)
Clip 6 (same settings)
Hero9 (probably decent first and last positions, the rest is unusable)
I was honestly not expecting the Hero8 to perform this well on such bad conditions. When I do these tests I also record the data with a GPS smartwatch as a benchmark, which tends to get coordinates always right, except on tunnels (props to them, I don't think they use cellular data, just GPS+GLONASS, so the same resources as the GoPro). In this case the Hero8 performed close to the watch, so I have excluded the watch from the analysis. I would have expected some of the glitches present in the H9 (clips 3 and 5), but the trend is very hard to ignore. Either my unit is broken or there is a serious firmware/hardware issue with the Hero9 GPS.
I would appreciate it if someone else shared their experiences, or even their files. You can create screenshots like these with https://goprotelemetryextractor.com/free/
on 09-29-2020 06:05 AM
I'll happily share my files with you. I have a double mount so I have been using the Hero 8 and Hero 9 side by side. I mentioned in another post that the GoPro Hero 8 GPS icon will light up and stay steady. The GoPro Hero 9 GPS icon will light up, go dark, light up, go dark as if it's acquiring a signal and then losing it again.
One thing I did read was that someone had GPS data on the first chapter of his recording but none on subsequent chapters. He found that disabling hypersmooth meant the rest of the chapters would have GPS data retained on them. I'm hoping this points to a simple firmware error with the GPS but so far, the Hero 9 is proving to be terrible when it comes to GPS data.
on 09-29-2020 11:29 AM
My Hero 9 GPS peformance is near identical to Irishmanpdx, and very disappointing considering that I have a Hero 8 that has no GPS Issues. I did do a test after doing a manual firmware reload, via the SD card, and doing Factory Reset. I took the Hero 9 out and just walked a mile without recording, but just watching the GPS indicator. For the entire walk the GPS indicator was solid white, which was really good to see! I than did the same walk with the camera recording. When I got back to the office I loaded the Hero file and ran Telemetry Extractor on it to analyze the data. While the 1080p 60 video looked just fine, the GPS was all over the place and not useable.
I'm trying to be patient to see what the Gopro support folks determine. I have about two weeks to play with before I have to send it back for a refund if this can't be resolved.
on 09-29-2020 11:51 AM
Thanks for your report @buckaroo60 This looks consistent with my experience (quick GPS lock, very bad GPS data).
Do you recall if you had the mounting fingers open or closed while recording? It's a very wild guess, but the touchscreen problem seems to have correlation with whether the camera is mounted or hand-held, so I was wondering if this could have a similar cause (magnetic disturbances?). Some people are showing good results with their Hero9 GPS data, which suggests the problem is only present in some faulty units (seeing other bad reports in the forum and on Facebook. Considering how niche the use of GPS data is, this seems quite widespread) or it only happens under certain conditions.
@irishmanpdx Would be interesting to have a look at your results too, especially that side by side comparison with the 8. Feel free to send them to my email or grab screenshots from the Map View in https://goprotelemetryextractor.com/free/
GoPro is looking into this. Obviously what we all want to know is if this is a hardware or a firmware problem.
on 09-29-2020 12:47 PM
When I did the test I had the Hero 9 camera mounted on a short hand support. The camera was not held at all in my hands, and was kept upright during the test. One thing that I forgot to mention, is I didn't see the same flickering on and off of the GPS indicator like I did earlier. For this test it stayed a solid white color, but unfortunately the data points are way off. I'm wondering if the Factory Reset that I did prior to this 2nd test did something to stabilize the signal lock. Note, the sky was overcast and not clear.
To address your question about electrical noise I just walked a similar route with my Hero 6 which I happen to have with me. As you can see in the attached graphic, after using your Telemetry Extractor, it did a pretty good darn job of tracking being a 6th generation unit! What's nice about the Hero 6 is the GPS strength/lock indicator is on the main front screen, so I can actually watch it as I'm recording. That's not the case for the Hero 9, where the indicator is located on the Preference screen.
What do you think?
09-29-2020 01:36 PM - edited 09-29-2020 01:40 PM
So my hypothesis was actually the opposite: that opening the fingers and mounting the camera somewhere could produce bad results, whereas hand holding the camera with the mounting fingers closed would record good GPS data. It's probably bogus, but I have seen some good hand-held samples and bad mounted ones. Probably a coincidence.
Other than that. These tests and comparisons are probably good data for the GoPro team working on this
PS: Factory resets and firmware reinstalls very rarely make any change. I know the support teams always recommends them as a first measure, but that's probably so the user has a default setup before they start debugging, or a last resort when support does not know what to suggest. Very rarely a solution
09-29-2020 01:48 PM - edited 09-29-2020 01:49 PM
I originally thought that maybe the internal processor couldn't handle all the tasks at once 4K 30, horizon leveling and GPS tracking, so I reduced the video to just 1080p 60fps to reduce the processor load. Unfortunately, I got the same results. Juan, you indicated that some Hero 9's are tracking just fine. If everyone is using the lastest firmware, it would definitely point to a hardware issue.