bitrate settings

[jauling]

By default, the git2 is set for bit rate of medium.

Do any of you guys always record at high bitrate? do you see a noticeable difference in quality? Is it much less compression artifacts? Is it worth it given the increased GB/hr?

I recall trying high bitrate a few months ago, but i didnt really do any scientific tests to determine if its worth it or not (compared to medium). I've got a pretty crappy desktop pc at home, which probably explains why it was choking really hard on playing these high bitrate files, so it was really tough to judge.

FWIW/FYI: I found a youtube video of a guy who went completely OCD and tested every single resolution and bitrate settings on the git2, resulting in this great spreadsheet. This video was posted on 2016-07-05, so I'm guessing he was using either v1.5 or v1.4 firmware (not sure if that matters?):

 

[Nigel]

The difference between medium and high isn't obvious unless you have a large screen to watch it on, but if you are going to make contrast/gamma/colour changes in a video editor then it is worth having the high bitrate to work on, even if you are going to output a much lower bitrate from the editor. Compression artefacts always become more visible when you make adjustments.

I normally leave mine set to high, only reason I would reduce it would be if I needed to fit more than 4 hours of video on my 64GB memory card. The video I keep after editing is often a lot smaller, and the video I use for playing on low performance devices and upload to YouTube is smaller again.

I think that table is a bit out of date for some of the values now.

 

[Jantteri]

As you can see from the spreadsheet Git2's bitrate is capped around 38Mbps, I guess that's the limit of Novatek's chipset. When there is a lot going on the video, that is to say there is a lot of information which is needed to compress to meet the highest available bitrate limit, you will see clearly all the compression artifacts especially if you pause the video. For instance driving in a forest will yield to high information input because the scenery is full of changing details and is demanding task to compress which means the more information has to be disregarded the lower the bitrate is, as compared to a simple, slowly changing scenery with little details.

Uploading the video to Youtube results even worse artifacts as for instance 1440p30 is capped around 10Mbps with H.264 codec for standard uploaders accumulating the quality loss. Because video compression demands much prosessing power video cameras use simpler compression algorithms on the fly and it will result poorer quality to bitrate ratio than more sophisticated CPU intensive desktop algorithms together with proper post-processing. As a result the higher the original bitrate is the better image quality you can maintain when compressed to a same file size, especially when the video includes intensive levels of variable details (= high level of information input from frame to frame).

But of course bitrate is just one factor together with the image resolution and the quality of optics and sensor as well.

 

[Jantteri]

Many riders wants to move to GoPro after watching the final YT video quality the Git2 produces.

Good starting camera, you can learn a lot and then jump to a high level camera.

I you want the best action camera on the market at the moment image quality wise I would suggest to take a look on Sony FDR-X3000R (4K@100Mbps & OIS), and not that much more expensive than Hero 5.

 

[Jantteri]

I'm getting the Hero5 as christmas self-gift 🙂...still not 100% sure because of the audio adapter, but thinking about selling it in the future is a better business than a Hero4. Hero5 is cheaper than Hero4 here.

Not a bad choice, you get very versatile camera and 60Mbps which is more than good still. However you might want to reconsider if you watch some reviews on Sony in Youtube 😉

 

[Ajusshi]

From a motorcycle rider POV, i won't recommend the Git2 if you want good quality video. It's just not at the level...for motorcycle dash cam purposes it's OK.

Detail is poor the more you accelerate (not even high speeds), highly detailed scenes look choppy even at low speeds the more you look further, artifacts appear at the minimum presence of shade pixeling the image (dark parts of the bike's dash or if you have a smoke/tinted windshield).

Exposure transitions are frenzy, not accurate/crazy/jumpy (a well discussed and proved problem)...if you go "fast" it's just a mess of blue circles in the sky.

WB is inaccurate at night wich can be forgiven, but is crazy inaccurate in daylight and not fixed using a preset. Use "sunny" and you will get orange tint everywhere, this will make you want to come back to AWB forcing you to have that blue tint in all of your videos. If you go from a sunny iluminated part of the road to a shaded part, the WB goes from warm to cold looking awful.

I'd recommend the Git2 to selfie people, people that makes stationary video, people that walk or as maximum use bicycle. Also to those who like timelapse from camera and raw photo at a low price.

For faster activities or if the quality is what moves you, use better cameras that look smoother, expose better and keep the detail when the movement is fast.

Regarding to bitrate, rise it if the video is important/going to recompress, if not use medium/low.

(High bitrate It may help reduce those mentioned definition defects)

These are my conclusions after months of use under different FW versions. Also what i have noticed on many motorcycle YT videos.

Many riders wants to move to GoPro after watching the final YT video quality the Git2 produces.

Good starting camera, you can learn a lot and then jump to a high level camera.

better cameras like gopro or what?

 

[tmvideo]

100 mbps can looks worst than 35 mbps. The quality of h.264 codec is also depends of I-frames frequency.
So, as I understood, the all I-frames 35 mbps coding will produce better action quality than the 100 mbps with the default IPB GOP-structure.

 

[Jantteri]

100 mbps can looks worst than 35 mbps. The quality of h.264 codec is also depends of I-frames frequency.
So, as I understood, the all I-frames 35 mbps coding will produce better action quality than the 100 mbps with the default IPB GOP-structure.

Of course the compression quality depends on the used method but there's no way you can substitute processing power with any method. The fact is that inherently small sized video cameras cannot never compete with raw processing power of desktop or cloud based computers simply because of laws of physics. And higher bitrate with same or comparable method will always lead to better quality, it's not really any rocket science in that regard.

However, finding the "best" method is a really complex task and is compromise between different factors and you will never find one ideal to meet every kind of source material.

Btw, as I have understood Sony uses here H.264 XAVC with highest 5.2 profile in 100Mbps mode. Not sure about Git2's Novatek codec but I would be a bit doubtful it could beat XAVC even at the same bitrate.