Why does VIRB Edit Export at Half Original Bit Rate?

Lesser levels of quality create smaller output files and accomplish the feat in a shorter amount of time - the output file is also smaller.

I believe I found my answer in another post where I learned the VIRB records at a constant bit rate (CBR) and VIRB Edit outputs at variable bit rate (VBR). I ran a test and confirmed that when viewed side by side, although there is a slight color difference, the 80 Mbps CBR original was comparable in quality to the 40 - 50 Mbps VBR file post edit.

Thanks.

I disagree that 40 Mbps is sufficient for export of 4K, much less the 5.7K Virb Edit is now capable of. Please add the ability for the user to determine the output bitrate or at least offer more options for better quality. during export. 4K video really needs 100 Mbps or more. This is really limiting an otherwise great product from producing high quality results!

I second some of the above comments regarding the low export bit rate. My 100 Mbps 4K source video exports at 40 Mbps at the maximum quality settings! The drop in picture quality is substantial and immediately obvious when viewing the rendered video. VIRB Edit is a quality product, but this flaw makes it all but useless for 4K editing.

Hi.

The way I was positively surprised with camera itself, the same, but negatively I'm disappointed with Virb Edit output options of encoded mp4 files. First of all I'm 100% supporting idea of letting me to choose bitrate, not only go from frop down list (this list is only for newibes with selection like medium or high quality...). But there is more. I see that used encoder (ffmpeg or that second one) is not utilizing computer power. I have 12 cores machine and I see it only uses 1 core. It's strange because ffmpeg is having built-in x264 codec, which is very good, has enormous numer or parameters you can change, but most of all is multi-thread and can saturate all available computer power.

So far options of encoder output are only for amateours. The output quality of video from camera doesn't make much sense if it's gonna be damaged by editing software.

By occasion: I have also Sony AV200 camera, wich comes with Play Memories Home (which is kind of software for amateuors), it also has miserable options of encoding video output, and it's dramatically slow, uses only 1-core and compresses output 10-times longer than source is with i7 3.5GHz core.

Let's see if somebody from Garmin comment on this.

Hello matroX,
For reasons I have not been able to discover, some machines do not utilize their full multithreading potential when exporting video. This appears to be quite rare, I've never actually witnessed a machine do this, but I have heard irritated reports from others on the forums. One of our testers has a 32 core AMD thread ripper and export used all 32 cores at 100%. The "ffmpeg" encoder option does indeed utilize x264 (unless you have a semi recent Nvidia card, then it will use ffmpeg's nvenc implementation). This is something I would love to fix if I can figure out how.

The encoder options are indeed for amateurs, which is by design. VirbEdit is not intended to be a pro or even semi pro software suite. Its target market is people who know almost nothing about video editing, and just want to quickly put together a video. It would be fairly straight forward to have a UI like Handbrake where we let you pick your rate control, specific bit rate or CRF, tune and presets etc. But there is really no desire to do so because that many options will overwhelm all but a small percentage of our users.

As for the quality of the output video, thats a subject on the forefront of everyone's mind, and we regularly make changes to try to improve things. There are a number of factors at play there. One of the most unfortunate is that when you add gauges to a video the over all encoding quality drops pretty significantly. We've found that Youtube suffers from the same problem. When you upload to Youtube, when it re encodes, the same video will end up looking much worse with gauges than the same video without them.

Hello matroX,
For reasons I have not been able to discover, some machines do not utilize their full multithreading potential when exporting video. This appears to be quite rare, I've never actually witnessed a machine do this, but I have heard irritated reports from others on the forums. One of our testers has a 32 core AMD thread ripper and export used all 32 cores at 100%. The "ffmpeg" encoder option does indeed utilize x264 (unless you have a semi recent Nvidia card, then it will use ffmpeg's nvenc implementation). This is something I would love to fix if I can figure out how.

The encoder options are indeed for amateurs, which is by design. VirbEdit is not intended to be a pro or even semi pro software suite. Its target market is people who know almost nothing about video editing, and just want to quickly put together a video. It would be fairly straight forward to have a UI like Handbrake where we let you pick your rate control, specific bit rate or CRF, tune and presets etc. But there is really no desire to do so because that many options will overwhelm all but a small percentage of our users.

As for the quality of the output video, thats a subject on the forefront of everyone's mind, and we regularly make changes to try to improve things. There are a number of factors at play there. One of the most unfortunate is that when you add gauges to a video the over all encoding quality drops pretty significantly. We've found that Youtube suffers from the same problem. When you upload to Youtube, when it re encodes, the same video will end up looking much worse with gauges than the same video without them.

So I guess the solution right now for people who use Nvidia is to use a previous version like 4.2.3? I've been testing and I can export at a bit rate of 100Mbps by adjusting the target quality. Only problem is a video that takes 45 minutes to export it 5.3 takes 5 hrs to export in 4.2.3.

When you upload to Youtube, when it re encodes, the same video will end up looking much worse with gauges than the same video without them.

Nice to know, that you are aware of that! I didn´t know that it was related to the gauges, but I experienced the same difference between local and uploaded videos.

Hopefully you´ll find a way to fix that. Much appreciated :)

@bramezan
Modern graphics cards have special modules built in that do video encoding, and they do it much much faster than CPU driven solutions like x264. If you have an Nvidia card, you aren't using all your CPU cores because you are using your graphics card instead. In 4.2.3, when you selected ffmpeg as your encoder, it was always using x264 regardless of your video card. In 5.3, we use your GPU if possible. Hence the speed difference in the two versions. Have you observed a quality difference between the two versions?

@HerrRiebmann I strongly suspect that this quality drop is due to how H264 does its compression. If that is the case, it is not something that can be fixed. One day I'll have to play with HEVC and see if videos look better with a different compression standard.