To followup with Hippo-Drones video crediting me with the theory, I made a video showing a bit more of the process. I made this just for small YouTube creators so I may not be releasing the instructional video public anytime soon. Additionally, I loaded the end result videos to YouTube each on their own in order to reduce variables.
Overall process video with reference to bit rate for all the resolutions YouTube likes (links to examples of the final footage in YouTube descriptions and below):
4k version of the 1080p60 test footage (let it load in YouTube itself before hitting play)*:
1080p60 as YouTube lets it be (again let it load):
Thank you again to @HippoDrone and Superballs Supervids. I also want to see what @scooterwuf can make of the process and results.
*Playing through a website can offer another variable so the examples are best viewed directly through YouTube. As mentioned in the video, other variables will be the viewer's device and their internet connection.
It does seem like YouTube throttle some vids on top of this, maybe if lots of people are trying to view the video at the same time or something? YouTube always seems to me to look better on Chrome than on ie too
It does seem like YouTube throttle some vids on top of this, maybe if lots of people are trying to view the video at the same time or something? YouTube always seems to me to look better on Chrome than on ie too
It does seem like YouTube throttle some vids on top of this, maybe if lots of people are trying to view the video at the same time or something? YouTube always seems to me to look better on Chrome than on ie too
Just occurred to me that while we can get the bigger bitrate, YouTube may not give all channels the same "width" for a pathway. So a smaller channel may not have as much traffic therefore the path is not as wide.
I recall that even a moderate sized channel can go viral and YouTube may need to adjust for the traffic.
I know that even using the process, I can see a difference as well as some members of the audience. Wait for Thursday when I release my small motovlogger for August video. I went a different route on res for that one.
This was very helpful. Not sure if it were things from my end, but I wasn't sure if I was seeing all the technical readouts you were mentioning on the screen regarding the rendering stats (this was for the first video). I'll watch the video again just to be sure.
Both versions (4K vs. 1080p) are pretty darn close. I'm not seeming much to make one markedly different. Blacks are rich, foreground detail on your dash is sharp. Colors are pretty close between the two versions with some softness on far sides (high bitrate detail) on the grass, trees and concrete along the waysides of the road. The 4K version may have a slightly brighter exposure to it that seemed to make it pop just a bit more. I think I'm seeing this as a slightly less level of gray in the overhead sky versus the 1080p version.
I'm willing to try some experiments on my own, but to see if there are any advantages for less than ideal shooting conditions -- low light in the morning, or overcast from a threat of rain.
This was very helpful. Not sure if it were things from my end, but I wasn't sure if I was seeing all the technical readouts you were mentioning on the screen regarding the rendering stats (this was for the first video). I'll watch the video again just to be sure.
Both versions (4K vs. 1080p) are pretty darn close. I'm not seeming much to make one markedly different. Blacks are rich, foreground detail on your dash is sharp. Colors are pretty close between the two versions with some softness on far sides (high bitrate detail) on the grass, trees and concrete along the waysides of the road. The 4K version may have a slightly brighter exposure to it that seemed to make it pop just a bit more. I think I'm seeing this as a slightly less level of gray in the overhead sky versus the 1080p version.
I'm willing to try some experiments on my own, but to see if there are any advantages for less than ideal shooting conditions -- low light in the morning, or overcast from a threat of rain.
Awesome about you testing it out too. Nothing like some peer review.
I have found if there are overcast skies, there is not a lot to be gained by processing up to 4k. Poor lighting just makes poor filming since details are lost on the capture anyway.
Just my experience with the GoPro Hero 4 Silver, you have different gear so you might have a different experience.
I'm looking at it on my office computer - iMac 21.5 inch machine, 1.4 GHz Intel Core i5 processor. Ethernet connection via the university where I teach.
Hope it's OK I'm bringing this up, but I would like to take a more technical approach for this matter.
consensuses:
The higher the bitrate of the video, the better
The higher resolution* and frame rate, the better
codec we prefer in ascending order: H264 (AVC1) -> VP9 -> AV1
Just so we are clear about what is what:
1920x1080p = FHD
2560x1440p = 2K
3840x2160p = 4K
7680x4320p = 8K
Test method:
In order the help us compare the quality without being subjected to personal preference and eyesight
in each video, just right click it the then go to "Stats for nerds"
The baseline for all videos is AVC1.
From romers, if your average video gets about 500-1000 views, you videos will be re-encoded in VP9 instead of AVC1.
(as far as I can see, I qualify and as a result my videos are re-encoded in VP9 automatically, no matter the original resolution)
If you don't have enough views, another way to "force" YouTube into re-encoding with VP9 is by uploading in 1440P instead of 1080P
In the past, if you would use the YouTube editor, enhance a video, and just trim 1 sec of the end of it just as you finish upload it to YouTube servers, It will be re-encoded in VP9 instead of AVC1.
This method no longer works.
AVC1 codec at 1080P60FPS:
VP9 codec at 1080P:
VP9 codec at 1440P:
What are the limitations of AVC1?
easy, see the YouTube official page: Link
Here we get out first hint about VP9 option, but more on that later.
With this info, we can determine how to set out render options.
for example, if I upload a 1080P video at 30 FPS with 30 Mbps, it would be a total waste! as YouTube will just re-encode it without our precious bitrate data.
What are the limitations of VP9?
unlike previously mentioned, VP9 isn't used just for 4K, but also in 2K Case Study: YouTube and HTML5 Video (page 2)
so, what are the limitations?
well, YouTube doesn't tell us....
But Google does!
VP9 is an open and royalty-free video coding format developed by Google, and as such, we can read more about it in here: https://developers.google.com/media/vp9
here is a snip of info about bitrate in VP9:
So what does it tells us?
Well, a lot!
with a bit of MS-Excel power, it will look like this:
* all number are in Megabyte per second, not Megabit per second.
So what do we do with this info?
well, not much
even if you get to be re-encoded in VP9, you still get re-encoded.
(so don't even bother to upload in VP9)
What is wrong with upscaling/interpolation?
when you render something at higher FPS/Res/Bitrate then what it was shot, you simply ask your editing software to add more data
but..... how?
well, there are basically 2 options:
1. replacte: more common in FPS, when you have a 30 FPS footage and want to upscale it to 60 FPS, just display every frame twice, and you got a 60 FPS video.
The problem with this is obvious, it does nothing to help out video look better. at all.
2. Guess/calculate: there are many interpolation algorithm, but the base logic is: I know what I have, let's try to guess using math what the added data would look like.
Sometime it's good, sometime not so much.
What's so great about VP9 anyway?
it's the future!
it weight less, compression is better
and the quality is better!
VP9, because less IS more!
downside of VP9, it's more intensive on the client than AVC1.
so slow/old PC/Phones will have hard time decode it.
What about AV1?
AV1 is everything good with VP9, now better!
YouTube is experimenting with it,
if you want to take in on the action, on your desktop
use Chrome V70 and above
go to: chrome://flags/#enable-av1-decoder and set is to Enabled
TLDR:
Right click on your video, click on "stats for nerds", if you see VP9 right now, you are golden.
if you see AVC1, maybe it's a good idea to upscale to 1440P to enjoy the benefit of VP9.
in both case, personally I don't see any advantage to render with setting above the overhead of the codec setting.
Just render as best as you can with the material you got.
@SM26 great research. Thanks for using my topic video as an example!
What is neat is how the example video for the 4k rendering in that video's description shows the vp09 and they both have about the same number of views.
I love watching the buffer health in the stats for nerds.
@SM26 great research. Thanks for using my topic video as an example!
What is neat is how the example video for the 4k rendering in that video's description shows the vp09 and they both have about the same number of views.
I love watching the buffer health in the stats for nerds.
@R-Rated
from what I can see, your 1080P videos get re-encoded in AVC1.
so for you, It make sense to render at 1440P, So you'll get the VP9 codec.
But I don't see a benefit to render it in 4K.
If I were you, I'll choose my render setting based on the Bitrate from the camera.
Personally, my main camera is Sena 10C, which output 1080P@30FPS with ~12 Megabyte per second.
So in my case, YouTube will always re-encode it to VP9 to save bandwidth,
I would use 1440P render setting only if the original footage has ~25 MB/s.
My hypothesis says that any less than this, will be indistinguishable in the output.
but I might be wrong.
@SM26 and @HippoDrone the Hero4 silver at 1080p60 has 29 Mb/s. At least that is what the data is showing on a couple of short clips on cloudy days when I pull it up the info via GoPro Quik.
From the chart from YouTube, it would appear that even at 1440, it loses 5 Mbps.
Still all of this is great info to know especially the buffer health. That would be on the viewer. Which is why we cannot always predict what the audience will experience.
We have a few hidden forums and features that you can't see until you login or sign up. We only send about 4 emails a year and never sell anything. We're a global community for bikers who like to record videos. We run this site for fun
This site uses cookies to help personalise content, tailor your experience and to keep you logged in if you register.
By continuing to use this site, you are consenting to our use of cookies.
