Hey guys Roman here from TechGuides and in
today's video I wanted to give you an update on my recommendations what the best bitrate is to render
your YouTube videos at to get the highest possible quality on YouTube at 1080p, 1440p, and 4K. Now
2 years ago I have already produced a video which is very similar to the one that I'm doing today,
however back then I didn't have access to any 4K footage - which I'm going to rectify today - and
I also didn't really properly test the quality in such a rigorous way that I'm doing today. So
in today's video we're really going to stuck our teeth into what the best possible bit rate is
to get the highest quality on YouTube regardless of the resolution that you're producing in. But
just before I jump right into the data I wanted to mention that this video is primarily targeted
at people that are producing content for YouTube that is mostly gameplay content. So the testing
footage consists of a one minute clip of Call of Duty Modern Warfare Warzone 2 recorded at 60
FPS. So if you're producing videos at only 30 FPS and that contain a lot of static images,
like the shot you're watching right now, then the recommendations in this video are probably
a bit on the high end and you probably want to about half the bitrate recommendations that I'm
providing in this video. And finally I'm not testing for CQP values. This would be an entirely
different video and it's kind of also not relevant because most people are likely using software like
DaVinci Resolve, which only allows you to render at a constant bitrate and therefore this is what
I tested for this video. One last mention is a big thanks to this community member for providing
a 4K footage of Call of Duty Warzone 2 for me to perform these tests with. And with that let's jump
right into results. Let's start off with 1080p and just briefly let me explain this graph: On the
x-axis or the horizontal you can see the render bitrate and on the y-axis or from the bottom
to the top you can see a quality metric which essentially is a measure of how good a distorted
video sample - such in this case from YouTube - is when compared to the raw original clip. Now
one last data point that I have on this graph is the raw data point which essentially equates to
directly uploading the raw recording directly from OBS onto YouTube. As you can see it is generally
a very good idea to actually re-encode your videos for YouTube instead of directly uploading a raw
recording because usually the raw recording has a very high bitrate. Moreover we can see that the
results are actually kind of a little bit jumpy and I guess the sort of plateau can be reached
after roughly 50 megabits per second. However, one big cevat when uploading videos at 1080p to
YouTube is the fact that these videos are not re-encoded in VP9 and instead are re-encoded in
AVC or h264. Now this is quite important, because VP9 is just vastly superior in visual quality -
and actually also produces much smaller file sizes for YouTube then AVC - and therefore you should
always make sure to make YouTube re-encode your videos in VP9. However, the only reliable way
to make YouTube re-encode your videos in VP9 is to actually upload your videos not in 1080p, but
at a higher resolution so either 1440p or 4K. So what you can see here is the video quality that
results when upscaling your 1080p footage to 2K in orange or 4K in yellow and then assessing the
video quality of the re-downscaled 1080p version from YouTube. And as you can see both of the
upscaled 2K and 4K versions vastly outperform form the 1080p version uploaded to YouTube in terms of
visual quality even after downloading the 1080p version back from YouTube. VP9 is actually so much
more efficient, that when we look at the bitrates that these videos have, we can see that they're
about a third lower than the 1080p version, however the visual quality is so much better.
Here's a side by side by side comparison of the 1080p, 1440p, and 4K version all at the same
render bitrate of 60 megabits per second. So my recommendation at 1080p would be to use a render
bitrate of 50 megabits per second. However, I would highly recommend anybody to actually
upscale your videos to 1440p, add a little bit of sharpener, and then maybe render that video
at about 60 megabits per second - if your upload speed can handle that and if you have enough
space on your system of course - or what I would recommend in the next segment which is going to
cover the native 1440p version. Speaking of which, here you can see the improvement in visual quality
when increasing the render bitrate for a native 1440p video. And as you can see from the blue
line this kind of plateaus at around 80 megabits per second where it's almost indistinguishable
from the raw video, which I actually recorded at the lossless setting in OBS which resulted in an
overall bit rate of 815 megabits per second - so that would be all the way to the right there on
the graph. Finally, I obviously also have to show you what the visual quality of a 4K upscaled video
would look like - if we download the 1440p version that is - and as you can see from the orange line
this completely blows the blue one out of the water. Once again it's clear to see that it's very
beneficial to actually upscale even a 1440p based footage onto 4K, in order to get higher quality
on YouTube. Now if we look at the YouTube bitrates that are assigned to a YouTube video that is
uploaded either in 1440p or upscaled 4K - and then using the 1440p version of that upscaled version
- you can see that the bitrates are somewhat all around the place and there's not really much
of a pattern here. I guess the upscaled version has ever so slightly higher YouTube bitrates
but generally I think it's true that YouTube just assigned the same target bitrate to a certain
re-encode of a video and will just be able to take more information from the 4K video that you've
uploaded to YouTube instead of the native 1440p version. So at 1440p my recommendation would be
to render your videos at 80 megabit per second, either directly in 1440p, or upscale them to
4K, add a little bit of sharpener and benefit from the much higher visual quality even on
your 1440p version on YouTube. And finally, let's move on to the 4K portion of this video
- or 2160p - and the quality of these types of videos you can see on screen right now. Now
here the line is a little bit more slanted, it's not so straight as it was for 1440p. And
the raw point was actually slightly above the highest possible bitrate that I tried for
this video at 220 megabits per second. But still you can see that ever higher bitrates
actually produce significantly and measurably better visual quality on YouTube because the
YouTube re-encode system has more information to work with. So the age old myth of people
uploading too high bitrate videos to YouTube, that then leads to actually bad visual quality,
is is completely busted here. Now as for a recommendation on what I would use in terms of the
render bitrate is kind of a bit of a tough one. I'd say that the curve becomes more horizontal
at around 120 megabits per second. However, I understand that this is a pretty high bitrate
and people are likely not willing to both store such large file sizes and also maybe don't have
the time to upload such large videos to YouTube. So I guess my recommendation would be to target
between 100 and 120 mebgabit per second for 4K footage. So in summarizing, if you have 1080p
footage then I would highly recommend to up-scale that to 1440p and render at about 60 megabits per
second. If you have native 1440p footage then I would recommend to upscale it to 4K and render
that at about 80 megabits per second. And finally if you're lucky enough to have a 4k monitor -
and you can actually also record at 4K - then my recommendation would be to render your videos
at 100 to 120 megabits per second. And that about wraps it up for today's video. Now I actually took
much longer to produce this video than initially planned. I actually started producing data for
this like a few months ago because I always ran into some issues with actually analyzing the
data and then it didn't make sense because YouTube kind of decides to re-encode VP9 not at 60
FPS but rather at 60.117 FPS and then the sample data - like my reference and the one from YouTube
- didn't match. So I had to kind of figure out how to deal with this and yeah it was just a huge
mess really this video. I'm sorry it took so long, however if you're interested in me re-doing
these tests for 30 FPS then definitely let me know in the comments down below. And I've
also planned to kind of redo these tests with other types of footage such as Drone footage or
footage from a GoPro, which I'll probably upload on my second channel, because this is much more
in line with what I'm doing there. And if this video is already up then I'll have it linked
right here. But that's it for today's video. Thank you so much for watching, have a wonderful
day, and I'll see you guys in the next video!!