[0:00] Hi, it's HandyAndy Tech Tips here.
[0:02] And today I'm going to tell you why H.265
HEVC, the latest video compression standard,
[0:09] is so awesome!
[0:10] Now, you know, video encoding has evolved
a lot through the years.
[0:14] From the very early MPEG-2, which was used
on DVDs way back in 1996, to the most recent
[0:20] H.264.
[0:21] But every new standard seems to promise the
same thing: identical quality to the previous
[0:27] one, but at half the bitrate.
[0:29] So how does H.265 achieve this?
[0:32] Well, there's two main methods used in video
compression.
[0:36] One is called "interframe", which means that
the previous and future frames are compared
[0:42] to the current one, and we only encode what's
changed.
[0:45] To do this, we start on an I-frame, which
is stored as a full image, much like a JPEG.
[0:51] Then, we divide it into small 16 x 16 pixel
areas - which, in the previous standard, were
[0:58] called "macroblocks".
[0:59] Now, we're going to advance to the next frame,
and compare its macroblocks with that of the
[1:05] I-frame.
[1:06] If there are several blocks which are roughly
the same, then we give the new frame the status
[1:11] of a P-frame: that is, a predicted frame.
[1:14] This means that what we can easily do is intercode
block A - so, because it's identical, we can
[1:21] just bring the pixel values directly across
- and then we can intracode block B. In other
[1:27] words, give it entirely unique pixel values
that are only for this frame.
[1:32] Now here is the H.265 difference.
[1:36] Macroblocks have now been renamed to "coding
tree units", and they can range from the original
[1:41] 16 x 16 size way up to 64 x 64.
[1:46] Now, why is this such a big improvement?
[1:48] Well, you've got to remember that, when H.264
was first standardized back in 2003, 1080p
[1:56] video was the height of technology!
[1:58] Now, of course, we've got 4K, and more efficient
encoding is needed.
[2:02] And this is exactly what larger macroblocks
give you.
[2:05] Now, the second big improvement has to do
with intra-frame compression - in other words,
[2:11] comparing areas within the same frame and
looking for redundancy.
[2:15] And the improvement is... wait for it... the
increase in prediction directions!
[2:19] What?
[2:20] Well, remember these things?
[2:21] Coding tree units, also called macroblocks?
[2:24] They can actually be split further, into things
called coding units, which can go down to
[2:30] 8 x 8 pixels.
[2:31] And these - get this - they can be partitioned
even further, cut up different ways into prediction
[2:37] units.
[2:38] Why would we bother doing this?
[2:39] So we can mathematically generate pixel values
in a block, instead of storing them.
[2:44] And, therefore, we can massively reduce the
size of each frame.
[2:48] Here's how it works - I've got my tiny little
prediction unit here, a little 4 x 4, and
[2:52] it's surrounded by two groups of pixels, which
I'll call Block A and Block B. Now, I can
[2:57] actually use various intra-prediction modes
on this structure.
[3:01] I can use, for example, DC, which will basically
fill this with the average of the surrounding
[3:07] pixels.
[3:08] This would be good on an outdoor scene, with
a blue sky, where there's no real pattern
[3:12] to it, it's more of a, just a single colour.
[3:15] Alternatively, I can use an angular function.
[3:17] So, say for example, if I have a line coming
down here, and I want to continue it through
[3:22] the block.
[3:23] Using my angular function, I simply point
to the direction where the line is coming
[3:27] from, and it will extrapolate, or continue,
that trend.
[3:31] Now, here again, there is a difference with
H.265, and it's a good one.
[3:35] Back in the old days, I only had 9 prediction
modes, but now I have access to - wait for
[3:41] it - 35 different modes!
[3:44] Now, the advantages to this are many, including
far greater accuracy when larger block sizes
[3:50] are used - say, for example, in 4K video - and
generally much better quality compression.
[3:55] So, they are the main reasons why H.265 is
better.
[4:01] Bit exhausting, hey?
[4:02] But they're all theoretical, and what you
might want to know is: when can you use it?
[4:07] And the answer to that, unfortunately, is
"not as often as you'd like".
[4:10] I mean, you can't upload videos encoded with
it to YouTube, and most web browsers don't
[4:15] support playing H.265 natively.
[4:17] But, it's been adopted as the default codec
on 4K Blu-Ray, and, of course, you can play
[4:23] it back on your PC, but you might need to
use software decoding.
[4:27] Anyway, I'm HandyAndy and thank you very much
for watching my video on H.265.
[4:32] Please subscribe to my channel for more tech
videos.