Contents
Introducing the “High efficiency video” Mode
S23 has a new “High efficiency video prioritise quality” mode, which uses HEVC (High Efficiency Video Coding). This mode is selected by default for recording videos.
This feature lets you store equal or better quality videos using lesser space than before.
In case you want to save even more space, we still support the "prioritise saving space" mode, although this may degrade the quality slightly depending on the properties of the video.
The “Camera Settings” menu is as follows and you can select the mode based on your preference.
Introducing “High Bitrate Video” Mode
There's another new feature in S23 that uses the HEVC codec.
It is the "High Bitrate Video" mode. This mode records videos that are difficult to compress, like videos with quick motion, jerky movements or other complexities, but still make them look like the original.
High resolution videos can take twice as much the space, but it's a great feature to have if you need to save high quality footage under any circumstances.
Here's how to set it up. Let’s take a look at the results.
The below video is a pretty complex one. The camera cannot resort to the default "refer to the previous video" action to compress the video.
The left default mode cannot encode the shape of grass properly. This results in blurring and artifacts.
But, the right "high bitrate mode" encodes the grass contours neatly.
H.265 (HEVC) Properties and Galaxy HEVC
"Prioritise quality" and "Prioritise space saving" modes use H.265 (HEVC), which has better coding efficiency than H.264 (AVC).
HEVC is standardized for 50% more compression than AVC. Naturally, it consists of a lot of new technologies that allows you to compress and retain the same quality at upto half the capacity.
In the past, we mostly watched videos in VGA (640x480) resolution, but now we have FHD (1920x1080), UHD (3840x2160), and the more uncommon 8K (7680x4320).
In the following figure, you can see the size of the video according to the resolution. It has enlarged with time.
Thus, HEVC codec needed to be built for large videos. The standardization encompassed multiple technologies to compress more effectively than AVC at large resolutions.
Take a look at the compression efficiencies by resolution measured for standardization.
Compression gets more efficient with the increase in resolution. This points to the fact that HEVC codec is suited for high-resolution videos.
In terms of storage space, you'll only need half the space to record AVC equivalent quality for an hour of footage.
There are many other features, but we will avoid going into too much depth, and just talk about the main features and the current state of Galaxy.
First, the basic unit of compression in a codec is called a "MacroBlock MB" or "Coding Unit" (CU). (It was MB until AVC, and it’s called CU starting with HEVC)
Until AVC, macro blocks were 16x16 pixels in size.
However, as the size of the video increases, there is a lot of wasted data when compressing to 16x16 pixels.
So HEVC introduced a concept called CU, which allows for larger sizes like 32x32 and 64x64.
At these larger sizes, we compress uncomplicated areas to 32x32 or 64x64 sizes, which can significantly reduce the amount of wasted data.
In the image below, you can see that the HEVC CU selects the appropriate CU size based on the video type.
The larger CU size in HEVC is advantageous for compressing high-resolution footage and allows for more efficient compression than AVC for normal footage too.
And the video encoder basically encodes the current video by referencing the "already encoded video", where encoding is the same as compression.
The "Current video for encoding" will look for a similar video in CUs in the "reference video". If it fails to find a similar video, it will use the neighboring pixel values of a similar image to create a value that is as close to it as possible.
This behavior of finding or creating videos similar to the current video creates a "predictive video".
This "predictive video" is very important.
This is because we're coding the "predicted video" data subtracted from the "current video for encoding".
As the image above shows, the "predictive action" performs multiple prediction tasks with different sized MBs (or CUs).
The resulting "predicted video" is on the right. It's kind of similar but off at the same time, right?
What's really important to note here is that "the compression performance pivots on how well the predicted video is made".
When we subtract the original video from the predicted video, we are left with "residual data". This residual data gets stored.
The residual data looks like this: it somewhat resembles noise.
There are still many steps after that, including DCT, quantization, and VLC, but we are not talking about codecs, so we'll skip those.
Once again, the key is to get the "predicted video" right, because it is closely related to the compression efficiency of the codec, i.e. the compression performance.
Below is a "predicted video" created with AVC and HEVC, under the same conditions.
Which one’s the better out of the two?
Doesn’t the HEVC predicted video looks slightly better? It has less residual data and is more compression efficient.
HEVC has become very complex as the features used for compression have been upgraded and new features have been added.
The decoding complexity has increased by about 1.5 to 2x, and the encoding complexity has increased by a whopping 10x.
Galaxy has not harnessed the full potential of HEVC yet. The same goes for x265, a commendable open source codec.
However, Galaxy codecs are constantly being upgraded and new features added to produce good "predictive videos".
In the two videos below, you can see that the S23 has better encoding quality than the Note10.
We strive to provide you with higher quality videos every year. Stay tuned for more in the future!
CamCyclopedia Index - Samsung Members
You can also check out CamCyclopedia at any time by going to Community -> Category (app) -> CamCyclopedia -> "CamCyclopedia Index".
Reference
[1] https://en.wikipedia.org/wiki/High_Efficiency_Video_Coding
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.