• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

new H.266 video codec

RickSanchez

Major Contributor
Cartographer
Joined
Sep 4, 2019
Messages
1,168
Likes
2,490
Location
Austin, TX

xykreinov

Senior Member
Forum Donor
Joined
Nov 16, 2019
Messages
424
Likes
678
Huh, interesting. I am curious to see how this competes with AV1- both technically and in adoption.
 

Fluffy

Addicted to Fun and Learning
Joined
Sep 14, 2019
Messages
856
Likes
1,423
I'm suspicious of broad declarations of "similar quality". There is always a loss when compressing, and the quality can only be better perceptually, if the data lost is less perceptible than in other codecs. H265 at the same bitrate of H264 loses quality in a different manner, I don't see it as "better quality".

Am I the only one who wishes we could stop with the numbers game and get lossless 1080P instead of 4 and 8K?

That would be rather pointless. Lossless 1080p at 24fps would be about 1.2 gb/s, which is still too high for most practical uses. You can compress it by a lot before any noticeable degradation takes place. That's exactly what specialized camera codecs do, and they never shoot uncompressed. 50 mb/s is already extremely high quality for 1080p.
 

maverickronin

Major Contributor
Forum Donor
Joined
Jul 19, 2018
Messages
2,527
Likes
3,308
Location
Midwest, USA
That would be rather pointless. Lossless 1080p at 24fps would be about 1.2 gb/s, which is still too high for most practical uses. You can compress it by a lot before any noticeable degradation takes place. That's exactly what specialized camera codecs do, and they never shoot uncompressed. 50 mb/s is already extremely high quality for 1080p.

If there was more interest in it, and thus more development, we'd have better lossless codecs that would bring those data rates down quite a bit.
 

bennetng

Major Contributor
Joined
Nov 15, 2017
Messages
1,634
Likes
1,692
When comparing codecs one should always consider the cost of encoding and decoding. For example, if the hardware being used is capable of realtime encoding/decoding or not. For instance you can always use a high quality H264 encoding preset to beat a low quality H265 encoding preset, it is just a matter of encoding time if you are doing it offline. Upon playback, if you are using a PC and a software-based decoder, and if your CPU is fast enough it is a non-issue.

However, when talking about SoC/ASIC/FPGA implementations, processing power, memory requirement, power consumption and so on are crucial. A video using a higher encoding profile/level may not be playable on these devices, even if the video has a lower bitrate than the one using a lower profile/level.

https://en.wikipedia.org/wiki/Advanced_Video_Coding#Profiles

For example, H264 actually has a lossless profile (High 4:4:4 Predictive Profile), and I used it to archive some of my gameplay captures for intermediate processing, but the encoded videos are only playable on my PC with software decoding, and CPU load is much higher than playing typical videos. My GPU's hardware decoder, the Smart TV in my home, and my phone can't play them, even if they are only at 480p and 55Mbps.
 

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
44,368
Likes
234,381
Location
Seattle Area
When comparing codecs one should always consider the cost of encoding and decoding.
Good point. This is actually the main reason new codecs get better. They assume more and more computational power available so change the algorithm to allow more freedom in its selection of tools. A simple 8x8 block can now be 4x8, 8x4, etc. which requires computational power to determine the best fit and hence better efficiency.
 

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
44,368
Likes
234,381
Location
Seattle Area
And you can guarantee streaming sites won’t keep the bandwidth the same and just allow better perceived bitrate, they will reduce bandwidth to where the perceived bitrate is the same as it was before.
No, that is precisely what the would use it for: lower the bandwidth. The only reason they would not is to support 8K resolution.
 

xykreinov

Senior Member
Forum Donor
Joined
Nov 16, 2019
Messages
424
Likes
678
If there was more interest in it, and thus more development, we'd have better lossless codecs that would bring those data rates down quite a bit.
How much improvement is enough?
VP8 and H.264 both possess lossless encoding capability, as do their successors VP9 and H.265, followed by AV1 and H.266 respectively. Efficiency in lossless encoding happens to improve to a decently comparable degree with each generation as efficiency in lossy encoding, too.
Lossless encoding comparison: https://www.texpion.com/2018/07/av1-vs-vp9-vs-avc-h264-vs-hevc-h265-1-lossless.html
I'm suspicious of broad declarations of "similar quality". There is always a loss when compressing, and the quality can only be better perceptually, if the data lost is less perceptible than in other codecs. H265 at the same bitrate of H264 loses quality in a different manner, I don't see it as "better quality".
While visual quality is somewhat subjective, arguably more than audio quality in most respects, quantification is in the works: https://www.texpion.com/2018/07/av1-vs-vp9-vs-avc-h264-vs-hevc-h265-3-quality.html
I believe that, if testing were to be done comparing people's preference in codecs at effectively equivalent encoding settings and bit rate in such a way that Harman compared people's preference in speaker tonality, we would see more conclusive evidence that variability in preference is rather small, and that next-gen codecs ultimately trump their predecessors.
 

Fluffy

Addicted to Fun and Learning
Joined
Sep 14, 2019
Messages
856
Likes
1,423
In my experience, H265 handles motion in video much better than H264.
I've seen weird motion artifacts in H265 videos. But my main complaint is that is smooths the image unnaturally. For example, in grainy material (like bluray of old movies shot on film), it smooths out the grain, and It looks weird.
If there was more interest in it, and thus more development, we'd have better lossless codecs that would bring those data rates down quite a bit.
There is huge interest in it – in the professional market. Ever since digital cameras were invented, there has been an intense race to achieve ever better codecs. Your assumption here that it's a neglected field is very incorrect.
 

Pluto

Addicted to Fun and Learning
Forum Donor
Joined
Sep 2, 2018
Messages
990
Likes
1,631
Location
Harrow, UK
…lossless 1080P…
I have seen genuinely lossless (i.e. uncoded) 1080p. Have you calculated the bandwidth necessary? Modest H264/5 coding reduces the bandwidth immensely and doesn't look all that different from the uncoded but totally impractical lossless version.
 

Berwhale

Major Contributor
Forum Donor
Joined
Aug 29, 2019
Messages
3,933
Likes
4,922
Location
UK
When comparing codecs one should always consider the cost of encoding and decoding

Indeed, at work we run ~140 Windows 10 virtual desktops per host server. The VMware Blast protocol is used to access the desktops and Blast supports both H.264 and HEVC encoding. These codecs can provide some improvement to the users desktop experience (especially over bandwidth limited connections). However the impact on CPU utilization is quite alarming, for example we'll often see a 15% jump in CPU utilization by enabling HEVC for the connection. This is within a single virtual desktop, so you can image the impact on the host of enabling it on the whole 140 VDI. A sensible company would of have purchased hosts that could accommodate graphics cards to offload and accelerate the H.264 and HEVC encoding...
 

Pluto

Addicted to Fun and Learning
Forum Donor
Joined
Sep 2, 2018
Messages
990
Likes
1,631
Location
Harrow, UK
Endgadget said:
It will be licensed by the Media Coding Industry Forum (MC-IF), a group with 34 major member companies. The aim there is to avoid the kind of licensing squabbles that plagued the H.264 codec a decade ago
I was distressed to read this…
Wikipedia said:
With 4 companies vying to be the patent pool administrator for VVC, the same problems that plagued AVC and HEVC licensing seem to be repeating themselves again
 
Top Bottom