Versatile Video Coding (VVC) / H.266: HEVC successor - Page 31

FranceBB · 12th December 2021, 17:39

Quote:

Originally Posted by ksec

It is no longer much a tech issue, but a cost issue.

CDNs ain't cheap and management likes the "same quality, less bitrate/less expenses" sentence...

benwaggoner · 13th December 2021, 05:30

Quote:

Originally Posted by FranceBB

CDNs ain't cheap and management likes the "same quality, less bitrate/less expenses" sentence...

Depends on the market. Premium subscription services are a lot more likely to spend bits to improve quality. UHD and 4K content are a key differentiator from broadcast and cable.

FranceBB · 13th December 2021, 12:36

Quote:

Originally Posted by benwaggoner

Depends on the market. Premium subscription services are a lot more likely to spend bits to improve quality. UHD and 4K content are a key differentiator from broadcast and cable.

Eheheheh you're at Amazon so you literally have everything implemented already. If there's something you don't fall short of is bandwidth in cloud servers scattered across the globe :P

birdie · 16th December 2021, 11:33

An interview with Dr. Dmitriy Vatolin is the Head of the Moscow State University Graphics and Media Lab about video codecs.

Blankmedia · 21st December 2021, 08:52

For all the known and unknown in this, all we can say in regard to vvc is that we are gonna need more Ram.

Envoyé de mon Pixel 6 en utilisant Tapatalk

Dann0245 · 7th January 2022, 13:17

Does anyone hear any thing about x266 encoder from Multicoreware at CES 2022?

If no, x266 seem to be heavily delayed.

benwaggoner · 7th January 2022, 23:41

Quote:

Originally Posted by Dann0245

Does anyone hear any thing about x266 encoder from Multicoreware at CES 2022?

If no, x266 seem to be heavily delayed.

CES would be an odd show to market a video encoder. That's much more of a NAB thing.

birdie · 8th January 2022, 04:45

Quote:

Originally Posted by Dann0245

Does anyone hear any thing about x266 encoder from Multicoreware at CES 2022?

If no, x266 seem to be heavily delayed.

To be honest I've got zero faith in this codec.

Recently I tried to recompress an SD source (interlaced MPEG2 at 10Mbps) using x264, x265 and libaom and x264 came out on top by a wide margin. Both x265 and AV1 simply destroyed the source and even x264 @3Mbps struggled. I didn't venture to go below 3000.

H.266 might excel at 4K/8K but for SD/HD x264 is still the king which is really weird considering all the new intra/inter algos in new codecs.

Dann0245 · 8th January 2022, 11:25

Quote:

Originally Posted by birdie

To be honest I've got zero faith in this codec.

Recently I tried to recompress an SD source (interlaced MPEG2 at 10Mbps) using x264, x265 and libaom and x264 came out on top by a wide margin. Both x265 and AV1 simply destroyed the source and even x264 @3Mbps struggled. I didn't venture to go below 3000.

H.266 might excel at 4K/8K but for SD/HD x264 is still the king which is really weird considering all the new intra/inter algos in new codecs.

I think the new codecs aren't impressive with SD resolution because of big block (x265 CTU and AV1 superblock). x264 in the same term call macroblock, is only 16x16.

So to make x265 handling SD resolution well, I use small CTU

Code:

--ctu 16 --max-tu-size 16 --qg-size 16

or

Code:

--ctu 16 --max-tu-size 8 --qg-size 8

disable recursion skip: --rskip 0 might force x265 deep dive into small detail.

raise --qcom from 0.6 to 0.8, 0.9 to improve high motion scene.

turn off smooth filter:

Code:

--no-sao, --no-strong-intra-smoothing

I think these settings should help with SD, except if your source have ton of noises

with AV1 encoder, particular aomenc, I can only reduce superblock to 64x64, and AV1 has too many filters which make result too blurry. As I test all AV1 encoders are so terrible at high motion scene.

All new codecs love high resolution with big block, by now the block increases to 128x128.

Boulder · 8th January 2022, 13:41

Quote:

Originally Posted by Dann0245

turn off smooth filter:

Code:

--no-sao, --no-strong-intra-smoothing

--strong-intra-smoothing is not a smoothing filter. It actually should improve the retention of details.

Dann0245 · 8th January 2022, 17:10

Quote:

Originally Posted by benwaggoner

CES would be an odd show to market a video encoder. That's much more of a NAB thing.

On MulticoreWare website has a place about CES 2022

https://multicorewareinc.com/news/join-us-at-ces-2022/

Because of that, so I think they will talk about their products, also something about x266

Quote:

Originally Posted by Boulder

--strong-intra-smoothing is not a smoothing filter. It actually should improve the retention of details.

I don't know that. I usually disable when encode grain or high bitrate.

ksec · 8th January 2022, 22:18

Quote:

Originally Posted by birdie

To be honest I've got zero faith in this codec.

Recently I tried to recompress an SD source (interlaced MPEG2 at 10Mbps) using x264, x265 and libaom and x264 came out on top by a wide margin. Both x265 and AV1 simply destroyed the source and even x264 @3Mbps struggled. I didn't venture to go below 3000.

H.266 might excel at 4K/8K but for SD/HD x264 is still the king which is really weird considering all the new intra/inter algos in new codecs.

New codec excel at low bitrate. Think HD at 1Mbps or 4K at 4Mbps. x264 had a heck of a team and time tuning for SD and 10Mbps bitrate. It is more of an encoder issue and investment into making it good at those scenario, rather than the codec itself.

Boulder · 9th January 2022, 13:19

The world is moving towards digital media from the old physical media. That's why the development concentrates mainly on creating a watchable low bitrate stream from the source instead of replicating it accurately.

birdie · 11th January 2022, 17:19

No idea what to make of it: https://www.techpowerup.com/290799/a...pool-licensors

18 months after finalizing the codec they are excited about ... licensing. Not adoption, not products, not software ...

That sounds almost [don't want to get a ban here].

benwaggoner · 11th January 2022, 22:47

Quote:

Originally Posted by birdie

No idea what to make of it: https://www.techpowerup.com/290799/a...pool-licensors

18 months after finalizing the codec they are excited about ... licensing. Not adoption, not products, not software ...

That sounds almost [don't want to get a ban here].

Hey, if we are getting a single patent pool for VVC, that is a huge development. HEVC had three, which added a lot to confusion after the simple, public terms MPEG-LA had for H.264.

There's no doubt that VVC is technically the best video codec standard available today, by a substantial margin. Licensing problems are the main reasons why it wouldn't be successful.

nevcairiel · 11th January 2022, 23:41

No doubt MPEG-LA will also have one again, so thats at least two again.

Unless the technical standardization also enforces patent sanity for a future codec, there will never be one single point for this again, its just too much money.

PCU · 12th January 2022, 10:29

I'm sure most streaming services will use AV1 or AV2 in the future unless MPEG-LA (or something like that) is behind it.

hajj_3 · 12th January 2022, 16:41

Qualcomm were the main company that prevented a single patent pool on h265, i highly doubt they will join access advance until companies like google pressure qualcomm. Google owns android which is critically important for qualcomm's products. Google could easily force qualcomm to agree with their power over android if they wanted to.

benwaggoner · 12th January 2022, 18:30

Quote:

Originally Posted by hajj_3

Qualcomm were the main company that prevented a single patent pool on h265, i highly doubt they will join access advance until companies like google pressure qualcomm. Google owns android which is critically important for qualcomm's products. Google could easily force qualcomm to agree with their power over android if they wanted to.

If Google had that much power over Qualcomm, there would be AV1 decoders in Qualcomm SoCs...

ksec · 12th January 2022, 23:09

Quote:

Originally Posted by hajj_3

Qualcomm were the main company that prevented a single patent pool on h265, i highly doubt they will join access advance until companies like google pressure qualcomm. Google owns android which is critically important for qualcomm's products. Google could easily force qualcomm to agree with their power over android if they wanted to.

That is not strictly true. Qualcomm was "one" of them, not the main. There were lots of companies in the MPEG-LA camp that prevented the single pool. Including Apple.

In terms of Smartphone, it is Qualcomm which has the power. You want 4G / 5G along with other Modem technology? You need to deal with Qualcomm no matter what. And I would be surprised if that patent agreement doesn't include video compression, since Qualcomm lumped everything from CPU, power management to other things all in one package anyway.

AV1 and VVC will likely live side by side in the future. Since the industry doesn't seems to agree on anything.

7th January 2022, 13:17	#606 \| Link
Dann0245 Registered User Join Date: Apr 2020 Posts: 23	Does anyone hear any thing about x266 encoder from Multicoreware at CES 2022? If no, x266 seem to be heavily delayed. Last edited by Dann0245; 7th January 2022 at 13:19.

9th January 2022, 13:19	#613 \| Link
Boulder Pig on the wing Join Date: Mar 2002 Location: Finland Posts: 5,673	The world is moving towards digital media from the old physical media. That's why the development concentrates mainly on creating a watchable low bitrate stream from the source instead of replicating it accurately. __________________ And if the band you're in starts playing different tunes I'll see you on the dark side of the Moon...

11th January 2022, 23:41	#616 \| Link
nevcairiel Registered Developer Join Date: Mar 2010 Location: Hamburg/Germany Posts: 10,329	No doubt MPEG-LA will also have one again, so thats at least two again. Unless the technical standardization also enforces patent sanity for a future codec, there will never be one single point for this again, its just too much money. __________________ LAV Filters - open source ffmpeg based media splitter and decoders Last edited by nevcairiel; 11th January 2022 at 23:43.

Thread Tools	Search this Thread
Show Printable Version Email this Page	Search this Thread: Advanced Search
Display Modes
Linear Mode Switch to Hybrid Mode Switch to Threaded Mode

16th December 2021, 11:33	#604 \| Link
birdie Artem S. Tashkinov Join Date: Dec 2006 Posts: 313	An interview with Dr. Dmitriy Vatolin is the Head of the Moscow State University Graphics and Media Lab about video codecs.

21st December 2021, 08:52	#605 \| Link
Blankmedia Registered User Join Date: Oct 2011 Location: Dans le nord Posts: 64	For all the known and unknown in this, all we can say in regard to vvc is that we are gonna need more Ram. Envoyé de mon Pixel 6 en utilisant Tapatalk

11th January 2022, 17:19	#614 \| Link
birdie Artem S. Tashkinov Join Date: Dec 2006 Posts: 313	No idea what to make of it: https://www.techpowerup.com/290799/a...pool-licensors 18 months after finalizing the codec they are excited about ... licensing. Not adoption, not products, not software ... That sounds almost [don't want to get a ban here].

12th January 2022, 10:29	#617 \| Link
PCU Registered User Join Date: Oct 2017 Posts: 324	I'm sure most streaming services will use AV1 or AV2 in the future unless MPEG-LA (or something like that) is behind it.

12th January 2022, 16:41	#618 \| Link
hajj_3 Registered User Join Date: Mar 2004 Posts: 1,105	Qualcomm were the main company that prevented a single patent pool on h265, i highly doubt they will join access advance until companies like google pressure qualcomm. Google owns android which is critically important for qualcomm's products. Google could easily force qualcomm to agree with their power over android if they wanted to.