BBC HD compression

[Ar]

Was interested to watch / record the Vienna New Years Day concert, which
was shown on BBC HD live earlier today.

The BBC show in 1080i, and the file size came to 10.5GB. However, many
German satellite stations show in 720p, and the same concert came to
13GB. Considering the file sizes and size of picture frame, does that
not indicate the BBC are doing a massive amount of compression compared
to the better looking progressive scanned but 720p German satellite
picture frame?

[John Legon]

No doubt about it. I've just been watching a recording of the gala
re-opening of the Russian Bolshoi theatre broadcast this afternoon by
the Arte HD channel on 19E. Looked fantastic at 720p because of the
high bitrate. The larger frame size used by BBC is irrelevant because
the detail simply isn't there.

[J. P. Gilliver (John)]

In message <1JGdnRxV3P3NsX7N...@brightview.co.uk>, John

What is the original material sourced as (i. e. what cameras, or at
least what is the lowest line number it gets during processing before
sharing)?
--
J. P. Gilliver. UMRA: 1960/<1985 MB++G()AL-IS-Ch++(p)Ar@T+H+Sh0!:`)DNAf

Every time I think I know where it's at, they move it.

[Ar]

On 02/01/13 12:35, J. P. Gilliver (John) wrote:
>> No doubt about it. I've just been watching a recording of the gala
>> re-opening of the Russian Bolshoi theatre broadcast this afternoon by
>> the Arte HD channel on 19E. Looked fantastic at 720p because of the
>> high bitrate. The larger frame size used by BBC is irrelevant because
>> the detail simply isn't there.
>
> What is the original material sourced as (i. e. what cameras, or at
> least what is the lowest line number it gets during processing before
> sharing)?

From what I've watched on German satellite, it is likely to be 1080p
sourced*, with most German HD stations transmitting in 720p. The
progressive scan of German satellite beats the UK's interlaced to a
pulp, despite the resolution being less in 720p.

The compression on UK HD is clearly hurting the sharpness of the video,
if you can say a 720p picture looks as good or better than full HD.

* The Vienna New Years Day concert is shot in 1080p, it's released in
1080 on BluRay for past few years.

[Mark Carver]

On 02/01/2013 13:07, Ar wrote:

> * The Vienna New Years Day concert is shot in 1080p, it's released in
> 1080 on BluRay for past few years.

1080p50 ? Are you sure, I didn't think BluRay could (routinely) support
that, the 'highest' formats are 1080p24 or 1080i50 ?


--
Mark
Please replace invalid and invalid with gmx and net to reply.

www.paras.org.uk

[tony sayer]

In article <50e4312d$0$7317$5b6a...@news.zen.co.uk>, Ar <A...@127.0.0.1>
scribeth thus

Lets be thankful that engineers have some influence in the world
somewhere;)....

--
Tony Sayer

[Ar]

On 02/01/13 16:15, Mark Carver wrote:
> On 02/01/2013 13:07, Ar wrote:
>
>> * The Vienna New Years Day concert is shot in 1080p, it's released in
>> 1080 on BluRay for past few years.
>
> 1080p50 ? Are you sure, I didn't think BluRay could (routinely) support
> that, the 'highest' formats are 1080p24 or 1080i50 ?

It's shot in 1080p, the BlueRay is 1080i presumed 25 fps as it's not
printed on the box, and I have not got software to tell what the format
actually is.

The live 720p video of the Vienna concert from German satellite reports
720p100, from BBC HD it reports 1080i50 (I assume interlaced as I've
never seen BBC transmit progressive).

[Andy Burns]

Ar wrote:

> The live 720p video of the Vienna concert from German satellite reports
> 720p100

p100?

> from BBC HD it reports 1080i50 (I assume interlaced as I've
> never seen BBC transmit progressive).

I thought thy now dynamically switched from i50 to p25 depending on the
material/source?

[Andy Champ]

On 02/01/2013 19:53, Ar wrote:
> It's shot in 1080p, the BlueRay is 1080i presumed 25 fps as it's not
> printed on the box, and I have not got software to tell what the format
> actually is.
>
> The live 720p video of the Vienna concert from German satellite reports
> 720p100, from BBC HD it reports 1080i50 (I assume interlaced as I've
> never seen BBC transmit progressive).

p100? It's in 100Frames per second progressive?? Seems a lot - or am I
misunderstanding your abbreviation?

I'd expect p25 (25 frames per second) or i50 (50 fields per second,
interlaced, so 25 frames per second) her in PAL land.

Andy

[Andy Furniss]

Yes - I have samples that are progressive and are flagged as progressive
at h264 frame lavel. The flags on one change to interlaced just for the
credits.

I doesn't change anything for my TV using internal decoder - it just
stays in 1080i whatever.

[Andy Furniss]

It's possibly more than just bitrate. To me it looks like they
filter/blur bbc1hd. Maybe it's because they derive bbc1sd from it - or
maybe I am way off the mark.

Whichever it is, look at these, the bbc were broadcast recently (dvbt2)
and the content is progressive animation which doesn't need many bits.
The comparisons are from a 6mbit h264 blu-ray rip - about the same rate
as the bbc.

http://www.andyqos.ukfsn.org/stars-bbc.png
http://www.andyqos.ukfsn.org/stars-rip.png
http://www.andyqos.ukfsn.org/grass-bbc.png
http://www.andyqos.ukfsn.org/grass-rip.png

[Andy Champ]

On 02/01/2013 20:38, Andy Furniss wrote:
> It's possibly more than just bitrate. To me it looks like they
> filter/blur bbc1hd. Maybe it's because they derive bbc1sd from it - or
> maybe I am way off the mark.
>
> Whichever it is, look at these, the bbc were broadcast recently (dvbt2)
> and the content is progressive animation which doesn't need many bits.
> The comparisons are from a 6mbit h264 blu-ray rip - about the same rate
> as the bbc.
>
> http://www.andyqos.ukfsn.org/stars-bbc.png
> http://www.andyqos.ukfsn.org/stars-rip.png

To me the main difference is that the BBC one has lower brightness.
Which one is the source, or even if neither is, I couldn't say.

Andy

[Andy Furniss]

I don't think the brightness of the clouds is that different, the stars
yes - but then that's what would happen to points next to black if a
blurring filter was applied.

[Ar]

On 02/01/13 20:24, Andy Champ wrote:
> p100? It's in 100Frames per second progressive?? Seems a lot - or am I
> misunderstanding your abbreviation?
>
> I'd expect p25 (25 frames per second) or i50 (50 fields per second,
> interlaced, so 25 frames per second) her in PAL land.

I would assume it is progressive as I cannot see ANY jaggies in any
movement, which is easy to spot on BBC1 HD / BBC HD with their 1080i.

http://oi49.tinypic.com/10pc5kz.jpg

(if the link fails.. http://tinypic.com/view.php?pic=10pc5kz&s=6 )

[Roderick Stewart]

In article <50e55517$0$1053$5b6a...@news.zen.co.uk>, Ar wrote:
> > I'd expect p25 (25 frames per second) or i50 (50 fields per second,
> > interlaced, so 25 frames per second) her in PAL land.
>
> I would assume it is progressive as I cannot see ANY jaggies in any
> movement, which is easy to spot on BBC1 HD / BBC HD with their 1080i.

There wouldn't be if the material was shot at 25fps, which seems quite
common for TV material nowadays. Displaying it at 50Hz, interlaced or
otherwise, would simply display the same information twice every 1/25th
of a second. It's "filmic", you see, which is why they think it's good.

Rod.
--

[Ar]

On 03/01/13 10:57, Roderick Stewart wrote:
> There wouldn't be if the material was shot at 25fps, which seems quite
> common for TV material nowadays. Displaying it at 50Hz, interlaced or
> otherwise, would simply display the same information twice every 1/25th
> of a second. It's "filmic", you see, which is why they think it's good.

Maybe there's a film effect on BBC HD (which I hate), but there is no
such effect on the German satellite stations from what I've seen, just
pin sharp and no jaggies in 720p.

[Jim Lesurf]

In article <kc2g9a$grt$1...@localhost.localdomain>, Andy Furniss

The BBC have had a policy for audio of applying a low-pass filter to get
the lossy encoder to 'direct its available bits to the lower frequencies'
on the basis that they do more good there in terms of audio quality. No
idea at present if they've done something similar for HD video.

I don't have an HD TV. So my own direct experience of HD is limited to
capturing HD Freeview using a computer tuner, then watching the ts file
played back with VLC, etc. When doing this the two things I noticed about
the 'New Year's Day Concert' were:

1) That complex detailed items like roses in the flower display tended to
look much more blurred than extended simply structures beside them.
Implying that features like simple geometric shapes were being defined more
clearly than features that might have needed a lot of detailed info.

2) That slow zoom/pans tended to break up as if getting parts of the image
from two different frames.

How much of that was VLC or my playback machine, I have no idea. Nor if
twiddling with VLC settings would do much to affect it. Not experimented as
yet as my main interest remains in the audio rather the video!

Slainte,

Jim

--
Electronics http://www.st-and.ac.uk/~www_pa/Scots_Guide/intro/electron.htm
Audio Misc http://www.audiomisc.co.uk/index.html
Armstrong Audio http://www.audiomisc.co.uk/Armstrong/armstrong.html

[John Legon]

Jim Lesurf wrote:
> In article <kc2g9a$grt$1...@localhost.localdomain>, Andy Furniss
> <sp...@andyfurniss.entadsl.com> wrote:
>> Andy Champ wrote:
>>> On 02/01/2013 20:38, Andy Furniss wrote:
>>>> It's possibly more than just bitrate. To me it looks like they
>>>> filter/blur bbc1hd. Maybe it's because they derive bbc1sd from it -
>>>> or maybe I am way off the mark.
>>>>
>>>> Whichever it is, look at these, the bbc were broadcast recently
>>>> (dvbt2) and the content is progressive animation which doesn't need
>>>> many bits. The comparisons are from a 6mbit h264 blu-ray rip - about
>>>> the same rate as the bbc.
>>>>
>>>> http://www.andyqos.ukfsn.org/stars-bbc.png
>>>> http://www.andyqos.ukfsn.org/stars-rip.png
>>> To me the main difference is that the BBC one has lower brightness.
>>> Which one is the source, or even if neither is, I couldn't say.
>
>> I don't think the brightness of the clouds is that different, the stars
>> yes - but then that's what would happen to points next to black if a
>> blurring filter was applied.

It seems clear to me that the two versions are more or less equally
bright overall, and that the stars are indeed less bright in the BBC
version due to a reduction in sharpness and hence bit-rate:

http://www.john-legon.co.uk/temp/rip-bbc.gif

>
> The BBC have had a policy for audio of applying a low-pass filter to get
> the lossy encoder to 'direct its available bits to the lower frequencies'
> on the basis that they do more good there in terms of audio quality. No
> idea at present if they've done something similar for HD video.

That's how it appears to me. No doubt the BBC would argue that the
viewer doesn't notice at typical viewing distances and screen sizes, but
I don't think the result can be described as full HD.

[Mark Carver]

They do, but only on DTT, D-Sat remains at 1080i50, whatever.

[Andy Furniss]

Jim Lesurf wrote:

> The BBC have had a policy for audio of applying a low-pass filter to get
> the lossy encoder to 'direct its available bits to the lower frequencies'
> on the basis that they do more good there in terms of audio quality. No
> idea at present if they've done something similar for HD video.

Maybe, but then isn't what the variable quantiser is supposed to do on
the fly in encoders.

There may be valid reasons but it seems a bit much on that sample.

>
> I don't have an HD TV. So my own direct experience of HD is limited to
> capturing HD Freeview using a computer tuner, then watching the ts file
> played back with VLC, etc. When doing this the two things I noticed about
> the 'New Year's Day Concert' were:
>
> 1) That complex detailed items like roses in the flower display tended to
> look much more blurred than extended simply structures beside them.
> Implying that features like simple geometric shapes were being defined more
> clearly than features that might have needed a lot of detailed info.
>
> 2) That slow zoom/pans tended to break up as if getting parts of the image
> from two different frames.
>
> How much of that was VLC or my playback machine, I have no idea. Nor if
> twiddling with VLC settings would do much to affect it. Not experimented as
> yet as my main interest remains in the audio rather the video!

By default I expect vlc will be displaying interlaced as weaved frames,
which is not nice. To make things worse if your screen is not HD scaling
down weave artifacts can make them even more obvious.

Chroma will also probably look worse than luma - mpeg does different
4:2:0 subsampling for interlaced and progressive and most software
players by default will assume progressive so the chroma bleeds between
fields.

VLC will have a choice of de-interlacers which will help, though the
better ones tend to eat CPU. It's also a pain that most computer panels
won't do 50Hz.






>
> Slainte,
>
> Jim
>