Wihich way to use Half D1- there are conflicting opinions ..

Archived from groups: rec.video.desktop (More info?)

In article <Lt6dnYruLvg3SBndRVn_iw@look.ca>, me@you.com
says...
> Strait to the point... Do you use half D1 to fit more on a DVD, or to use
> same bitrate and get better quality?

Yes.

> In my experience, encoding to half D1 doesn't reduce the file size to half,
> so if you use half the bitrate to encode, you will lose quality in the mpeg
> file.

Only if you haven't tweaked your encoding parameters.
(Motion Search Precision is the key one that needs to be
considered, and probably a few others.)

> I haven't got numbers, but encoding to half D1 only reduces the file
> size by about 25%. (which makes sense because you're only cutting one of
> the two dimensions in half - the vertical)

Uhhhhh... the *only* factors that affect the size of a
DVD clip are the bitrate and the run length. Resolution
(full-D1 or half-D1) has *zero* impact on file size.

.... and I'd suggest you go back and re-visit Geometry
101 (704x480 is 337920 pixels, 352x480 is 168960 pixels
or *half* as many pixels)

> So, you're only getting a bit
> more time on a DVD.. For VHS, isn't it better to keep the same bitrate and
> take advantage of less info per frame, thus better quality encoding at that
> bitrate? For example; VHS to a DVD at 4000kbps gives 2 hours at
> not-so-good quality, vs., VHS to DVD at 4000kpbs at half D1, which needs
> less compression to fit on a 2 hour DVD because it's smaller. Less
> compression means better quality (for VHS).

> I have read that using half D1 at a high bitrate is useless. I have done
> tests and found that a video encoded at half D1 - 6000kbps has less mpeg
> artifacts than a full D1 file at the same bitrate, and it looks better,
> (Again, with VHS source)

Half-D1 probably stops showing improvement after 4000-
6000. However, where you draw the line depends on the
quality of the encoder, VBR vs CBR, Motion Search
Precision, and any other tweaks that have been made in
the encoder settings.

The fact that you noticed that half-D1 encodes better at
a given bitrate then full-D1 is a bit silly. It's half
the pixels, so naturally it will require a lower bitrate
then the full-D1 to achieve the same quality.

> So, how do we use half D1? ... To save space, or to get better quality
> encoding for VHS archiving?
>
> Sanman
>
>
>

Archived from groups: rec.video.desktop (More info?)

"Toshi1873" <toshi1873@nowhere.com> wrote in message
news:MPG.1aeef5552897e54b989872@news-50.giganews.com...
> In article <Lt6dnYruLvg3SBndRVn_iw@look.ca>, me@you.com
> says...
> > Strait to the point... Do you use half D1 to fit more on a DVD, or to
use
> > same bitrate and get better quality?
>
> Yes.
>
I agree, yes to both parts of your question.

> > In my experience, encoding to half D1 doesn't reduce the file size to
half,
> > so if you use half the bitrate to encode, you will lose quality in the
mpeg
> > file.
>
> Only if you haven't tweaked your encoding parameters.
> (Motion Search Precision is the key one that needs to be
> considered, and probably a few others.)
>
Very important points, and they can be, and most often
are, impacted if not controlled by your capture HW&/ SW.
(While this seems self evident, these factors are seldom
included in the selection/consideration of hardware or
software for this purpose.)

> > I haven't got numbers, but encoding to half D1 only reduces the file
> > size by about 25%. (which makes sense because you're only cutting one
of
> > the two dimensions in half - the vertical)
>
> Uhhhhh... the *only* factors that affect the size of a
> DVD clip are the bitrate and the run length. Resolution
> (full-D1 or half-D1) has *zero* impact on file size.
>
Basically true, but using the lower resolution can allow a
lower bitrate, and still handle image changes better than
would be the case at a higher resolution. As long as the
source image (analog) can be digitized without significant
loss, at a any particular resolution (digital); a "true to
source" result is possible. Increasing the resolution above
what is needed for reproduction of the source, shows little
if any improvement, in terms of the resulting image detail
or other factors associated with the benefits of higher (digital)
resolution. There is only so much image data in an analog
TV or VCR signal.

> ... and I'd suggest you go back and re-visit Geometry
> 101 (704x480 is 337920 pixels, 352x480 is 168960 pixels
> or *half* as many pixels)
>
Ok;
Now what is here, is what was made of the original analog image
by the A/D chip (as a baseband, probably close to 100Mbps)
and then, for hardware sent to the MPEG Encoder chip or for
software provided in some "uncompressed" DV format.

If NO compression was applied then :

337,920 x 8bits = 2,703,360bits + 6,400bits (audio @192kbs)
= 2,709,760 bits per frame x 30fps = 81,292,800bps

168,960 x 8 bits = 1,351,680bits + 6,400bits
= 1,358,080 bits per frame x 30fps = 40,742,400bps

Every pixel rendered at 8 bits. ( of course there are 10 bit and
greater A/D encoder chips.)

Of course you never see the above, as MPEGs whole point
is to provide a compression to the images.

> > So, you're only getting a bit
> > more time on a DVD.. For VHS, isn't it better to keep the same bitrate
and
> > take advantage of less info per frame, thus better quality encoding at
that
> > bitrate? For example; VHS to a DVD at 4000kbps gives 2 hours at
> > not-so-good quality, vs., VHS to DVD at 4000kpbs at half D1, which needs
> > less compression to fit on a 2 hour DVD because it's smaller. Less
> > compression means better quality (for VHS).
>
> > I have read that using half D1 at a high bitrate is useless. I have
done
> > tests and found that a video encoded at half D1 - 6000kbps has less mpeg
> > artifacts than a full D1 file at the same bitrate, and it looks better,
> > (Again, with VHS source)
>
> Half-D1 probably stops showing improvement after 4000-
> 6000. However, where you draw the line depends on the
> quality of the encoder, VBR vs CBR, Motion Search
> Precision, and any other tweaks that have been made in
> the encoder settings.
>
> The fact that you noticed that half-D1 encodes better at
> a given bitrate then full-D1 is a bit silly. It's half
> the pixels, so naturally it will require a lower bitrate
> then the full-D1 to achieve the same quality.
>
All true, but I don't see why it's silly? It's kinda the point
isn't it?

> > So, how do we use half D1? ... To save space, or to get better quality
> > encoding for VHS archiving?
> >
> > Sanman
> >

Both, it's part of a balancing act that includes the points
that "Toshi1873" mentioned.

Luck;
Ken

Archived from groups: rec.video.desktop (More info?)

"Toshi1873" <toshi1873@nowhere.com> wrote in message
news:MPG.1aeef5552897e54b989872@news-50.giganews.com...
> In article <Lt6dnYruLvg3SBndRVn_iw@look.ca>, me@you.com
> says...
> > Strait to the point... Do you use half D1 to fit more on a DVD, or to
use
> > same bitrate and get better quality?
>
> Yes.
>
> > In my experience, encoding to half D1 doesn't reduce the file size to
half,
> > so if you use half the bitrate to encode, you will lose quality in the
mpeg
> > file.
>
> Only if you haven't tweaked your encoding parameters.
> (Motion Search Precision is the key one that needs to be
> considered, and probably a few others.)
>
> > I haven't got numbers, but encoding to half D1 only reduces the file
> > size by about 25%. (which makes sense because you're only cutting one
of
> > the two dimensions in half - the vertical)
>
> Uhhhhh... the *only* factors that affect the size of a
> DVD clip are the bitrate and the run length. Resolution
> (full-D1 or half-D1) has *zero* impact on file size.

Yes, I'm aware of that. My mistake, I was thinking of capturing
uncompressed video at lower resolutions, not compressing.

>
> ... and I'd suggest you go back and re-visit Geometry
> 101 (704x480 is 337920 pixels, 352x480 is 168960 pixels
> or *half* as many pixels)

Good point. Lose the sarcasm.

>
> > So, you're only getting a bit
> > more time on a DVD.. For VHS, isn't it better to keep the same bitrate
and
> > take advantage of less info per frame, thus better quality encoding at
that
> > bitrate? For example; VHS to a DVD at 4000kbps gives 2 hours at
> > not-so-good quality, vs., VHS to DVD at 4000kpbs at half D1, which needs
> > less compression to fit on a 2 hour DVD because it's smaller. Less
> > compression means better quality (for VHS).
>
> > I have read that using half D1 at a high bitrate is useless. I have
done
> > tests and found that a video encoded at half D1 - 6000kbps has less mpeg
> > artifacts than a full D1 file at the same bitrate, and it looks better,
> > (Again, with VHS source)
>
> Half-D1 probably stops showing improvement after 4000-
> 6000. However, where you draw the line depends on the
> quality of the encoder, VBR vs CBR, Motion Search
> Precision, and any other tweaks that have been made in
> the encoder settings.
>
> The fact that you noticed that half-D1 encodes better at
> a given bitrate then full-D1 is a bit silly.

I disagree. Encoding half the data with the same bitrate will render less
artifacts in the encoding than with full D1. It makes perfect sense that
the encoder will have less information to compress at the same bitrate, and
will have to compress *less* to stay within that bitrate.

>It's half
> the pixels, so naturally it will require a lower bitrate
> then the full-D1 to achieve the same quality.

Exactly, which is why it would look better at the same bitrate, other than
it being lower resolution. My tests show that there is far less motion
artifacting with Half D1 at the same bitrate as Full D1. The difference was
clear, especially during <pause>.

Thank you. You've answered my question.

Sanman

Archived from groups: rec.video.desktop (More info?)

On Tue, 20 Apr 2004 03:47:40 -0400, "Sanman" <me@you.com> wrote:

>Strait to the point... Do you use half D1 to fit more on a DVD, or to use
>same bitrate and get better quality?

Both. For instance, I'm encoding material from VHS as half-D1 in 100%
CQ mode. This usually amounts to a bitrate between 6500 and 7000,
although at times it can even go beyond 7000. I get the best in
compression terms, and some benefit in storage ones.

Archived from groups: rec.video.desktop (More info?)

On Tue, 20 Apr 2004 23:49:44 -0400, "Sanman" <me@you.com> wrote:

>My tests show that there is far less motion
>artifacting with Half D1 at the same bitrate as Full D1.

Obviously. But the half-D1 / full D1 question is a complex one. If one
needs encode at very low bitrates, say 2500, because of storage
considerations, one would use half-D1 in all cases, because that
bitrate is very low for a full D1. But at higher bitrates, and for the
same given bitrate, half-D1 gives you better encoding quality while D1
gives you a sharper image. And there are factors that can very much
affect the result. For instance, low motion would favour the use of D1
(for a better detail), while fast motion favours half-D1 (reducing
artifacts). The use of filters (denoise, sharpening...) would also
favour one or the other. Commercial movies use D1, thus evaluating
better the resolution factor than the compression one. But that is
because they use of 8000 and above. At lower bitrates things may be
not so clear. I do not have enough experience to aseverate on this,
and would gladly hear from those that have it, but wonder if a 3000
kbps half-D1 may not look better than a 4000 kbps D1, in spite of the
lower bitrate -at least, in some cases.

Archived from groups: rec.video.desktop (More info?)

Bariloche wrote:

> because they use of 8000 and above. At lower bitrates things may be
> not so clear. I do not have enough experience to aseverate on this,
> and would gladly hear from those that have it, but wonder if a 3000
> kbps half-D1 may not look better than a 4000 kbps D1, in spite of the
> lower bitrate -at least, in some cases.

I still think that in most cases, the Half D1 at 3000 will look better
than Full D1 at 4000. But it may depend on your source material.
High motion scenes are better represented with Half D1, and low motion
with Full D1. And if your source material isn't high resolution and
very sharp to start with, I see no reason to go Full D1 for your output.

See some examples for yourself:
http://dormcam.mine.nu:8080/restest


-WD

Archived from groups: rec.video.desktop (More info?)

On Wed, 21 Apr 2004 14:45:20 GMT, Will Dormann
<wdormann@yahoo.com.invalid> wrote:

>I still think that in most cases, the Half D1 at 3000 will look better
>than Full D1 at 4000. But it may depend on your source material.
>High motion scenes are better represented with Half D1, and low motion
>with Full D1. And if your source material isn't high resolution and
>very sharp to start with, I see no reason to go Full D1 for your output.
>
>See some examples for yourself:
>http://dormcam.mine.nu:8080/restest

Nice and instructive photos. Pity the DVD standard does not allow a
per-scene resolution!

Archived from groups: rec.video.desktop (More info?)

One more consideration: if you multiply 1150 (the Vcd bitrate) by 4,
you get 4600, which would be the bits/pixel equivalent of a 704x80
mpeg-2 for Dvd. For 720x480, it becomes 4705 kbps. Thus, 4000 kbps is
a really low bitrate for D1.

Now, I'm still wondering if Rempeg2 can do a better job than
DvdShrink, as it would use a resolution more appropriate to the
lowered bitrate; the con is that it re-encodes, which is a factor of
degradation. As Rempeg2 doesn't work for me, I may have to do the test
myself by a different and longer route.