file size and frame size - correlated?

[Guest]

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

I took a 20 sec 704x480 uncompressed RGB AVI (formerly a VOB file that I
opened, deleted all but 20 sec, and saved w/vdub) and then opened it up and
compressed it in vdub using the "DivX MPEG-4 (Low-Motion)" codec using a
"Data Rate" slider setting of 510 Kilobits/s at 3 different frame sizes:
640x480, 320x240, and 180x120.

To my surprise all three clips were almost equal in size, about 1600KB.

I had assumed the clip size would increasr proportionately with frame size.
What am I missing?

Doctor Zaius

 

[Guest]

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

Doctor Zaius <DoctorZaius@ApeCity.gov> wrote:

>I took a 20 sec 704x480 uncompressed RGB AVI (formerly a VOB file that I
>opened, deleted all but 20 sec, and saved w/vdub) and then opened it up and
>compressed it in vdub using the "DivX MPEG-4 (Low-Motion)" codec using a
>"Data Rate" slider setting of 510 Kilobits/s at 3 different frame sizes:
>640x480, 320x240, and 180x120.
>
>To my surprise all three clips were almost equal in size, about 1600KB.
>
>I had assumed the clip size would increasr proportionately with frame size.
>What am I missing?

Bitrate and running time are really the only factors which determine
the size of the resulting file. The resolution only affects how high
a bitrate you will need in order to achieve acceptable results.

 

[Guest]

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

On Fri, 27 Aug 2004 12:36:20 GMT, DeepOne@ix.netcom.com wrote:

>Bitrate and running time are really the only factors which determine
>the size of the resulting file. The resolution only affects how high
>a bitrate you will need in order to achieve acceptable results.

Thank you for your reply. Is there a rule of thumb or direct calculation
for determining how much higher a bitrate one would need to use when going
from a lower frame size (which played fine, w/good quality) to a higher
frame size?

IOW, theoretically speaking, would a bitrate of twice that used to produce
a good 320x240 clip be required when producing (compressing in vdub) a
640x480 version of the same clip?

 

[Guest]

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

Doctor Zaius <DoctorZaius@ApeCity.gov> wrote in
news:24f1j0t11sf86b9k7j79g1j10fevplqn8h@4ax.com:

> On Fri, 27 Aug 2004 12:36:20 GMT, DeepOne@ix.netcom.com wrote:
>
>>Bitrate and running time are really the only factors which
>>determine the size of the resulting file. The resolution only
>>affects how high a bitrate you will need in order to achieve
>>acceptable results.
>
> Thank you for your reply. Is there a rule of thumb or direct
> calculation for determining how much higher a bitrate one would
> need to use when going from a lower frame size (which played fine,
> w/good quality) to a higher frame size?
>
> IOW, theoretically speaking, would a bitrate of twice that used to
> produce a good 320x240 clip be required when producing
> (compressing in vdub) a 640x480 version of the same clip?
>

Probably four times

You've doubled both dimensions.

HTH,
Gino

--
Gene E. Bloch (Gino) phone 650.966.8481
Call me letters find me at domain blochg whose dot is com

 

[Guest]

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

Doctor Zaius <DoctorZaius@ApeCity.gov> wrote:

>On Fri, 27 Aug 2004 12:36:20 GMT, DeepOne@ix.netcom.com wrote:
>
>>Bitrate and running time are really the only factors which determine
>>the size of the resulting file. The resolution only affects how high
>>a bitrate you will need in order to achieve acceptable results.
>
>Thank you for your reply. Is there a rule of thumb or direct calculation
>for determining how much higher a bitrate one would need to use when going
>from a lower frame size (which played fine, w/good quality) to a higher
>frame size?
>
>IOW, theoretically speaking, would a bitrate of twice that used to produce
>a good 320x240 clip be required when producing (compressing in vdub) a
>640x480 version of the same clip?

The following web page has graphs of appropriate bitrates for
different DVD-compliant resolutions (the resolutions you mention are
not DVD compliant): http://www.digitalfaq.com/capture/avivsmpeg.htm
(Scroll down about half way for the graphs)

Those graphs are for MPEG(2). You mentioned that you were using DivX,
and I have no idea what bitrates are appropriate for that except that
I think they might be lower.

 

[Guest]

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

"Gene E. Bloch" <hamburger@NOT_SPAM.invalid> wrote in
news:Xns955395D3A18A7Astrolabe@204.127.199.17:

> Doctor Zaius <DoctorZaius@ApeCity.gov> wrote in
> news:24f1j0t11sf86b9k7j79g1j10fevplqn8h@4ax.com:
>
>> On Fri, 27 Aug 2004 12:36:20 GMT, DeepOne@ix.netcom.com wrote:
>>
>>>Bitrate and running time are really the only factors which
>>>determine the size of the resulting file. The resolution only
>>>affects how high a bitrate you will need in order to achieve
>>>acceptable results.
>>
>> Thank you for your reply. Is there a rule of thumb or direct
>> calculation for determining how much higher a bitrate one would
>> need to use when going from a lower frame size (which played
fine,
>> w/good quality) to a higher frame size?
>>
>> IOW, theoretically speaking, would a bitrate of twice that used
to
>> produce a good 320x240 clip be required when producing
>> (compressing in vdub) a 640x480 version of the same clip?
>>
>
> Probably four times
>
> You've doubled both dimensions.
>
> HTH,
> Gino
>

Over the weekend, I thought about my answer above. I realize that in
truth I know nothing about this, and my naive response could be
quite wrong, depending on how these things are encoded. In
particular, there's the interlaced/non-interlaced problem...

Anyone care to comment? Politely, if you would

Gino

--
Gene E. Bloch (Gino) phone 650.966.8481
Call me letters find me at domain blochg whose dot is com

 

[Guest]

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

"Gene E. Bloch" <hamburger@NOT_SPAM.invalid> wrote:

>"Gene E. Bloch" <hamburger@NOT_SPAM.invalid> wrote in
>news:Xns955395D3A18A7Astrolabe@204.127.199.17:
>
>> Doctor Zaius <DoctorZaius@ApeCity.gov> wrote in
>> news:24f1j0t11sf86b9k7j79g1j10fevplqn8h@4ax.com:
>>
>>> Thank you for your reply. Is there a rule of thumb or direct
>>> calculation for determining how much higher a bitrate one would
>>> need to use when going from a lower frame size (which played
>fine,
>>> w/good quality) to a higher frame size?
>>>
>>> IOW, theoretically speaking, would a bitrate of twice that used
>to
>>> produce a good 320x240 clip be required when producing
>>> (compressing in vdub) a 640x480 version of the same clip?
>>>
>>
>> Probably four times
>>
>> You've doubled both dimensions.
>>
>> HTH,
>> Gino
>>
>
>Over the weekend, I thought about my answer above. I realize that in
>truth I know nothing about this, and my naive response could be
>quite wrong, depending on how these things are encoded. In
>particular, there's the interlaced/non-interlaced problem...
>
>Anyone care to comment? Politely, if you would

If you follow the link I posted and look at the graphs, it seems that
your assumption was fairly accurate.

 

[Guest]

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

DeepOne@ix.netcom.com wrote in news:QvNYc.3100$8d1.552
@newsread2.news.pas.earthlink.net:

> "Gene E. Bloch" <hamburger@NOT_SPAM.invalid> wrote:
>
>>"Gene E. Bloch" <hamburger@NOT_SPAM.invalid> wrote in
>>news:Xns955395D3A18A7Astrolabe@204.127.199.17:
>>
>>> Doctor Zaius <DoctorZaius@ApeCity.gov> wrote in
>>> news:24f1j0t11sf86b9k7j79g1j10fevplqn8h@4ax.com:
>>>
>>>> Thank you for your reply. Is there a rule of thumb or direct
>>>> calculation for determining how much higher a bitrate one would
>>>> need to use when going from a lower frame size (which played
>>fine,
>>>> w/good quality) to a higher frame size?
>>>>
>>>> IOW, theoretically speaking, would a bitrate of twice that used
>>to
>>>> produce a good 320x240 clip be required when producing
>>>> (compressing in vdub) a 640x480 version of the same clip?
>>>>
>>>
>>> Probably four times
>>>
>>> You've doubled both dimensions.
>>>
>>> HTH,
>>> Gino
>>>
>>
>>Over the weekend, I thought about my answer above. I realize that
in
>>truth I know nothing about this, and my naive response could be
>>quite wrong, depending on how these things are encoded. In
>>particular, there's the interlaced/non-interlaced problem...
>>
>>Anyone care to comment? Politely, if you would
>
> If you follow the link I posted and look at the graphs, it seems
that
> your assumption was fairly accurate.
>

Cool! I should've followed your link sooner...without needing to be
led by the hand. Oh well.

Looking at the shoulders of the graphs, the desired bit rate follows
the pixel area quite linearly, given the resolution of the graphs.

As a reward, I present you with Bloch's Lemma: Everything is linear,
to a first approximation.

(Insert smiley here.)

Thanks,
Gino

--
Gene E. Bloch (Gino) phone 650.966.8481
Call me letters find me at domain blochg whose dot is com