TMPGEnc- Frame rate 15vs29, file size SAME ?

[jmc]

TMPGEnc 4.0 XPress, mpeg-4 file output,
Ave bitrate 1Mbs
Max bitrate 4Mbs
2 pass Variable bitrate
---------------------------------------

15 FPS vs 29 NOT smaller. "point" 5% 4412 to
4390, 22Kb smaller, Half the frames.
I put the resulting file back into TMP to make
sure the changes happened.

So I have 500 frames the same size as 1000 frames.
Is this normal. I assumed half the frames would
equal half the file file size.

Also setting the res from 720x480 to 360x240...
less then 1% smaller file.
and CABAC vs CAVLC 4398 - SAME file size.

From what I have been reading I should not be
getting basically the same file size
with half the res or half the frame rate or CABAC
vs CAVLC

Any idea what is going on?

The adjustments (outside of the ones at the top)
are done one at a time.

Thanks,
jmc

[John Williamson]

The final size of the file is determined by the bitrate setting, not the
frame rate or size. If the duration is the same and the bitrate is the
same, then the filesize will be the same. Bitrate can be set
independently of frame size and rate.

To reduce the file size, keeping all else constant, reduce the bitrate.
Just note that the lower the bitrate, the worse playback will look.

--
Tciao for Now!

JOhn.

[jmc]

Hmm, Are the file getting "padded" to keep them
the same size (approx)?

To keep things simple for me... lets go with a 1
second file. one at 15 FPS and
one at 30 FPS.

There are 15 still images displayed in one second
at a set resolution (720x480).

And then you have 30 still images displayed in one
second at a set resolution (720x480)

Assume that nothing changes in the video. Each
"still image" should be the same size.

Don't see how 15 still images and 30 still images
(of the same thing) end up
having the same file size.

The file results back up what you are saying.
So I must be really looking at this in the wrong way.

I read of people saying to drop the resolution to
drop the file size...
Drop the dvd quality 720x480 to 512x320 ( or so)

That does not work here. (tmpgenc X4 mp4 files)
And the FPS, that seems so obvious yet I am
obviously wrong.

Thoughts VERY welcome!
jmc

[John Williamson]

No, they're just being compressed by a different amount. The original 30
frame sequence contains twice the data of the 15 frame sequence, and
both sequences contain a lot more data than the output file. As the
total amount of data in the output file is the same, the 30 frame
sequence "throws away" more data so it will fit in the same space.

> To keep things simple for me... lets go with a 1 second file. one at 15
> FPS and
> one at 30 FPS.
>
> There are 15 still images displayed in one second at a set resolution
> (720x480).
>
> And then you have 30 still images displayed in one second at a set
> resolution (720x480)
>
> Assume that nothing changes in the video. Each "still image" should be
> the same size.
>

They are *shown* at the same size, but the sequence of 30 frames
compresses each frame more. In other words, each frame in the second
case is represented by less data. So, if each frame in the 15 frame
sequence is represented by 20Kb, each frame in the 30 frame sequence is
represented by 10Kb. The playback unit expands them both to the same
frame size, but the 10Kb frames have less detail visible than the 20Kb.

> Don't see how 15 still images and 30 still images (of the same thing)
> end up
> having the same file size.
>
> The file results back up what you are saying.
> So I must be really looking at this in the wrong way.
>

Try looking at each frame seperately. In your case where you have 30
frames in a sequence, each frame takes half the space of the case where
there are 15 frames. This means that each of the 15 frames is, in
effect, less blurred than the 30 frames.

> I read of people saying to drop the resolution to drop the file size...
> Drop the dvd quality 720x480 to 512x320 ( or so)
>
> That does not work here. (tmpgenc X4 mp4 files)
> And the FPS, that seems so obvious yet I am obviously wrong.
>

If you have a 720x480 DV compressed (As used by MiniDV tape based
recorders) file, then reduce the size to 512x320, the file size reduces,
because each frame is compressed by the same percentage. With MP4, the
output file size is determined by the bitrate, independently of the
frame size. DVD format (Mpeg2 )also compresses each frame by the same
percentage, so reducing the frame size will normally reduce the file size.

Hope this helps. It's not exactly the easiest thing in the world to get
your head round.

--
Tciao for Now!

John.

[jmc]

John Williamson wrote:
> jmc wrote:

SNIPS

Hmm, Think I'm understanding some here. (To just
forget all the MPG stuff)
and If I can stand to watch less FPS then 29 I
will get better image quality per frame.

I've been changing a lot of setting (one at at
time) in TMPGEnc and see very little
or no positive change in the image...

Except for GOP. TMP defaults to 33 but that is awful!
The textures on the walls/floor etc SHIFT every
second or
I guess every 33 frames.

Jumped GOP up to 300 and all the shimmer was GONE!
And it reduce the file size 1414 to 1401 KB.

Glad to have finally found a useful setting
outside of bit rate that
made a real difference!

Thank you very much!
jmc

[John Williamson]

jmc wrote:
>
> Hmm, Think I'm understanding some here. (To just forget all the MPG stuff)
> and If I can stand to watch less FPS then 29 I will get better image
> quality per frame.
>

Given that you keep the same bitrate, yes.

> I've been changing a lot of setting (one at at time) in TMPGEnc and see
> very little
> or no positive change in the image...
>
> Except for GOP. TMP defaults to 33 but that is awful!
> The textures on the walls/floor etc SHIFT every second or
> I guess every 33 frames.
>
> Jumped GOP up to 300 and all the shimmer was GONE!
> And it reduce the file size 1414 to 1401 KB.
>

It also increases the length of noticeable problems if there's a glitch
in the playback stream, but you may be able to live with that if you're
using a reliable playback method. It's not a good idea if you're
streaming over the web or a busy network, for instance. The image on
screen can freeze for up to about ten seconds, as against freezing for
up to one second at thirty fps.

It may also help to use a two pass encoding technique, if you've not
tried that already. The first pass analyses the footage, and passes
control codes to the actual encoding mechanism, which varies the bitrate
while encoding, keeping the average bitrate the same, more or less.
Simple scenes get less space in storage, while complicated, fast moving
scenes get more.

> Glad to have finally found a useful setting outside of bit rate that
> made a real difference!
>
> Thank you very much!

Glad to help.

[jmc]

Seem to have lost my newsgroup message listing...
trying to rebuild this one.

>�John Williamson wrote:

>It may also help to use a two pass encoding
technique, if you've not

>tried that already. The first pass analyzes the

footage, and passes
>control codes to the actual encoding mechanism,
which varies the bitrate
>while encoding, keeping the average bitrate the
same, more or less.
>Simple scenes get less space in storage, while
complicated, fast moving
>scenes get more.

>> Glad to have finally found a useful setting
outside of bit rate that
>> made a real difference!
>>
>> Thank you very much!

>>jmc

>Glad to help.

My results in compression and time TMPGEnc 4x
100 Meg file MPG2 DVD 720x480. MAIN, 2PassVar.
1100 bps ave. 4000 max
At minimal settings........................1
minutes 51 seconds

One change at a time.

MotionSearchRange63 to 256 +43 bytes +1
second time
(A Quality setting... can't tell)

GOP 33 to 300, +2900
bytes +2 seconds
(Got rid of texture shifting)

B frame 0 to 3
0 bytes +1 second
(compression setting- NO effect)

Reference
Frame1 to 2 -
5840 bytes +4 seconds
(compression setting -worked,
3,4,6 less compression lot more time)

CABAC
+1603 bytes +4 seconds
(compression setting...
better compression- NOPE)

MotionEstimate 1 to 1/4 Pixel - 954 bytes
+16 seconds
(compression setting - worked
1 whole KB less out of 100 Megs @ 14% increase in
time!
--------------------------------------------------------------------------------------------------------

So does not seem to matter what the manual says
Somethings that are suppose to help compress
the file in fact increase the file size.

I was working with a 10 second file (300 frames) and
GOP actually did shrink the file with a 300
setting vs 33.
And I thought maybe a 10 sec 6 meg file was too small
for the different setting to really show what they
can do.

So did the 100 Meg file.

GOP-300 is a "must have" to control texture
shimmer. (bigger file or not)

Reference Frame-2 seems to be worth it.

B Frames-3 did NOTHING... disappointment.

MotionEstimate 1/4 pixel not worth the 14% time
increase.

The settings would not be there if there were no
use for them.
I just have not found it. Maybe fast sports shows
or something
else I don't work with.

Thanks,
jmc