TMPGEnc File Size Not Changing

[dykesc]

I encoded a 320x240 avi at 2250 kbps to mpeg-2 with TMPGEnc and got a 9.68 MB final file size. I changed the resolution to 160x120 leaving the bitrate at 2250 kbps in TMPGEnc and suprisingly this setting also yielded a final mpeg-2 file size of exactly 9.68 MB. I then tried lowering the bitrate and still ended up with a 9.68 MB file size. I changed the settings in System/System Stream/Type to various settings, including mpeg-1 non standard, and still always ended up at a 9.68 MB final file size.

Obviously all my setting changes are being done without reliance on the "wizard" feature.

Why is the final file size not changing when I change the resolution and bitrate parameters in TMPGEnc?

Thanks.

[yoda313]

Hello,

Could it be the audio??? Is it vbr or cbr??? Is it pcm?? If it's uncompressed it will remain huge! Ac3 and mp2 can be changed from 92 to 384 (maybe lower and higher depending on the application - 448 is the highest I believe).

Try changing the audio bitrate as well to lower file size

Kevin

[jimmalenko]

Changing the resolution won't change the file size .... ever. It's all about bitrate. Try setting the System Stream Type to MPEG-1 System(automatic) while lowering the bitrate.

[dykesc]

Jim,

Would you please explain why changing the resolution doesn't affect the final file size.

Thanks.

[jimmalenko]

Would you please explain why changing the resolution doesn't affect the final file size.

Because the size calculation is as follows:

size = running time X bitrate

Your encoder will still allocate the same number of bits per frame (the number of which you specify by what you set your bitrate to), regardless of what resolution you use.

[vhelp]

I think that the confusion is with the "resolution" difference, that this
is always confusing to most peoples.

(
.. I think that some time ago, someone came out with a pixel-for-pixel
.. calculation count - something like (ie, 300,000 per frame *times* fps, or
.. something like that) But it was nice to see such info/stats
)

Anyways.

*** size = Running time X Bitrate

I did a small test encode of 100 frames (again.., I seem to just love doing
nothing better than test test test) anyways..

MPEG-2 w/ CBR 9000 bitrate encode:
** 352 x 480 .. 2,569 kb
** 720 x 480 .. 2,891 kb

With a difference of 322 kb's. I would guess, that onces you factor this
into the info (see above request) the pixel-to-pixel numbers will add up
coensidentaly.
.
But, the above 322 kb difference makes sense because you have a different
size container (or more (or lesser) cubes to fill)
.
But this ratio should always be the same when talking about the formula:

*** size = Running time X Bitrate

-vhelp 2972

[jimmalenko]

*** size = Running time X Bitrate

I did a small test encode of 100 frames (again.., I seem to just love doing
nothing better than test test test) anyways..

MPEG-2 w/ CBR 9000 bitrate encode:
** 352 x 480 .. 2,569 kb
** 720 x 480 .. 2,891 kb

With a difference of 322 kb's. I would guess, that onces you factor this
into the info (see above request) the pixel-to-pixel numbers will add up
coensidentaly.

Not to discount what you're saying, and I know that this test is only anecdotal, but IMO 100 frames isn't a big enough test sample, and quite frankly I'm very surprised that there's a 322KB difference over a sample that's less than 4 seconds long. One only has to extrapolate that to the average film length of 90 minutes (161870 frames @ 29.976 NTSC) to work out that there would be a 509MB difference between Half D1 and Full D1 should the pattern as displayed in the test sample be continued, which fully highlights the fragility of such a small test sample.

[vhelp]

Opps

My mistake. After double-checking, I found that I had the encoding mode
set to CQ

-vhelp

[jimmalenko]

Opps

My mistake. After double-checking, I found that I had the encoding mode
set to CQ

Those figures look much more possible now ...

[vhelp]

Alright.. alright

I re-did the test (by now, you know I love tests)

I did a small test encode of 1829 frames (1 minute) ..

MPEG-2 w/ CBR 9000 bitrate encode:
** 352 x 480 .. 69,385 kb
** 720 x 480 .. 69,387 kb

With a difference of 2 kb's.
( ..that better now ? )

-vhelp

[jimmalenko]

Alright.. alright

I re-did the test (by now, you know I love tests)

I did a small test encode of 1829 frames (1 minute) ..

MPEG-2 w/ CBR 9000 bitrate encode:
** 352 x 480 .. 69,385 kb
** 720 x 480 .. 69,387 kb

With a difference of 2 kb's.
( ..that better now ? )

sounds good !