my first Xvid encoding

[movmasty]

just a shorth clip.......
- using nandub - min / max drf 5 =4,325k
-using virtualdub-divx fast motion=4,354k
-using virtualdub-xvid,quant mode 5-H.263
luma mask on , motion search 5 = 4,861k

any comment?

[rui]

What was your desired file size?
What was your source? A vob file or other video? (i made myself some tests using already compressed avi's, but i not doing that anymore, only work with vob's. If you want to do a small test, rip some movie chapter or the trailer).
Beware that, for personal experience, making tests with such small files isn't very representative.
When you say Divx fast motion, you mean divx3 fast motion??

[movmasty]

rui....
only divx 3.11 got the fast and low motion codecs

-What was your desired file size?
i said i encoded in constant quantization mode, there you cant set a bitrate....
and this to test the size output of the codecs at same compression....

note that the drf 5/5 that i used in nandub is equivalent to divx fast
ot using the constant quantization 5 in Xvid

source was an already compressed small avi.

i will do other tests, and hope that other people will do.

[-h]

Try MPEG quantization as well. I might run a few tests on this, can't say I've seen DivX 3.11 outperform XviD by that far a margin, despite DivX 3.11's cheating with a prefilter ;)

-h

[movmasty]

yes, i knew just now that mpeg quantizers compress better
i believed that were the h.263 to do..........

[movmasty]

just did with mpeg quant:
new size is 4958k

as i said i used a quantizer of 5, so rather high.........

[-h]

Just did a few tests:

Quantizer = 5:

XviD, H.263 = 1.83 MB
XviD, MPEG = 1.67 MB
DivX 3.11 = 2.02 MB
MP42 = 2.03 MB

What kind of clip are you encoding?

-h

[rui]

Originally posted by movmasty
rui....
-What was your desired file size?
i said i encoded in constant quantization mode, there you cant set a bitrate....
and this to test the size output of the codecs at same compression....

i will do other tests, and hope that other people will do.

Sorry, after reading your first post, i thought that you mean that you used quantized motion search precision 5.

Well, the tests i am doing are always about 2 pass encoding, not constant quantization mode.

[movmasty]

ummm, carl lewis doing jumps, the 1991 cm...9.95...
dont think this is important........

but, didnt you said that h.263 gives less size at high quantizer?
5 is high..........

[-h]

but, didnt you said that h.263 gives less size at high quantizer?
5 is high..........

Nope, MPEG should always result in a smaller file size for identical quantizers.

The difference is in how that frame actually looks. With H.263, it will appear blurry, whereas with MPEG it will appear sharp (and blockier).

-h

[movmasty]

ok sorry

[movmasty]

but, maybe divx3 and mpg4 quantizers are not the same??
as doom pointed on, quality of xvid is comparable to divx at the same bitrate, then........

[-h]

but, maybe divx3 and mpg4 quantizers are not the same??
as doom pointed on, quality of xvid is comparable to divx at the same bitrate, then........

The DivX 3.11 quantizer is just standard H.263 - I'm at a loss as to what the bitrate savings are coming from.

Must do more testing...

-h