Video Riddle: 153=115 ?

Myouh · Post by **Myouh** » Sun Mar 18, 2007 6:37 am

2 video files:

VIDEO 1:

Audio:
10min28s
704X360
183kbps
MPEG Layer-3

Video:
23frame\second
244kbps
24bits
DivX

Size: 153 MO

----------------------------------

VIDEO 2:

Audio:
10min5s
704X369
193kbps
AC-3 ACM Codec

Video:
23frame\second
191kbps
24Xvid
XviD

Size: 115 MO

THE QUESTION IS: WHICH ONE HAS A BETTER IMAGE QUALITY?

Purge · Post by **Purge** » Sun Mar 18, 2007 6:58 am

do you slreasy know the answer or is this a serious question?

Corran · Post by **Corran** » Sun Mar 18, 2007 7:09 am

You can't tell simply based on bit rates/codecs alone...

Different sources and scenes require different settings.

I don't recommend using ac3 for your audio since a lot of people probably won't have the necessary software to play it back. Those bitrates seem a bit too low (especially since I assume the footage was not cleaned up in avisynth based on the overall structure of your post/nature of the question)

BTW, those are some funky resolutions (Aside from the fact that you listed resolution under audio...) Both height and width should be divisible by 16.

Myouh · Post by **Myouh** » Sun Mar 18, 2007 7:57 am

Purge: I don't know the answer.

Corran: these videos are just sources, not final exports.
in other words, the question is which one is best to work with?

(pay attention to the size)

Willen · Post by **Willen** » Sun Mar 18, 2007 8:41 am

The brutally honest answer? Neither. DivX or XviD video as sources are generally poor, especially at those bitrates.

Let's take a closer look at those videos. Examine them with MediaInfo or GSpot. The video bitrates you listed are suspiciously too low. At those filesizes, a typical 10 minute video should have rates at least 5 to 10 times greater (if not more).

That being said, your eyes would be the best gauge. Numbers alone aren't enough to judge by. I can make a very high bitrate video that by the specs may look good, but has massive blocking, poor color, and other problems. On the other hand, a well encoded video that is filtered properly for better compression will look much better even at lower comparative bitrates.

Scintilla · Post by **Scintilla** » Sun Mar 18, 2007 9:48 am

Willen wrote:The video bitrates you listed are suspiciously too low.

That's because Windows Explorer actually reports KBps instead of Kbps for video bitrates, which is odd because nobody else ever does. Multiply by 8.

Willen · Post by **Willen** » Sun Mar 18, 2007 12:02 pm

Scintilla wrote:
Willen wrote:The video bitrates you listed are suspiciously too low.
That's because Windows Explorer actually reports KBps instead of Kbps for video bitrates, which is odd because nobody else ever does. Multiply by 8.

Someone at Microsoft should be shot for using 'b' (bit) when it is supposed to be 'B' (byte) if that is the case.

Even examining a file I have for reference, MediaInfo gives me this:

Code: Select all

Video #0
Codec                : XviD
Family               : MPEG-4
Info                 : XviD project
PlayTime             : 1mn 13s
Bit rate             : 4834 Kbps
Width                : 704
Height               : 396
Aspect ratio         : 16/9
Frame rate           : 29.97 fps
Resolution           : 8
Chroma               : 4:2:0
Interlacement        : Progressive
Bits/(Pixel*Frame)   : 0.579
Writing library      : XviD0046

GSpot says the bit rate is 4838 kb/s.

AVIcodec says 4885 Kbps.

Window Explorer tells me 615 kbps (615 x 8 = 4920 kbps?). I guess everyone counts slightly differently.

Back to the original videos, factoring the 'error' of Windows reporting, we get about 1952 kbps for video 1 and 1528 kbps for video 2, which for the length and resolution are typical bitrates. Although, the resolutions (assuming no typing errors) are kinda weird for what I am assuming are 16:9 widescreen videos. 704 horizontal pixels are usually used for widescreen distros and the vertical for 16:9 is 396 (although 400 is better since it is mod16). 704x360 is very unusual: that comes out to 16:8.2. 704x369 (16:8.4) is just not normally possible since encoders (and resizers) generally work with even numbers. If these are 4:3 videos then, well...