Quantize vs. Two-Pass

[Joz3d]

What does quantization do? For example, in fairuse wizard there is the option when encoding to xvid to do a set mb-limit file and select two pass OR you can select Quantize and then a number, default being 4.

Which is better between quantize 4 and two pass 700mb? What does quantizing mean? Having trouble finding answers to this through google.

[jimmalenko]

https://www.videohelp.com/glossary?Q#Quantisation

That makes my head spin


My understanding is that Quantization or Quality-based encoding is a one-pass VBR that intends to retain the same quality throughout the encode, with or without constraints such as max and min bitrate. 2 (or more) pass VBR uses an initial pass to deduce what bitrate it will need to use, subject to settings of min, ave and max, and then uses the 2nd (and subsequent) pass(es) to actually encode the file so that in the end, the encoded file meets the constraints as set. I am unsure of (but it would be logical) that using more than 2 passes simply means that every pass except the last is an analysis pass such that only the last pass does the actual encoding.

The problem with quality-based encoding is that filesize is generally unpredictable, although some will claim that experience helps in guestimations of output filesizes. However, multiple pass VBR is generally within a couple of MBs of the calculated filesize, depending on the encoder (and calculator) used of course.

[Joz3d]

Oddly enough, fairuse wizard doesn't have an option for 2-pass quantizing, only quantizing followed by a number, OR two pass and setting a file size limit. I wonder why? Or is that number following the Quantize option the number of passes? That doesn't make as much sense though as the lower the number the higher the quality/file size.

[jimmalenko]

Oddly enough, fairuse wizard doesn't have an option for 2-pass quantizing, only quantizing followed by a number, OR two pass and setting a file size limit. I wonder why?

Because it doesn't make any sense. Quantization mode is one pass VBR.

OR you can do 2 pass VBR.

The number following is a quality value on a scale of 1 = highest quality down to 31 = lowest quality.

[guns1inger]

The number after the Quantize function is the relative quality setting you want the encoding to aim for. As explained in the Jimmalenko's post, quality based encoding is a single pass method. You tell the encoder what level of quality to try to maintain, and that is what it aims for.

[Joz3d]

I think I understand now, thanks!

[jagabo]

In 2-pass VBR mode the codec needs two passes. During the first pass it examines the video to determine how "complex" each frame is. During the second pass it uses this information to apportion bits. Frames that need more bits get more bits, frames that need less get less. In the end you get a file with the average bitrate (file size) you selected -- but you don't really know what the quality will be.

Constant quantizer mode encodes each frame to the quality you specify (via the Quantizer value) using whatever bitrate (at each frame) is necessary to achieve that quality. You know what quality the file will have but you don't know the file size until it's done.

If you encode a file in constant quantizer mode, then go back and encode in 2-pass VBR mode and specify the same file size, the two files will have approximately the same quality.

[vhelp]

Evening guys.

In MPEG (and other compression formats.. ie, divX/XviD/QT,etc) their specific
C.Quantizing; Motion Compensation; min/ave/max; subroutine algorithems will
vary in terms of expected quality from these routines.

Although the Math and Logic seems to dictate what should be, given a actual
comparison between MPEG A vs. MPEG B encoder, the two will produce different
outcomes, or whatever you happen to be comparing.

So, because of this, no matter how many (in this case/example) MPEG Encoders
you test a given scenario on, say the following MPEG Encoders:

min/ave/max -- Main Concept
min/ave/max -- CCE
min/ave/max -- TMPGenc
min/ave/max -- Procoder
.
.

..will ultimately produce FOUR different quality outcome from the other.
Some encoded mpegs in the above process will have different:

** quality levels
** file sizes

This does not mean that any one of the above MPEG Encoders are faulty, but
rather each have their own distinctive capabilities -- some good, some bad,
or some strong, and other, or weak etc.

-vhelp 4148