Question about Quality between I-P-B frame

[Starkian]

I was wondering about the quality reduction between I-P frames and P-B frames .
What does it means to set the Quantizer factor to a certain level (1.4 for I-P frames for example)
If I understand correctly
an I frame is a independant frame
a P frame can contain p-macroblock which is motion vector of a reference frame and I-macroblock which is independant macroblock.
Does a I-macroblock in a P-frame have the same quality that a I macroblock in a I-frame ?
What makes a P-frame smaller in size? I thought it was only because of p-macroblock, in that case why set a quantizer factor ?

[Ethlred]

What makes a P-frame smaller in size? I thought it was only because of p-macroblock, in that case why set a quantizer factor ?

Gee, a part I can answer. P-frames do not have all the information to make a frame. They refer back to the I or key frame for much of the information. Blocks that don't change for instance are taken from the key frame.

[jagabo]

Think of the quantizer as how much information is thrown away. The bigger the quantizer the more information you lose and the lower the quality.

I frames encode the entire image, much like a JPEG picture.

P frames encode only the differences between the current frame and an earlier frame. This is what makes them smaller than I frames. For example, a talking head shot may have a static background and only the speakers lips are moving. So only the moving lips are updated in the P frame. P frames usually use the same quantizer as I frames.

B frames encode the differences between frames and can reference earlier or later frames. They are usually also encoded with a higher quantizer (lower quality) to make them even smaller than P frames. The idea being that it's OK (you notice too much) to have a frame or two of lower quality because it will be cleaned up by an I or P frame soon.

Both P and B frames can use motion vectors to reduce the bitrate even further.

I'm not sure where you're getting the "quantizer factor" terminology from. It may let you specify how much higher a quantizer is used for B frames. Xvid for example has a quantizer offset and a quantizer ratio. The B frame quantizer is calculated as Q*ratio+offset.

[Starkian]

ok, let's me ask this way.

ipratio

Default: 1.40 Sets the target average increase in bitrate for I-frames as compared to P-frames. Seen as 'keyframe boost' in xvid. Higher values increase the quality of I frames. This makes them better references, which can improve the overall image quality. The problem is that the extra bits taken by the I-frames are taken from the P and B-frames, which makes this variable a balancing act.

pbratio

Default: 1.30 Sets the targest average reduction in bitrate for B-frames as compared to P-frames. This variable works more or less the same as ipratio above.

my question is how does these parameters affect the frame? let's say you have 3 differents frame A-B-C

frame A is 10% different from a reference frame
frame B is 30% different from a reference frame
frame C is 50% different from a reference frame

If for each frame you encode only what is different then you encode respectively 10%,30% and 50% and you will save the difference (90%-70%and 50%) but the way I see it that difference have nothing to do with ipratio or pbratio.

For the part that need to be encode , if I have a ipratio of 1.4 does it means that the part that need to be encode will have less quality but the result will be even more compression.

If i set the ipratio and pbratio to 1 , am I gonna have a reduction in frame size because it's only encoding the difference from a reference frame ?

for the terminology of "quantizer factor" i get that from the tool tips in MeGUI's x264 configuration dialog, in the quantizers ratio part of the advanced tab.

[Ethlred]

ok, let's me ask this way.

Let me answer this way.

Leave the bloody Quantisizers box settings at default. They were set the way they are by people that actually understand this crap to get the best result in most situations. Basically if you have to ask about that box leave it alone. I would have to ask myself. So I leave it alone as is suggested in every guide I have seen. The other boxes on the Advanced tab are fair game. Lots of things you can play with there so as to make the best mess possible.

Of course you could do a lot of tests and post your results. I promise to appreciate it.

However if you want to learn far more than you need to know to simply use the tool go to Doom9's forums.

Here are some links, non doom9 first by necessity:
http://en.wikipedia.org/wiki/X264

This is old but it might help a bit:
http://www.digital-digest.com/articles/x264_options_page1.html

Couldn't get on Doom9 at first

MPEG-4 Encoder GUIs:
http://forum.doom9.org/forumdisplay.php?f=78

MPEG-4 AVC:
http://forum.doom9.org/forumdisplay.php?f=77

Sometimes you can get answers from members of the x264 team as Doom9 is their primary discussion site. I have even seen guys from Microsoft's VC1 team there.

[Ethlred]

I just updated my copy of Megui. They have changed it more than usual.

It has three tabs instead of four, the former second tab is gone.

It has a setting for deadzones.

Adaptive Quantisizers.

The Custom Command Line that was on the 2nd tab is now on the Advanced tab.

The Me setting now has SATD exhaustive for the self-flagellants out there. Or perhaps they just have a dual processor quad core system.

While looking through the profiles I noticed that most have the same Quantisizers settings. A few do change them. The PS3-Xbox360 for instance.

Usually its I:P =1.4 and P:B =1.3

For a few its I:P =1.1 and P:B =1.1

I think its a matter of the needs for the systems where its changed from the standard. There are a couple of times where the minimum is changed also.

Glad I previewed this. I had to turn off smilies or I would be sticking my tongue out.