Maximum Bit Rate - MeGUI

[doomleox99]

Hi, first of all forgive me if I dont speak (actually write...) english very well, because it isnt my first language.

Here is my question: Is maximum bit rate something really important?

I encode mostly ANIME videos of 300 MB to 100 MB. So I ALWAYS use tune animation and Automated 2pass but my doubt is if I should use the MEDIUM PRESET or the SLOWER PRESET (not Slow preset, Slower). I know that the second one obviously requires a lot more time, but I dont care because I can leave my computer all night encoding, so time is not a problem.

Here is an example. I encoded a 300 MB video two times, one with the MEDIUM PRESET and other with the SLOWER PRESET:

http://i.imgur.com/psm52RL.png



WHY WITH THE SLOWER PRESET I GET A LOWER MAXIMUM BIT RATE?

WHY THIS THING HAPPENS? IS IT IMPORTANT? A LOWER MAXIMUM BIT RATE MEANS THAT IN SOME SCENES THE QUALITY IS GONNA BE LOWER RIGHT? SO WHAT SHOULD I DO? ENCODE WITH THE MEDIUM PRESET SO I GET A HIGH MAXIMUM BIT RATE OR USE EVERYTIME I CAN THE SLOWER PRESET AND GET A LOWER MAXIMUM BIT RATE?? :confused:

Im kinda of confused. I hope you understand everything I wrote, my problem, my doubt and you can help me. THANK YOU ALL VERY MUCH.

[detmek]

MediaInfo can not measure maximum bitrate correctly so just ignore that info. If you can afford lower compression speed use slower preset as it will give you, more or less noticeable, better quality because you use same average bitrate.

P.S. Lets assume that MediaInfo can measure maximum bitrate correctly. Lower maximum bitrate for preset slower could aslo be the result of higher compression efficiency. Using slower preset x264 was able to compress same scene using less bitrate.

But, stick with original explanation - MediaInfo does not show maximum bitrate correctly. For that, it would need to analyze entire video, which MediaInfo does not.

[hello_hello]

Have a look with Bitrate Viewer (http://www.winhoros.de/docs/bitrate-viewer/).

[Sharc]

Even the "Max. bitrate" is - strictly speaking - an average.
The time interval for the measurements is just reduced to say 1 second instead of the entire clip. The value depends on the borders (start-stop time) of this test interval. If the borders change depending on the encoder settings we will obtain different results.
The measurement would be more consistent if the number of bits of a frame divided by the frame duration would be used.
Does anyone know which principle (time base) the tools are using for calculating the Max. bitrate ?

[Groucho2004]

Does anyone know which principle (time base) the tools are using for calculating the Max. bitrate ?
As for the bitrate viewer mentioned above, several methods are available. Frame based, GOP based, ...
For pro tools like Streameye, check the website for features.

[Sharc]

As for the bitrate viewer mentioned above, several methods are available. Frame based, GOP based, ...
For pro tools like Streameye, check the website for features.
Thanks. Interesting. Aparently there seems to be no standard. So unless we know how it has been measured we cannot compare the numbers between different tools.

[doomleox99]

MediaInfo can not measure maximum bitrate correctly so just ignore that info. If you can afford lower compression speed use slower preset as it will give you, more or less noticeable, better quality because you use same average bitrate.

P.S. Lets assume that MediaInfo can measure maximum bitrate correctly. Lower maximum bitrate for preset slower could aslo be the result of higher compression efficiency. Using slower preset x264 was able to compress same scene using less bitrate.

But, stick with original explanation - MediaInfo does not show maximum bitrate correctly. For that, it would need to analyze entire video, which MediaInfo does not.

Have a look with Bitrate Viewer (http://www.winhoros.de/docs/bitrate-viewer/).

Even the "Max. bitrate" is - strictly speaking - an average.
The time interval for the measurements is just reduced to say 1 second instead of the entire clip. The value depends on the borders (start-stop time) of this test interval. If the borders change depending on the encoder settings we will obtain different results.
The measurement would be more consistent if the number of bits of a frame divided by the frame duration would be used.
Does anyone know which principle (time base) the tools are using for calculating the Max. bitrate ?

As for the bitrate viewer mentioned above, several methods are available. Frame based, GOP based, ...
For pro tools like Streameye, check the website for features.


Thanks all of you for answering, I used bitrate viewer and put all the screen captures together so its easier to compare:


http://i.imgur.com/5NEpf2Z.png




So, aparently, MediaInfo is reliable because it gives almost the same Maximum bit rate as bitrate viewer, now that we clarify that, what do you think?

Should I use the SLOWER PRESET everytime I can?? Why using this preset gives you a lower Maximum bitrate?

Im not sure if this is important but remember that I encode almost only ANIME.



P.S. Lets assume that MediaInfo can measure maximum bitrate correctly. Lower maximum bitrate for preset slower could aslo be the result of higher compression efficiency. Using slower preset x264 was able to compress same scene using less bitrate.

But a higher compression efficiency means less bitrate right?, so how this could be better quality if it has less bits? Am I wrong?

THANKS AGAIN FOR YOUR ANSWERS.

[hello_hello]

Try clicking on the Bitrate Viewer title bar (icon, top left corner) and selecting "GOP based" and/or "Frame based" to see if things look much different. Well, they will.... the file I used for a quick test shows a maximum bitrate of 4728 kbps, 4178 kbps or 26056 kbps depending on the mode Bitrate Viewer is in. Depending on how the maximum bitrate is calculated the two encodes might look a lot more, or lot less different. You're currently looking at the "Second based" maximum bitrate calculation.

From what I can tell from the Bitrate Viewer screenshots you posted they look fairly similar. The first seems to have a slightly higher bitrate spike near the beginning but they don't look massively different. If the bitrates were exactly the same from start to finish you'd have to wonder what the point of different x264 speed presets might be. ;)

I'm no x264 expert but I'd guess the slower x264 speed preset producing a slightly lower peak bitrate here and there would be a reflection of the slower preset's ability to make better decisions and compress a tad more efficiently. That'd probably be a good thing, and not necessarily an indication the quality is lower. The bits being saved by the slower preset (in respect to peak bitrate) aren't wasted or thrown away. They're being used elsewhere. Both encodes use the same over-all (average bitrate). The bits are just distributed a little differently.

You could look at the differences in minimum bitrate and stress about that as an alternative. ;) The slower encode appears to have a slightly higher minimum bitrate.

[Sharc]

For the same files size (=same average bitrate) the slower setting will normally produce a better quality. Whether you can really see the difference and whether you are willing to wait longer for getting this quality difference is your decision. Don't worry about the Max. bitrate. If it becomes too high it may eventually produce stutter during playback in worst case.

[kuchikirukia]

From what I see, the difference between Medium and Slower is:
Adaptive B-frames goes from Fast to Optimal
Reference Frames goes from 3 to 8
RC lookahead from 40 to 60
Motion Estimation algorithm from Hex to Multi-Hex
Motion Estimation subpixel refinement from 07 to 09
And Trellis from 1 to 2

That's a lot.
The difference you're seeing could just be one B-frame.

Oh, and neither of yours are simply using the presets. You have 6 and 9 reference frames vs 3 and 8, and 5 B-frames vs 3.

[hello_hello]

Oh, and neither of yours are simply using the presets. You have 6 and 9 reference frames vs 3 and 8, and 5 B-frames vs 3.

That'd be the animation tuning wouldn't it? Increasing he number of B and Ref frames? And I guess the encoder would then limit the Ref frames as required to keep it Level 4.0 compliant. Or something......