H.264 Encoding and delivery Best Practices @ MAX 2009

Only two weeks and the 2009 edition of Adobe MAX will begin.
This year in my presentation ( Encoding Best Practices for H.264 Video Using Flash. 5 October 2:00 pm Room 506 ), I’ll focus on changing the usual encoding perspectives proposing a dynamic approach in choosing the best resolution – bitrate mix.
Dynamic is always better than static, you know, and it’s much better in video encoding: reach a wider audience and save money!
I’ll also focus on playback best practices (full-screen acceleration, visual enhancements and so on).

Because a picture is worth a thousand words, today I want to show you the real potentialities of these approaches…

Me vs YouTube

In this test, I have downloaded from YouTube the video clip of Bjork “All is Full of Love”. This clip is an High Definition video encoded by YouTube in 720p (1280×720) at around 2Mbit/s.

The same (already encoded) video has been then re-encoded at a much lower bitrate of 500Kbit/s with a specific set of encoding parameters and pre-processings. I think the result is very good, and strangely enough, somewhat better than the original.

How can 500Kbit/s be better than 2000Kbit/s ? How can the quality improve re-encoding a damaged video at a quarter of the original bitrate ? Well, you are all invited at MAX 2009, 4-7 October in Los Angeles to discuss this and other topics.

Here you find the demo page

Note: decoding two HD video at the same time requires a fast computer. On my Core2Duo @ 2.2GHz Windows laptop it works very well.
Remember also that IE is faster in Flash Video decoding than FF.

But now let’s take a look by yourself and don’t forget to send me a feedback on your experience. What is the better looking video in your opinion, both in browser and at full-screen ?

9 thoughts on “H.264 Encoding and delivery Best Practices @ MAX 2009

  1. Fabio,

    There is a flaw in your comparison. The original video uploaded by the YouTube member has been encoded at 1000kbps. YouTube does not look at the bit rate of the source video and blindly re-encodes at their own bit rate setting for the various dimensions they provide.

    You seem to be implying that Flash media encoder does some magic with its limited options (as regards the H.264 codec).

    I’ve taken a close look at the two videos and I can say with absolute certainty that the YouTube version is better. The edges of the various objects are a telltale sign for starters.

    Unfortunately, that version itself is so crappy that it becomes difficult to see a difference.

    Of course with a video like this (hardly any movement, no drastic change of scene etc. one can afford to use a lower bit rate. But honestly, in my opinion, the YouTube version is not HD to begin with but the fault lies with the YouTube member who uploaded the original.

    1. I never mentioned to use Flash Media Encoder.
      We are talking of this comparison right ? http://www.progettosinergia.com/flashvideo/comparison.htm
      How can you say with “absolute certainty” that the left video in this page has better edges ? I know that “de gustibus non disputandum est”, but hundreds of people say the opposite (I speak of the side by side comparison).
      I know that the original youtube video is damaged, and infact my intent was to show that with a proper encoding pipeline and post processing in Flash the final effect is better even at a lower bitrate.

  2. you seem to mis the basic point. The YouTube video might as well as been encoded at 1000kbps since that is what the original is encoded at.

    Yes, I compared the two videos side by side and I can see the edges of your video to be smudgy. The YouTube video is too, but your version is a lot worse. I guess your concept of what is better differs from most people.

    Think of it this way:
    Over the years, Blu-ray disks have been encoded with higher and higher bit rates to get better quality HD so much so that you can look at older Blu-ray videos and tell the difference in picture quality as compared to newer videos. The players have had to be improved in order to handle the higher processing requirements.

    So from most of your posts, it would seem the better solution would be to simply add post processing in the mix and you’ll get better image quality? The TV could do that.

    Flash player is doing nothing innovative, but you seem to imply that in your posts.

      1. Fabio, those two images you are presenting do not seem to be the same in the video sequence. First you can see that the left one is at 0:30 while the right 0:32. Second you can tell that from the image content. For example, comparing the arches on the top part of the images, the left image shows a lot more content than the right one.
        Can you verify that both images are compressed with the same type, i.e., I/P/B picture?

      2. you are right, the pictures are not the same, but the sequence is “still” in that point, only robot arms are moving so there is no motion blur in action that alter the crispness of original picture. Anyway it was only a screenshot to respond to a comment. To compare the two videos by yourself follow the instructions in the article

  3. Sonnati, Can you share with us, your video preset for do this convertion? I would like to make the test using FFMPEG and your setting configuration and compare….

    Thanks, and regards from Argentine.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s