YouTube the hard way (x264 command line)

[Zacariaz]

Hi everyone

After much deliberation, trial and error, I've decided on the x264 encoder, nero aac and mp4box for my YouTube purposes and I'm going to make proper use of the command line. Seems simpler, how odd that may sound.

My input is taken directly from fraps. It doesn't seem to be a problem, but if you feel otherwise, please let me know.

As for the nero aac I haven't had any problems, but again, if you can argue that I should use faac or something else, I'm listening.

I'm assuming that mp4box is an insignificant detail, but again I'm very interested to hear what you have to say.

I've decided upon following the recommendations set by YouTube:

Container: .mp4

No Edit Lists (or you may lose AV sync)
moov atom at the front of the file (Fast Start)

I'm assuming that the above will not be an issue, or at least it has nothing to do with the x264 encoder.

Video Codec: H.264

Progressive scan (no interlacing)
High Profile
2 consecutive B frames
Closed GOP. GOP of half the frame rate.
CABAC
Variable bitrate. No bitrate limit required, though we offer recommended bit rates below for reference
Color Space: 4.2.0

The source material is taken from fraps, 720p@30 fps and again following YouTubes recommendations I've decided upon a bitrate of 5.000.

I've read a lot and gathered that --b-pyramid and --no-b-adapt is a no go, and that 3 reference frames is the right way to go, so with that in mind my settings look like this:

--bitrate 5000
--output-csp i420
--bframes 2
--b-adapt 2
--keyint 15
--ref 3

The result looks good enough, but could be better, thus I'm writing here to beg you to voice your opinions.

What should be different, what is missing and what have I misinterpreted?

Also I would like to go for at two pass encoding. I know it probably won't make much difference, but you've got to learn one way or another. Unfortunately I've been unable to comprehend how this is in fact achieved, or I have no been able to find the right source of information. Take your pick.

It should be noted that I'm by no means an expert, though I did mess around with such things years ago. It certainly was a different time then. Much has changes, so I'm sure you can understand my confusion. At this time I've been messing around with this many hours a day for about a week, I feel that I'm at a standstill and my head is both square and spinning.


I do hope that you'll take the time to throw a hint or two and on that note...


Best regards.

Edit:

Here is a small sample:

http://www.youtube.com/watch?v=mgY3OkT4Rk4

The source is about one gig, so I'm afraid I won't be able to upload that for comparison.

Edit 2:
regarding Fraps color space.

Some say that it's YV12, some say that it's RGB, RGB24, RGB32 and some say that it records in YV12 but saves it as RGBxx.

Suffice to say that I'm confused.

For now I'm going to assume that this is not an issue, but I would really like clarification.

Edit 3:

I've been so very very stupid. Apparently it's only when you check the "Force lossless RGB capture" that color space is an issue

Edit 4:
I've realized that this may not be the right place to pose these questions. If so, I would much appreciate being pointed in the right direction.