Using H.264 to compress HD video for upload to FTP

[Richard Sanchez]

First off, I would have cut in a native 720p sequence and then used compressor to crop and resize to a 4×3 aspect ratio, since it will do a much better job than Final Cut, but that’s already done, so I won’t grill on that one any further.

Given that your client asked for a data rate of 9000 kbps, you shouldn’t set your encoder settings to automatic. Did the editor ask for a total date rate of 9000 kpbs, or a video data rate of 9000 kpbs. If it’s a total data rate of 9000 kpbs, you would set your audio to 192 kbps AAC and Set your Video Date Rate to somewhere in the 8700 to 8800 (allowing a little head room as there will be flucuations).

You should use Compressors frame controls to resize your video. Obviously, setting it to best quality resize will be the best method but it will take a very long time. You could set it to second best (Linear I think) and it will still do a great job and not take nearly as long. Shane has a good tutorial explaining frame controls at https://lfhd.blogspot.com/2006/10/dv-to-dvcpro-hd.html

Now the issue of broadcast quality arises. That really is a term that means nothing. Broadcasters will put up DV and less if their program requires it, and if it the network clears it. Being that your client gave you a specific data rate, use that. In response to the time length and file size, your client mentioned the data rate they wanted, and since the datarate will dictate the filesize, you’re doing what your client asked for. In regard to the time it requires to compress it, that’s just how it goes. Compressor does a great job, but it will take a while. I’ve heard that episode goes a little faster, but have never used it myself. You can buy different peripherals that will speed up encoding, but they tend to be not cheap.

Richard Sanchez
North Hollywood, CA

“We are the facilitators of our own creative evolution.” – Bill Hicks

[Carlton Hathcoat]

Thanks you for your post. The client only asked for a video data rate of 9000 kpbs. It was his recommendation. I have to admit I don’t know what that means, and why that’s important.
The issue we have with this is that we are in the business of doing Video News Releases, shooting, editing and delivering a 5-10 minute video to news stations all over the country, and thus we have very little time to wait for conversions. Having said that, I realize that these things take time. I am just simply looking for the fastest and best way to shoot the video on P2, edit in FCP (I chose to edit in a Standard Def DV timeline to minimize the size, Compressor seems to take too long. We then compress the video so that it is still broadcast quality 720 x 486(I totally get your point on what is exactly “broadcast Quality” I agree), and a file size that won’t take hrs to upload, or force us to spend ungodly $$ on an FTP that can handle the sizes. We really need these 5-10 minute videos be in the ballpark of 500mb and smaller.

[Herb Sevush]

Carlton –

H.264 takes a lot of time to encode. For that reason there now exist external USB accelerators to help speed up the process. I use the Elgato Turbo264 excelerator and it speeds up the proces by a factor of 4 to 1 over FCP. It costs under $100.

https://www.elgato.com/elgato/na/mainmenu/products/Accessories/Turbo264/product1.en.html

Herb Sevush
Zebra Productions

[Jeremy Doyle]

Does that elegato speed up 4 to 1 if I’m using an octo mac?

[Chris Paul]

Elgato Turbo is not any faster than normal H.264 export on an octo. Be aware also that Elgato only works with specific presets- it does not accelerate custom frame sizing in Compressor. I do use Elgato to accelerate rendering on an older computer that I offload some jobs to.

Chris Paul
POV

[Richard Sanchez]

What’s up Carlton?

In a nutshell, 9000 kbps, kbps standards for Kilobits per second. That means, every second of video, you are using 9000 kilobits to describe the video. Just calculating that out, 9000 x 60 (seconds) x 10 (minutes) equals 5,400,000 kilobits. A byte is about 7 bits, so divided by 7 that is 711,428 kilobytes which is around 700 Megabytes. Enough math! I did test to simplify.

I used a program to quickly spit out an h.264 mov with a datarate of 9000 kpbs and an audio data rate of 192 kbps. The resulting file was 657.24 MB That was a 10 minute file, so I was testing the extreme high end of your deliveries. Obviously 5 minutes would be in the 325 arena. The filesizes will be large due to the requested datarate, so to reduce that, you could ask them if they can take a lower datarate.

To alleviate the need to have to downconvert the footage, you could shoot standard def on your P2 cameras. That way, you wouldn’t have Final Cut doing the downscaling (since it scales pretty poorly). Grant it, your footage wouldn’t be future proof anymore. The reason Compressor is going to take so long encoding (especially with frame controls on) is because it is going to analyze your footage meticulously to preserve as much detail as possible in the downconversion process. If you’re not downconverting, by shooting standard def, it’s one less very processor intensive job for compressor to do. You should still expect 3 to 4 x realtime encoding in H.264 (since it’s fairly complex codec) but that would be faster than using compressors frame control options.

Telestream Episode is another compression tool that is known for being quick to encode, but I’ve never used it. Google it and check it out, that might be a good solution for your turn-around needs.

Richard Sanchez
North Hollywood, CA

“We are the facilitators of our own creative evolution.” – Bill Hicks

[Tim Irwin]

If you haven’t done it yet, you should take advantage of that octocore by creating a virtual cluster in qmaster. It will speed up any compressor task significantly. see https://www.slccut.com/Apple-Compressor-Tutorials/enable-multiple-core-processing-in-apple-compressor.html for details.

-tim

[Carlton Hathcoat]

Thanks for the info on the kbps. We definitely would have shot this on DV if that was the only use, but unfortunately the clients want the footage in HD for future marketing use, this is becoming common with these big companies. The problem is we need to send it out SD for stations that aren’t HD yet. Also, the file sizes of a ten minute HD video, unless you know another way, are too large, thus the conversion is time consuming as is the upload to the server and download for the stations.

[Rafael Amador]

Hi Carlton,
[carlton hathcoat] “shoot the video on P2, edit in FCP (I chose to edit in a Standard Def DV timeline to minimize the size, “
This is killing your DVCProHD footage.
You need to render to a codec. at least as good as DVCProHD.
Go to 8/10b Unc or Proress, then you can make the H264.
Rafael

http://www.nagavideo.com

[Ken Jones]

That is an awesome article. Thank you VERY much for posting the link.
I just rendered a 30 minute sequence to an m2v file in five minutes!