Compressing file for purchased download

[Craig Seeman]

Regarding file size.

https://forums.creativecow.net/readpost/20/870185

[Ivan Myles]

Although the source material is anamorphic, you will want to encode using square pixels for web delivery.

The lower limit for compression of H.264 web video is about 0.07 bpp. Therefore your footage would correlate to the following:

(853x480x29.97 pix/s)*(0.07 bpp) / (1024 bits/kb) = 839 kbps
(640x480x29.97 pix/s)*(0.07 bpp) / (1024 bits/kb) = 629 kbps

A high quality target for H.264 web delivery is about 0.20 bpp, which correlates to 2400 kbps for SD 16:9 and 1800 kbps for SD 4:3.

Based on the data in your post it looks like the 16:9 version in iTunes is about 1.43 Mbps or 0.125 bpp. Depending on your desired level of quality, set the target video bitrate between 840-2400 kbps for your 16:9 480p version, and 630-1800 kbps for your 4:3 480p version.

The audio track would typically use AAC compression at 160-320 kbps.

[Christopher Dunning]

Thank you for the suggestions, Ivan. And thank you for making the point about file size, Craig.

I’m not sure I was clear in my original post so I’ll clarify here, though I suspect Ivan’s recommendations stand.

I am coming into the tail end of a very old project done by other people. I am trying to figure out the best way to get this 1 hour film available for purchase download via a website.

What I have as options are:

1) Open original FCP 6 file and output/export to Compressor
2) Output a standard QT file from FCP 6 and then run it through Compressor
3) Use the existing 15 GB file and run it through Compressor

So, I will follow Ivan’s suggestions in terms of target video bit rate settings and see what I get, but which of the three approaches is likely to yield the best results? Or is that a trial and error kind of thing?

And do people here think H.264 with a .mp4 container is a good approach? (That is what was recommended by someone else.)

Thank you again,
Christopher

[Ivan Myles]

It depends on the quality of the footage and the time available.

– You will get the best quality by rendering the original project file and encoding directly with Compressor, but this will require more time.
– A mezzanine file will be helpful if you plan to make multiple versions of the video at different bitrates and/or resolutions. EDIT: This is essentially your DVCPro files, which have 4:1:1 chroma sub-sampling. If desired, use the project file to create 4:2:2 ProRes files instead.
– Re-compressing the 15GB file will be the fastest option. The quality should be OK, but not as good as the other two alternatives.

Ultimately, try encoding a short test file using each method to see which one produces the most desirable results.

H.264/MP4 is a good approach, but there might be licensing implications for videos longer than 12 minutes.

[Christopher Dunning]

Thank you again, Ivan. This is really helpful.

And of course it makes sense to do a short test file, but I hadn’t thought of that. The 15 GB DV/DVCPRO -> 7 GB H.264 Compressor conversion took a couple of hours.

But if we want to play around with the quality / bitrate settings and not have to wait 2-5 hours (or more?) just to see if it works, we should set in and out on the FCP timeline and output that to Compressor. Or make a small sample version of the full 15 GB file (only a couple of minutes perhaps) and test settings with that.

Thank you for this.

[Ivan Myles]

You’re welcome.