16:9 - a formula for exporting?

[Leslie Knox]

Hi, I'm using 16:9 DV Cam PAL footage and want to compress it as 16:9 for the web. I'm not sure how big i want it to be on the website but i want to try out different sizes so i can get the best picture quality....

Is there a mathematical formula for working out a way of making the video smaller or larger but keeping the 16:9 ratio without distorting it?

I hope I've made myself clear from the description of what I'm after!

Thanks

[Adam Stanislav]

Originally Posted by Leslie Knox

Is there a mathematical formula for working out a way of making the video smaller or larger but keeping the 16:9 ratio without distorting it?

Well, assuming square pixels, if you want it to have a certain height, then width = 16 * height / 9. Or if you want it a certain width, then height = 9 * width / 16. And preferably you choose sizes that are integer multiple of those dimensions. For example, 640 is a multiple of 16, so it makes a good width. And then, following the the second formula, the height would be 9 * 640 / 16 = 360. So, 640x360 would preserve the 16:9 ratio.

[Sareesh Sudhakaran]

other than 16x9, you'll find you also need to worry about:

1. frame rates
2. bit rates
3. compression codecs
4. the player you'll be using on the website

Can't give you a formula since even the pros struggle with it. Every project is different, and every video is different. If you're doing it without help, expect to spend some time before you hit the sweet spot.

[Jeff Pulera]

When encoding using the 16:9 export sizing, be sure to choose "square pixels" or "1.0 pixel aspect", and also "deinterlace" anything encoded for the web

Jeff Pulera
Safe Harbor Computers

[David Johns]

Originally Posted by Leslie Knox

I'm using 16:9 DV Cam PAL footage and want to compress it as 16:9 for the web. Is there a mathematical formula for working out a way of making the video smaller or larger but keeping the 16:9 ratio without distorting it?

Also be aware that standard-def PAL footage isn't actually 16:9, it's fractionally wider. So, when converted to square pixels, rather than the mathematically-obvious 1024x576, it's actually 1048x576 - the extra pixels on the edge are just not displayed on a TV screen.

When you export said footage at 16:9 using square pixels, you will get a very slight letterbox effect top and bottom, unless you crop the lot to a true 16:9 ratio, or are willing and able to use an "expand video to fill screen" option like Sony Vegas has.

Regards
Dave

[Edward Shaw]

Take this for what it's worth, because I am still looking for the holy grail.

RE: For on line distribution. You Tube, Vimeo, Vzaar.
XDCAM EX, MPEG-2 Long GOP Codec, 35 Mbps (high)
To: FCP 6.06
So far, the best I have been able to do is the following:
Video Select 720/30P HQ NTSC (1280X720 29.97 Progressive)
Export QT Movie using QT Conversion
(Select options: H.264, NTSC720X480, key 24, no letterbox, fast start, Broadband Medium)

Here's the thing: you want to get as close to the site decoder settings as you can with the
fewest compression steps. After a while, this will be obvious to you, but not right away it isn't.
The above scheme passes through the encoding codec only once before uploading; namely,
FCP's QT Conversion.
The chances are slight we both use the same equipment.
If you think FCP is a moving target, get ready for You Tube. They are constantly 'improving'
and what works today may not work the day after tomorrow. So, what I am getting at with
this is trial and error on your part. Also, better luck with Vimeo. I think You Tube should get
the Nobel Prize, but is has gotten so big. Another thing, those hosts have tricks to refine
your upload. Processes are at work even days after the upload. Think about it: data stream is big dollars
to them. So don't give up on a presentation the same day. Let it cook on there a couple of
days before you yank it off and try something else.