|
Video Delivery Tutorial
Quality and
Size
Just choosing one video codec (QuickTime,
Windows Media Video, MPEG, RealVideo) over another means
you're going to alienate a small percentage of your
audience. Some content managers provide a couple of
formats, just to be sure. You might want to do that
yourself and see for yourself what your visitors prefer.
Another consideration is bandwidth
and CPU speed. A visitor downloading and viewing a 2MB
file that's 2 minutes long will be exercising their
CPU less than when downloading and viewing a 90MB file.
On older machines with slower CPUs and less RAM (and
that write slower to their cache and maybe with older
browsers), that 90MB file is going to choke and cause
some problems. If you're expecting them to download
a 90 MB file, you should provide some lower bandwidth
alternatives.
Let's say you have 10 minutes of video
you want to serve. On one hand, 10 minutes should be
long enough to download a large file, so the visitor
can watch while it downloads. But 10 minutes of streaming
video a la RealVideo, that's just as much being downloaded
but less is being retained in memory. A user with a
slower CPU and less memory would prefer a longer program
being distributed in smaller packets.
As for size, 320 x 240 is the average
these days for online video. If it's a shorter piece,
like under 30 seconds, you could go 640 x 480.
Frame rate is another variable that
affects quality and size. Television is displayed at
29.97 fps. Online video looks fine at 15 fps, and a
little choppy at 12. Anything less than 12 is useless.
18 is clean if you can afford the space.
|
Special
Considerations
Video viewed online and
viewed at 320 x 240 should be lit, filmed, edited, and
compressed differently than video destined for TV or
film. MPEG compression tries to take advantage of the
relative homogeneity of video: most of the screen is
still for several frames. Usually only a few items on
the screen are moving: a character, a vehicle, animal,
etc. MPEG compressors don't bother to "redraw"
static items on the screen, thus saving lots of room
and effort. Other compressors have started to catch
on and instead of each frame being an accurate representation
of what is on the frame pixel for pixel, newer compressors
are substituting pixels for math algorithms or the video
equivalent of "ibid", or "and so on."
Knowing that your video
will perform better and be smaller if you constrain
the movement to just a few areas of the screen means
instead of constantly panning, zooming, or doing other
typically television-like shots, the quality of your
video will improve. Instead of fancy wipes or dissolves
that seem pretty stunning on a large TV screen, stick
to cuts or very quick fades. Each fade means a huge
spike in pixel changes and the compressor has to work
harder.
If your video is 240
pixels tall, you can't sustain a full-length shot of
a person. Their head would be about 40 pixels high and
their mouth about 6 pixels vertical and 15 pixels horizontal.
Not a very easy task to see what's going on. Stick to
shoulder shots.
Subtle gradients are
harder to compress as well. Instead of that photography
studio-like velvet curtain with the angled light, stick
to something flatter. And don't even bother with mood
lighting or trying to set a "tone". It'll
just look dark and muddy. Go for the high contrast,
well-lit stuff. This ain't TV, it's a new medium with
new rules.
You can, of course, ignore
these guidelines. And it might turn out great. But when
working with multimedia online, it's safer to underestimate
the technical fortitude of your audience.
Gabino Travassos, Mote Interactive
Creative Director
MORE
TUTORIALS
|
