Andreas Haugstrup Pedersen | solitude.dk

This video contains no audio. Feel free to leave your own music running.

Networks are noisy, the world is noisy. Loss is one manifestation of this noise. Rather than hide from this, or pretend that isn't there I think the very aesthetic of our work should celebrate noise as the productive expression of difference.

Videoblogs are not digital TV or DVD. Noise is an integral part of the medium, and the sooner people realise that we can free ourselves from statements such as “compression ruined my sunset shot”. Because, honestly, no one gives a shit. Not as eloquently as Adrian Miles above, but same difference.

Pollas asked for some tips on compression a couple of days ago. When I found out Michael Verdi was about to post a new compression tutorial on Freevlog I held out. Michael has taught me everything I know anyway. Michael's tutorial uses 3ivx, but I think you can get similar results using the h.264 codec that is shipped with Quicktime 7. 3ivx has the advantage of being playable in Quicktime 6 — it's up to you to decide what you want.

I uses Michael's settings for 3ivx with some important differences. Freevlog has always reccommended a datarate of 600 kbits/sec (75 KB/sec). The result will be files that clock in at around 4-5 MB/min. This is huge overkill in my opinion. Personally I use 160 kbits/sec (20 KB/sec) — you can still get quite crisp video with 3ivx. These files end up around 1-1.4 MB/min. As Adrian puts it: Any modern codec can make any bit of video look good at 4MB a minute. The real advantage of these new codecs is that you can now compress harder.

Nb. Quicktime changed their interface from version 6 to 7. Quicktime 6 lists the datarate in KB/sec, Quicktime 7 uses kbits/sec.

Unless I have a lot of fast movement in my video I also halve the frame rate. Since my source footage is 24 fps I compress it to 12 fps. If you're shooting in PAL the source is 25 fps so you can halve that to 12.5 fps. Unless you're making a video of a hummingbird in flight no one can tell the difference.

The last difference is that unlike Michael I leave the minimum quality slider in 3ivx at 10% instead of bumping it to 50%. Andrew Baron recommended setting the maximum/minimum sliders to 95%/6% for a dratically smaller file size. I haven't tried this, but it's worth a shot.

Finally Adrian Miles recommends not forcing a keyframe at a set interval. Both 3ivx and h.264 are supposed to handle this automatically for best results. In addition Adrian has put up a comparison of 3ivx and h.264 at low datarates (Quicktime 7 required).

Steve Garfield pointed me to some random guy who is thoroughly unimpressed with videoblogging:

Amateurism may be democratic and in keeping with Long Tail theory, but it doesn't do much to ensure quality.

However, there are some interesting videos here worth watching if you have the time and bandwidth to spare. Surely the Steven Spielberg of videoblogging is out there somewhere… but so far, he or she is keeping a very low profile.

Since he offers no examples of what he wants it's a little hard to help the guy out. One thing though: If he's expecting feature films he's misunderstood the whole concept. In the same way you don't see the next Jane Austen hacking away in a LiveJournal, you don't see videobloggers creating the next Indiana Jones (or The Celebration if you're into low-budget). Blogging is not the same as authoring a book, and videoblogging is not traditional filmmaking. For now don't decry the apparent lack of quality in videoblogs. Rejoice in the fact that there are so many people worldwide who love their cameras, and like to experiment with how they can use them (and the World Wide Web).