Live visualism needs a format that a) isn’t so huge it eats your hard drive, b) can be decoded and processed by your computer without bringing your CPU to its needs (DV compression causes some issues here), and c) still looks good, especially when scratched. We’ve had a great discussion going on the forums, in case you missed it:
Here’s the basic consensus:
- Codec: Motion JPEG (for interlaced footage) or Photo JPEG.
- Compression ratio/quality: Quality 80 is a decent baseline for JPEG, though you can crank as high as 97 to improve quality.
- Keyframes: Encode a keyframe on every frame so it’s “scratch-ready”.
- Alpha channels: For video containing alpha channels, PNG is the format of choice. (See comments.)
Now, I have successfully mixed two DV streams at full resolution; modern machines can handle the decoding. But, of course, this stuff is far more practical for use in terms of efficiency and size.
Next up: tips for improving performance in live video streams. I’ve got some new hardware and software I’ll be working with, and will share what I learn. In the meantime, any nominations (or questions) in that regard?