I just sent out notes towards a 1.0 rev of the IJS spec. As a protocol, it's been pleasantly stable, and I get the impression that there are quite a few people using it without a lot of fuss and bother. This is good; I don't consider protocol design a shortcut on the road towards riches and fame.
We got an Apex 1100W DVD player for Christmas, which is a pretty cool toy. One of the things I've been playing with is transcoding DVD's to SVCD format. When done right, the quality is quite good, and I'm a fan of the CD-R format. It's quick, cheap, and easy to duplicate, readers are nearly universal, and it's good for archival as well.
Unfortunately, some DVD's are difficult to transcode with existing free software. I've had the best success with transcode and mpeg2enc (from the mjpegtools suite). But sometimes, particular for cartoon sources, there's jerky motion, visible interleaving artifacts, and A/V sync problems. I did a little digging and think I know what the problem is.
Most films are encoded into MPEG2 using "3:2 pulldown" telecine. Each 1/24th second film frame is converted into either 3 or 2 1/60th second video fields. By alternating between the two, the frame rates come out about right. (In fact, in the NTSC standard, the field period is exactly 1001/60000 second, which, if ignored, can lead to fun drifting of sync).
And indeed, most of the time, decoding of the MPEG2 stream will yield the original frames at (roughly) 24fps. Just scale to SVCD res, re-encode with the 3:2 pulldown flag set (-p option to mpeg2enc), and you're golden.
Only problem is, 3:2 pulldown is not exactly one flag in MPEG2 streams. In fact, it's implemented as two flags per "picture": "repeat first field" and "top field first" (described in more detail here). And, as one might expect, some DVD's use a fancier mix of these flags than vanilla 3:2 pulldown. For example, "Elmo in Grouchland" starts with 60 black frames with no pulldown, followed by the movie proper with 3:2 pulldown.
And the problem is that, as far as I can tell, there's no way to preserve the RFF and TFF field pattern from input (DVD) to output (SVCD) MPEG2 streams using free tools. In particular, mpeg2enc hardwires these flags on output to either all-0's (no pulldown) or a 3:2 pulldown pattern with a repetition period of 4 frames.
Modifying mpeg2enc to support arbitrary RFF and TFF, and transcode to provide these fields from MPEG2 input, would seem to be a very good idea. I'm sure whoever implements this will get love and appreciation from legions of users.
It's also possible that I'm missing a good way to do this with existing tools, in which case I'm sure someone will point it out.
Alan's school resumed today, and I start up with the homeschooling. I'm eager to try electrolysis of Hydrogen as a fun chemistry experiment. Since my goal is not (at present, anyway) to make a real table of elements, but to teach the basic concept, I think I'll go for the variation where the mixture of O2 and H2 gases fizz into soap bubbles, which are then be ignited for (hopefully) impressive yet safe effect. I'll let you know how it goes.
Max continues to astonish us with his language development. Over the past couple of weeks, he's pretty much mastered the alphabet song, and is just starting to be able to learn his individual letters (he knows "M", "A", and "X" pretty well now :). But he still managed to impress me with his facility at Alchemy, in which you have to match symbols and color. Some of our friends have been impressed merely that he's able to use the mouse. They should watch him clicking away at the thing like it's perfectly natural for a two-and-a-half year old. He's going to be a computer geek like his dad, I just know it.