Divx question

[Enrico Ng]

Is there a way to get AVI_IO to use MP3 compression for audio?
I would use VirtualDUB, but I get dropped frames when I try to capture using the same settings

[Walrus]

(going to try to answer a lot of questions from a lot of people at the same time)
- I HIGHLY recommend the IBM 5400 or 7200 rpm drives. ALL of my 20 giggers be they 5400 or 7200 can handle full resolution YUY2 capture piped through huffyuv (which compresses the video losslessly and stores the perfect copy in ~ 8 megs/sec as opposed to 20 megs/sec) in VirtualDub. A lot of people on here (including Dr. Mordrid) have had great experience with the Maxtor's, though as a network consultant, I've seen too many of them die to recommend them. My 5400 rpm drives can handle ~ 10-12MB/sec captures, while my 7200 rpm drive can easily go to 20megs/sec. (all of them 20 gig models) However, with HuffYUV, you never need to exceed 10 mb/sec. Incidently, IBM's UW SCSI drives are also VERY nice (but much more expensive)
As for mp3 capture... It's HIGHLY discouraged. Only VirtualDub can correct for the codec skew introduced, and even then, I doubt it does it well in realtime. I'm working on a lossless ACM codec (ala HuffYUV) that could be useable for realtime captures in theory. What you should do is compress the video at the capture phase, and capture the audio raw. Once it's done go into VirtualDub and direct stream copy the video and compress the audio into mp3 at this point. (no further compression will be done to the video) For the BEST results capture the video with HuffYUV through YUY2 and the audio raw. Take the video into VirtualDub, denoise it, deinterlace it, color correct, etc, and then export it to your final video/audio codecs from VirtualDub.
As for YUY2 in W2K... if you were able to that is incredible as I understand that it doesn't support it. (then again, my W2K is refusing to boot right now so I can't check)
Also, I'm not convinced that YUV->RGB->YUV need be a lossy process (though it IS time consuming). I'm working on something...
As for debates about AVI_IO vs VirtualDub... I prefer the latter myself (I own a license to AVI_IO and I haven't used it in months), but the playback and timed capture features of AVI_IO are VERY nice. (supposedly it is also a bit better at audio synch, though I have no problem with VirtualDub in this regard even with hour+ captures)

[Enrico Ng]

I only capture video for two purposes.
capturing video for editing and output back to tape and making divx movies for distribution over the net.
Obviously, for the first, the quality needs to be higher. I'm still not sure how to capture for that, as MJPEG doesnt cut it, yet I dont have tons and tons of HD space.
I think there are some interlacing problems with my mjpeg capturing
As for capturing to send back to tape, I really dont have any idea how to do this.

[This message has been edited by Enrico Ng (edited 19 June 2000).]

[Walrus]

Enrico Ng: I don't have a G400, so I don't know if HuffYUV outputs to a tv set via DVDMax.. if it does, then I'd recommend for max quality, capture into huffyuv via yuy2 and then edit/process/etc in that format as well. Output to the TV using AVI_IO. If it doesn't support HuffYUV as a DVDMax codec, then capture/edit/process in HuffYUV and when you're done, render to Matrox MJPEG and then output via AVI_IO. These are the IDEAL configs if you need as little as possible degredation (scenario 1 should have virtually none at all) though even at ~ 8 mb/sec, it will eat up gobs of hard drive space. However, realistically for vhs tapes, the compression artifacts from 1 or 2 generations of MJPEG won't be too bad. The analog artifacts should mask any of the MJPEG artifacts and should be more than aceptable.

Of course, your mileage will vary.