Slashdot Mirror
Video Capturing Guide at Ars Technica
Deffexor writes "For those of you who read Ars Technica, but do not visit our forum, we have an active Audio/Visual Club where we talk shop about everything ranging from TVs to Stereos to Speakers to Videocards and everything in between. Lately, there has been a lot of interest in capturing broadcast television and converting old VHS home movies to a more timeless digital format, such as VCD, SVCD, and DVD. As more and more people become interested, it becomes increasingly difficult to educate everyone on how to do this properly. Tapping the collective consciousness of the Ars A/V forum, we bring you the 1st part of the Ars Technica Guide to Video Capturing, Cleaning, and Compression."
1. Capture source (& clean on the fly with the new Video Soap in the MMC 8.1 software). .vcr file to mpeg2 (ATI likes capturing in it's own format better). :)
2. Export
3. Run mpg file through FlaskMPEG to convert to DivX video with MP3 audio.
4. Cut commercials in VirtualDub and save using Direct Stream Copy (on Audio & Video settings)
5. Enjoy your capped copy of the (hopefully not) last episode of Farscape
Sometimes I doubt your commitment to Sparkle Motion.
If you want control and easy scripting, get Linux and capture with something like "dvgrab" and compress it with "ffmpeg" or "transcode" (search on Google, they pop right up). You can view with "xine" or "mplayer", and there are a bunch of editing solutions for Linux as well (although probably not as good as the commercial stuff).
If you want a no-frills, no-thoughts solutions, just get a Mac and use iMovie. It lets you capture, do some edits, then compress and burn to disk. Very easy to use (but nowhere near as flexible as the Linux solution).
I've been working on doing this very same thing - transferring some 8" & 12" laserdiscs that will never be released on a modern format (80's music videos, mostly).
One thing I've saw is that the article specifies a 40gb hard drive as a minimum. That's laughably small. I have twin 80gb drives spanned via RAID, and I filled them up with most of one side of a movie (about 50 minutes of video). Not only do I need more room for the 2nd side of the movie, I also need room for producing the final DVD MPEG files before burning them. Next paycheck I'm buying a couple of 200gb drives to replace them with, and I'm concerned that even they might not be large enough.
It also doesn't hurt to have the fastest CPU available. I'm on a Athlon XP 1800, and mastering/producing takes longer than the source material is (15min of material takes ~20min to produce). Don't think dual CPUs will help, as the production process is pretty much single-threaded.
Chip H.
If you add firewire and use an external capture device (like a DV camera or DV bridge) you can do quite a bit with an older processor. However, any video editing you attempt will be slow and painful. If I use my XP 2000+ to edit things go fast enough to not make it bothersome, but rendering can still take awhile. I've also used my wife's dual G4 and while it isn't any faster, the OS and software is so much better to work with, you wouldn't believe it.
Personally, I'd prefer Linux to finish polishing the end user experience for Office and Desktop Publishing software (because that's what I use most) before getting into the video editing relm.
1. Capture in as lossless and high resolution a format as you can. It's much easier to discard information than to make more in upscaling.
2. Halving the resolution means you can reduce the picture size by 4 times. But this does not mean you can quarter the bandwidth. Smaller pictures contain more detail per macroblock of 8x8 or 16x16.
3. Lots and lots of disk space. I purchased another 60 gigs just for the capture space. Never mind the processing space.
4. Since the article stays in Windows, try avisynth to do some of the post-processing. It saves quite a bit of disk space, but at the expense of time if doing two stage encoding.
5. If using Linux, transcode is fairly good, but it lacks the configurability of avisynth and Virtualdub with filters. It's just not as complete a set.
6. Interlacing bites. And an analog TV signal will definitely have an interlaced signal. You don't notice it on television because of the permanance of phosphorence. On a monitor that will do 85Hz, it's glaringly obvious. So do an inverse telecine on the video before encoding.
7. Big iron box. Encoding with any nontrivial filters (like an unsharp mask, or worse yet, noise smoother) will take a lot more CPU time than you could have imagined. Thank goodness that encoding is one of the most parallelizable things to do out there though.
I'd post more but I think this is enough noise for today.
Doing the Right Thing should not be preempted by making a buck.
Use at least a lossless compressor like HuffYUV or even, with a good machine, you could DivX or something on the fly at a high bitrate, and you /could/ even capture the audio in MP3 directly, or else just at 44.1Khz/16bit.
There's no way you should be needing that much hard disk space for a 50 min capture. I only have 80 Gig in all, and I captured and compressed a 2 hour film in MPEG-1 format, high quality, on the fly, on a PII - 350MHz @ 400x300 or something around that don't remember the exact figure. If I had a better CPU I could do better resolution and MPEG-2 on the fly.
Conversion Rate Optimisation French / English consultant
Yes, you can get a capture card cheaper ... but it requires there be working drivers for not only the system you run now, but the one you run in a couple years.
The Canopus ADVC100 (which I own) or the other "bridge" type products require no drivers other your computer having a working Firewire setup the same as you'd have with a DV cam.
--
BTW. These products are not TV tuner cards. They only convert an input video source. If you want TV capture, they're not the right solution.
"Hey, I'm a complete moron who, in between late-night one-handed surfing sessions, likes to make inane and ignorant slashdot posts bashing anyone with the initiative to make a contribution to the online tech community. I didn't actually read the article, so I wouldn't know that it was put together by a group of volunteers who donated their time and effort so that people like me can have easy access to technical information."
"Oh, by the way, I also have no clue about Ars in general, so I wouldn't know that the entire site (with the exception of the forums) runs on a single server, and that the guys who own it, run it, and contribute to it have day jobs in order to support themselves so that they can spend their precious free time creating high-quality web content that they give away for free. I would get a life, but it's just too easy to sit back and fire off a cynical post to Slashdot, hoping someone will mod me up and I'll have my very own flaccid little moment of poseur fame--a moment that, unlike the folks who contributed to the article I'm bashing, I didn't actually have to do any work for. "
Senior CPU Editor | Ars Technica | http://arstechnica.com/
I don't know any good guide on how to do this, but using MPlayer under Linux you "just" issue the following command:
mencoder -tv on:driver=v4l:input=1:width=640:height=480:adevice =/dev/dsp:amode=0 -ovc lavc -lavcopts vcodec=mpeg4 -oac mp3lame -lameopts abr:br=128 -o myvideo.avi
with the correct video capture device loaded (e.g. bttv), video cable plugged to the video input, sound cable to the soundcard, and you obtain a 640x480 DivX with MP3 stereo audio, on the fly, using a 600MHz+ machine :-)