DV-AVI - are they all the same?

gus2150 wrote on 12/7/2003, 8:06 AM
I created a video file (a short title with moving text) in Windows MovieMaker 2 and saved it my my hard drive as a DV-AVI (NTSC). It was 8.9 seconds long and the resulting file size was 33.75 MB or 3.8 MB per second of video. This comes out to 13.68 GB for one hour of video, a number I have read for AVI files.
However, when I import this AVI file into VideoFactory 2, then, without any editing, I go to "make movie" and "write your movie to file on your disk". Here I have two AVI options, AVI for Windows and NTSC-DV. The size of the AVI for Windows is 358.64 MB (about 11 times the size of the original) and the size of the NTSC-DV is 33.21 MB, essentially the same as the original.
This tells me that the DV-AVI output by MM2 is highly compressed, to about 11% of the uncompressed AVI. Interestingly, I have seen this 11% when viewing "Properties" of an AVI file played in Windows Media Player.
So what's going on here? I naively thought an AVI was an AVI and if I used that format I could rest assured that it was uncompressed and I could edit and save, edit some more and save, etc. without losing quality. But if its a file compressed to 11% of the original, and it gets compressed again each time I save it, wow, I'll have junk in no time.
Bottom line: I want to capture digital video from my Sony Digital 8 camcorder, bring it in to Sony ScreenBlaster MovieStudio 3, do some very minor editing (just trimming), then render to MPEG-2 and burn to DVD. (I have both Sonic MyDVD 5.2 and Ulead MovieFactory 2). What is the recommended workflow and what file formats should I use? Oh, I also have Scenalyzer which I can use to capture. I like the way it creates separate files for each scene captured from the camcorder. That makes it easy to just toss out unwanted scenes, and only work on the ones I want. Thanks in advance for your help.

Comments

Chienworks wrote on 12/7/2003, 9:02 AM
DV is a compression codec that can be used with AVI files. In fact, DV is the compression that your camera uses when recording the signal to tape. It's a rather minor compression and the loss is minimal. Some folks refer to it as 5:1 compression but i have no idea why since it's very plainly obvious that the compression is closer to 8:1 or more as you've seen with your file sizes.

Once a video has been compressed to DV (which has already happened in your camera) then any further DV recompressions suffer very little quality loss. If you're only going through one generation of adding titles and other editing then don't worry about it. In fact, SONY's DV codec is so good that it can easily withstand 99 generations without noticeable degredation. I've posted the results of my experiment here http://www.vegasusers.com/testbench/files/generations/ so you can see the results for yourself.
gus2150 wrote on 12/7/2003, 10:32 AM
Thank you, that makes sense. Can I then assume that it doesn't matter what program I use to capture the video from the camera - since the camera is outputting DV-AVI and all the program is doing is putting the data stream into a file, or several files, and storing them on my hard drive?
However, If I select to capture at some other format, say MPEG-2, then it probably DOES matter which program is used to capture the data - since now the program is rendering the DV-AVI data from the camera to the MPEG-2 format. Would that be a correct assumption?
Which leads me to another question - sorry for all the questions, but I promise to learn quickly! Since I will be doing minimal editing, should I capture directly to MPEG-2? Nice because of the much smaller file sizes. I read in another post that MovieStudio 3 does a better job of rendering to MPEG-2 than either MyDVD or MovieFactory. However, the authoring program will very likely re-render the video anyway, thus voiding any advantage, and possibly making the quality worse due to multiple rendering. Again, what's the workflow that will give me the BEST quality on DVD viewed on HDTV?
Chienworks wrote on 12/7/2003, 12:28 PM
Let's see, i think the answers would be "Yes", "Yes", and "NO RUN AWAY RUN AWAY!!!!!"

Yes: As long as you're capturing through a firewire/1394/ilink interface, all you're doing is transferring a data stream. There is no rendering, recompressing, processing, etc. going on. You'll get an exact copy of what the camera recorded.

Yes: Some programs are much worse at rendering to MPEG than others are. In any case, software rendering in real time is probably going to be pretty bad because the software has to cut corners and take shortcuts in order to be able to do it in real time. Some capture cards offer hardware based realtime encoding, but at the consumer level even these aren't really great.

No: Save the conversion to MPEG as the last step of the editing chain. While you are editing in AVI there will be (almost, except of a few situations) no loss in quality. If you start with MPEG files then these must be decompressed and recompressed each step of the way. Not only will the quality loss build up noticeable even with the first generation, but the process is so slow that you'll think you're editing on an old 20MHz 386SX computer. If you value your time at all then you'd be better off buying as much hard drive space as you need to capture and work in DV .avi instead of MPEG.
hbwerner wrote on 12/7/2003, 1:13 PM
To Chienworks -
I appreciate all the work you went to in looking at loss during rendering, and for sharing it via the link.

I went through a process on one production of editing, saving back to DV tape, then having to reload that tape to edit more (hard drive failure), and then once again from the subsequent DV save for final editing (file had been screwed up during defrag). Finally back to DV tape (to transfer to VHS) and to DVD.

I'm sure I lost quality in the video in the process, and from what you say it would be in the going back and forth to the DV tape, with compressions happening each time. I see blocking and loss of sharpness in the final version. I also think the transfer to VHS is better from the final DV tape than the DVD, maybe from loss of resolution going through the MPEG2 and then writing to disk.

I thank you for clearing up for me where I am losing video quality. I see where it's best to work always from one upload from the DV tape for all the subsequent editing and renderings. Helps to not have a hard drive crash.
Bryan
Chienworks wrote on 12/7/2003, 1:29 PM
'Welcome!

However, if you're transferring DV through firewire to tape and back, there won't be any quality loss there. It's just copying the data stream with no processing. True, there may be dropouts that will cause lost frames here and there, but those frames you do successfully capture will be identical even after infinite generations of transfers.
gus2150 wrote on 12/7/2003, 3:14 PM
Thanks again - I thought that would be the case, especially #3. I was in a computer store the other day (shopping for MovieStudio) and met a gentleman looking for the same. He seemed like an expert on a lot of things, and one of the things he told me was to ALWAYS capture directly to MPEG-2, and do ALL your editing in MPEG-2. I'm glad I didn't take his advice, and I appreciate your setting this straight.
Yes, my connection is 1394 and now I'm ready to go to work. I will have no more questions! ha!