Am I losing quality?

[Adam Smadaf]

Hello, All.

It's my very first post here; I hope I'm following all the rules.

There's one rather basic thing I'm wondering about; but I will provide what may seem to some to be extraneous details, because I usually find that that precludes a time-consuming exchange in which those trying to help me have to ask for more information, and I have to give it for them, before they can offer a useful response.

BACKGROUND

For years, I've had use of a Digital8 camcorder with which I hoped to make good digital copies (on hard disk / optical discs) of my analog NTSC Video8 tapes. Now that I finally have a FireWire card and a FireWire cable, I can get the capture quality that the camcorder's USB connection didn't provide.

GOAL

My first goal is simply to archive to single-layer DVD-Rs the unaltered files that result from using Sony Vegas 7.0's included DV capturing function, which is Sony Video Capture 6.0.

FIRST METHOD TRIED

Because I want to burn the DVD-Rs in ISO form, which can't handle any file of more than 2 GB, I decided to divide the capacity of a single-layer DVD-R by four and capture the footage from my tape in segments that took up just a bit less than a quarter of a DVD-R's capacity, so that each DVD-R would hold four such files. In my initial experiment, this meant files that each were 5 minutes and 18 seconds long. I captured a whole two-hour tape in this way. To ensure that I got everything clearly, each segment overlapped its predecessor and follower by a few seconds. Altogether, there were 24 such files.

When I went to burn four of these to a "DVD-ROM (ISO)" in Nero Burning ROM, I learned that they were slightly beyond the capacity of my DVD-R, and that the only way to try fitting them all on the disc was to overburn, which I found unacceptable.

When Windows XP Professional displays the Advanced version of the Summary tab in the Properties panel for one of these AVI files made by Sony Video Capture 6, it gives only this information:
Image: Width, 720 pixels; Height, 480 pixels
Audio: Duration, 00:05:18; Bit Rate, 1024kbps; Audio sample size, 16 bit; Audio format, PCM
Video: Frame rate, 29 frames/second

At least until yesterday afternoon, I was under the impression that these files involved no lossy compression—that they could be treated sort of like the video equivalent of .bmp, .png, or .tif, in terms of one's ability to trim and augment them and save the results without any reduction in quality compared to the original file.

Anyway, I didn't want to waste nearly 25% of each DVD-R's capacity by burning only three such files to each disc. So I decided to experiment: my plan was (1) to put four of the capture files together in Vegas; (2) to cut out the overlap so that they essentially ran as one long, 20-minute video; (3) to set a loop region of 5 minutes and 12 seconds (this, multiplied by four, seemed likelier not to exceed a DVD-R's capacity); (4) to overlap these loop regions by 1 second; and (5) to render each loop region as a new AVI, with no loss in quality compared to the original file, that I could burn to DVD-R.

When I went to render the first such AVI as described in point 5 in the previous paragraph, I chose, in the "Template" area of the "Render As" dialog box, "Default Template (uncompressed)", which has 44.1-kHz, 16-bit, stereo audio and 29.970-fps, 720×480×32 video.

Well, before Vegas had even finished rendering the 5:12-long file, it was 10 GB, nearly 1,000% as big as I wanted it to be.

I compared the properties of this new rendering with those of one of the files created by Sony Video Capture 6, and noticed a few differences.

I tried again, and succeeded in making a rendering that matched the properties of the original capture files—by instead using the "NTSC DV" template and changing its audio sample rate from 48 kHz to 32 kHz. The speed with which Vegas rendered such files left me with the feeling that a loss of quality was unlikely.

But the great file-size difference that depended on whether I chose to make a "Default Template (uncompressed)" AVI or an "NTSC DV" AVI got me suspicious about (1) whether I was getting the highest-quality capture that my equipment would provide in the first place, and (2) whether I was losing quality by stringing the captures together, paring them down, and re-rendering them.

Wikipedia says that the DV standard's compression is lossy.

As far as I can see in Sony Video Capture 6, there is no way to have it capture to anything other than an "NTSC DV" AVI.

MY ACTUAL QUESTIONS (AT LONG LAST)

1. Am I creating a second generation of quality loss by using the method described above?

2. If I am, do I have any option other than simply to capture from my tape all over again, in shorter segments?

I've halted the project for now, until I have more certainty about this, because I don't want to waste time capturing more tape, and waste DVD-Rs by burning files to them that don't have the full quality of the original AVIs that Sony Video Capture made from my tapes.

Thanks to all those who've taken the time to read my epic epistle and can help.

Adam Smadaf

[John Cline]

DV is indeed a lossy format. It uses approximately 5:1 compression with 4:1:1 color sampling. The compression algorithm is similar to JPEG image compression. The 4:1:1 color sampling is going to give you trouble on highly saturated solid colors (particularly red) on solid color backgrounds, especially if these regions are diagonal. Nevertheless, DV is a reasonable compromise between image quality and disc space.

If you capture as DV, trim the project as DV and make absolutely no changes to the files by color correcting or adding titles and then render the files as DV, Vegas will "smart render" the files with no loss whatsoever. It will just be copying the trimmed data and no actual rendering or recompression will take place.

By the way, DV audio can be either 32k non-linear 12bit or 48k 16bit stereo. This can be set in the camera, 48k 16bit is preferred. Vegas defaults to 44.1k, which is incorrect. You must make sure that the project and the render is set to 48k 16bit stereo.

[Adam Smadaf]

Originally Posted by John Cline

DV is indeed a lossy format. It uses approximately 5:1 compression with 4:1:1 color sampling. The compression algorithm is similar to JPEG image compression. The 4:1:1 color sampling is going to give you trouble on highly saturated solid colors (particularly red) on solid color backgrounds, especially if these regions are diagonal. Nevertheless, DV is a reasonable compromise between image quality and disc space.

If you capture as DV, trim the project as DV and make absolutely no changes to the files by color correcting or adding titles and then render the files as DV, Vegas will "smart render" the files with no loss whatsoever. It will just be copying the trimmed data and no actual rendering or recompression will take place.

Thanks, John, for your great reply. I had expected to see something like JPEG's/MPEG's 'blockiness' in these files, but hadn't yet observed it—so, yes, certainly a reasonable compromise. Perhaps I would see it if I were to zoom in greatly without anti-aliasing.

Your description of saving trimmed but otherwise unaltered footage in Vegas seems to confirm my hopes, which is a relief. Does your description hold true when the new rendering contains data from two different DV capture files simply trimmed and butted together but otherwise left untouched?

Also, would I be right in thinking that, because my camcorder, in converting the Video8 tapes, is outputting a DV signal, there would be no point in capturing to a file with any greater quality?

[Adam Smadaf]

Originally Posted by John Cline

By the way, DV audio can be either 32k non-linear 12bit or 48k 16bit stereo. This can be set in the camera, 48k 16bit is preferred. Vegas defaults to 44.1k, which is incorrect. You must make sure that the project and the render is set to 48k 16bit stereo.

Just saw your audio addendum. I will investigate those options with my camera, a Sony DCR-TRV480 NTSC. Although the sound on my tapes is the analog of Video8, I'd like to capture it as faithfully as possible: do you suppose that I'm missing anything by using only 32k, rather than 48k?

[John Cline]

DV (as opposed to MPEG2 or AVCHD) uses intraframe compression, which means that each individual frame in the file stands alone and can be freely cut and trimmed. Yes, you can cut and trim as many different DV files as you want.

Also, while DV compression is similar to JPEG, it is much more sophisticated than JPEG and the blocky artifacts are reduced as a result. There is a video codec which uses straight JPEG compression and it's called "Motion JPEG" or MJPEG. It's been around longer than DV and it has a couple of advantages, you can adjust the quality just like you can with JPEG and it uses 4:2:2 color sampling. Regardless, you are capturing DV and you would gain nothing by converting it to another codec. The DV compression is being done in the camera and output that way via Firewire, stay with DV.

Yes, absolutely, you will be missing quite a bit by using the 32k sample rate. Set it to 48k 16bit stereo. The TRV-480 has this setting.

[Adam Smadaf]

Originally Posted by John Cline

DV (as opposed to MPEG2 or AVCHD) uses intraframe compression, which means that each individual frame in the file stands alone and can be freely cut and trimmed. Yes, you can cut and trim as many different DV files as you want.

Also, while DV compression is similar to JPEG, it is much more sophisticated than JPEG and the blocky artifacts are reduced as a result. There is a video codec which uses straight JPEG compression and it's called "Motion JPEG" or MJPEG. It's been around longer than DV and it has a couple of advantages, you can adjust the quality just like you can with JPEG and it uses 4:2:2 color sampling. Regardless, you are capturing DV and you would gain nothing by converting it to another codec. The DV compression is being done in the camera and output that way via Firewire, stay with DV.

Yes, absolutely, you will be missing quite a bit by using the 32k sample rate. Set it to 48k 16bit stereo. The TRV-480 has this setting.

Thanks again. Sorry about asking so many questions.

Intraframe compression: now I understand what that means.

"stay with DV": good.

48k 16-bit stereo audio sample rate in TRV-480: the only setting I've found so far is one in the "Audio recording system" (that's how the manual's "Specification" section describes it), which I've changed from 12 bits ("Fs 32 kHz, stereo 1, stereo 2") to 16 bits ("Fs 48 kHz, stereo"). Just to confirm: changing this setting will affect the quality with which the camera samples the analog sound on Video8 tapes to add to the DV signal sent to my computer; right?

[Mike Kujbida]

Originally Posted by Adam Smadaf

48k 16-bit stereo audio sample rate in TRV-480: the only setting I've found so far is one in the "Audio recording system" (that's how the manual's "Specification" section describes it), which I've changed from 12 bits ("Fs 32 kHz, stereo 1, stereo 2") to 16 bits ("Fs 48 kHz, stereo"). Just to confirm: changing this setting will affect the quality with which the camera samples the analog sound on Video8 tapes to add to the DV signal sent to my computer; right?

Sorry but no.
That setting only affects what gets recorded and does nothing to material already on tape.
If it was set to 12-bit, that's what your audio got recorded at.

[Adam Smadaf]

Originally Posted by Mike Kujbida

Sorry but no.
That setting only affects what gets recorded and does nothing to material already on tape.
If it was set to 12-bit, that's what your audio got recorded at.

It seems there may have been one point of misunderstanding.

These are Video8 tapes, not Hi8 and not Digital8. They're completely analog. They have what counted as "hi-fi" stereo sound for Video8 in the mid and late 1990s. The presumably relatively low audio quality of these old recordings is why I asked whether I was missing out by digitizing the sound at only 32 instead of 48.

In trying to find the camera's 32k-versus-48k setting that John mentioned, I found only one setting that seemed perhaps to be the one he had in mind; it is in the record settings. This seems certain to affect the audio with which the camera records sound onto Digital8 tapes when it is MAKING digital recordings. What I'm wondering about is whether this setting also affects the quality with which the camera digitizes the analog audio that it plays back from my old Video8 tapes on its way out the FireWire port to the computer.

[John Cline]

Yes, it does affect the sample rate that the Digital8 digitizes old analog recordings. You found the right setting.

[Seth Bloombaum]

Good advice all.

Adam, one suggestion on workflow - it seems to me that you're probably finding capturing 5:18 segments one at a time somewhat tedious.

You might consider just capturing whole tapes, putting up the clip on the timeline, and dropping markers or regions every 5:18 and render those selections. Probably go much faster, with less cycling of your old tapes - which is a good thing!

If you do use regions, investigate the batch render script if you are using a version of vegas that supports scripting (basically, all non-"studio" versions). This can really speed up these sorts of repetitive tasks.

I'm always thinking about workflow...

[Adam Smadaf]

Originally Posted by Seth Bloombaum

Good advice all.

Adam, one suggestion on workflow - it seems to me that you're probably finding capturing 5:18 segments one at a time somewhat tedious.

You might consider just capturing whole tapes, putting up the clip on the timeline, and dropping markers or regions every 5:18 and render those selections. Probably go much faster, with less cycling of your old tapes - which is a good thing!

If you do use regions, investigate the batch render script if you are using a version of vegas that supports scripting (basically, all non-"studio" versions). This can really speed up these sorts of repetitive tasks.

I'm always thinking about workflow...

Thanks, Seth.

Capturing longer segments at a time: exactly what I had in mind, once I learned that I could then cut them up without quality loss.

I am nowhere near having explored all of Vegas's functions (though I have used it for more than just capturing), and so don't yet know about regions (beyond setting a loop region and having only it rendered) and scripting; but I certainly will investigate them. With about sixty 8-mm tapes to digitize, and then well over a hundred VHS ones (through the Digital8 camcorder), I'll be glad of anything that reduces my involvement in the details at this stage, for there will be plenty of detail work to do later.

Thanks again.
Adam