how to cram a 3 hour vhs onto a dvd

bwanaaa wrote on 3/6/2007, 10:17 AM
After capture, this vhs comes to 10.7 gig (although the wmv file is <3 gig). When rendering the video to mpg, the standard settings are used (~8mbps). Has anyone figured out how much this kind of video can get compressed without appreciable degradation? For example, to get it to 4.44 gig (and allow some overhead for the dvd menus), DVDA specifies a bitrate of 3.5 mbps. Rather than reinventh the wheel, what do people here do when capturing vhs.

Comments

johnmeyer wrote on 3/6/2007, 10:45 AM
There have been DOZENS of posts about this in the past few months, and HUNDREDS in the past few years. You'll get a lot more answers, more quickly, by searching. Here is one thread that is active -- in this forum -- right now at this moment:

OT: How Much Will Fit?

Here are a whole bunch of other posts:

This one is from two weeks ago:

Fitting 2 hours on DVD??

and this from a long time ago:

Fit Video to DVD

Here's another person who asks pretty much the identical question to yours:

DVD Video Length?

I could keep going, but you can do the search.



TLF wrote on 3/7/2007, 1:28 AM
VHS is low quality. First, you can transfer it to your computer at a resolution of, roughly, 300x200 (whatever is nearest to that).

That will enable you to cram about 3hours onto a DVD - so long as the DVD is set to be the same resolution.

Worley
GeorgeW wrote on 3/7/2007, 3:39 AM
<<<
VHS is low quality. First, you can transfer it to your computer at a resolution of, roughly, 300x200 (whatever is nearest to that).

That will enable you to cram about 3hours onto a DVD - so long as the DVD is set to be the same resolution.
>>>

That won't really help -- you really have to adjust the bitrate accordingly (CBR or VBR-Average Bitrate ~3100kbps with AC3 audio -- for DVD5 disc).
johnmeyer wrote on 3/7/2007, 7:46 AM
Yes, it is a common confusion that lowering spatial quality changes the amount of video that can be put into a given space. The ONLY thing that determines this is the average bitrate.

Average bitrate determines the size of your encode. Period.

However, as you lower the bitrate, you will begin to notice more and more problems when things move quickly across the screen. If you lower the spatial quality by reducing the frame size from 720x480 (NTSC) to 352x480 (which is a legal DVD size), you will then have more bits to encode the movement and therefore see fewer artifacts around moving objects at any given bitrate. However, you will also lose details in each frame. It's a tradeoff.

Since VHS, in theory, cannot store more than 352x480, reducing to that resolution should not harm the detail. In practice, I have found that you will suffer some small loss of detail if your VHS original is SP and in really good shape. The only way you can know is to encode 30 seconds or so at 720x480 and the same at 352x480 -- both at the same high bitrate (choose a really high bitrate for this test because you want to test the spatial quality independent of the temporal quality). If you can't see any difference, then by all means encode at 352x480.

bwanaaa wrote on 3/7/2007, 11:36 AM
thank you for your replies. Yes the search function is a wonderful thing and the threads you kindly pointed me to have some basic background information that is helpful. but this is SUCH A COMMON TASK, I would have thought there would be a sticky for it.
(For example, the trick of connecting the vhs to a camcorder and then taking the firewire cable to the pc is a nice way of digitizing analog video. Personally, i am going directly into my ATI video card and the audio is going into the line in jack.)

My issue is the compression and rendering cycle. I have been studying on this issue a bit. It seems that noise in the image really confuses the compression algorithms-noisier sources compress much more poorly and with poorer quality than clean sources. To minimize the compression artifacts, I'd want to minimize this by 1)-noise reduction after capture 2) saving the capture file in an uncompressed format for import to DVD architect. This way the compression step in vegas is avoided.

To wit, What do you all recommend as the best noise reducer?
And, which format is should be used after capture?

Also, another stupid question, the 352x480 size that was recommended is confusing me. Wont the picture be squished? how would i setup up the vegas capture/render to achieve it.
johnmeyer wrote on 3/7/2007, 12:01 PM
Answers:

1. Capture at 720x480 (assuming NTSC). Then downsize to 352x480 during render. Better quality that way. Use Best render quality whenever resampling to a different spatial resolution.

2. Noise reduction will "free up" the bits to encode REAL video rather than chasing random dots around the screen. Therefore noise reduction can help a lot. Which one? There are dozens of posts, lots of them by me. The simplest is to use Mike Crash's temporal denoiser. Keep the settings very low if you use this (you'll think it is doing great stuff when you turn the strength way up, but later you'll be sorry you did when you start to spot all the artifacts it introduces). My favorite, by far, is an FFT denoiser, but you have to learn and use AVISynth, and not many people are up for that.

3. Don't bother with uncompressed format. Since DVDA is going to have to render to MPEG-2 in order to go to DVD, it makes a LOT more sense to just simply render to MPEG-2 from Vegas. There are no downsides, and actually plenty of upsides to doing the MPEG-2 encoding in Vegas. Make sure you use one of the DVD Architect templates in Vegas.
MPM wrote on 3/7/2007, 4:45 PM
", i am going directly into my ATI video card and the audio is going into the line in jack.)"

If possible (ATI drivers & MMC are a mess I know), capture at top CBR bitrate mpg2 -- should be 25. Overkill for VHS, but as you're going to recompress, probably resize, you'll not lose any of the original data so if you've got the disk space, why not? Using ATIs mpg2 you'll also avoid some potentially costly color conversions.

"It seems that noise in the image really confuses the compression algorithms-noisier sources compress much more poorly and with poorer quality than clean sources."

If it helps, encoders are going to try to find the max amount of data they can throw out. One way they do this is to look for solid, unchanging areas of the picture. If there's noise, then that same block gets recorded with more detail. One of the things you have to watch for is de-noising filters doing too good of a job -- areas can get too similar, in extreme cases almost cartoonish.

"the 352x480 size that was recommended "

A quick ATI tip: If considering less than final full frame video, capture a test at the reduced frame size, & try it with remove 3/2 pulldown enabled (still to max bit rate mpg2). A LOT depends on your software versions & card, but you may find you get excellent results -- better than full frame even. I've seen the pulldown removal work very, very well, but only at reduced frame sizes. [go figure?] If it works, use DGIndex with force film, then VFAPI to create a file that will open in Vegas. Treat it as 24P, encode with pulldown to mpg2 -- you will save space, reducing file size for a given bit rate.
bwanaaa wrote on 3/8/2007, 4:12 AM
thank you for the insightful replies. I do not use mmc because the software is so buggy. Rather, this is a plain vanilla 1900 xt card. The cables allow s video/composite in/out and the sony application as well as windows movie maker and dscaler recognize the input. i use svideo out from the svhs deck but am unclear how to capture 352x480.

wont this squish the video? would be happy to try if someone pointed out the relevant settings in vegas to tweak. i just dont want to mess with the template. if i use anything other than the dvda ntsc template, the image is no good.
MPM wrote on 3/8/2007, 7:43 AM
RE: ATI...
If your card will run MMC, & if you don't need the latest drivers, you can find info on a stable setup using prior versions at rage3d.com in the multimedia forum section. Since you have VIVO I'd guess you have a 3rd party card using the ATI chipset -- many of these have MMC support. You might also check for AVIO support given the 1900, which if present provides hardware encoding &/or conversion.

MMC lets you capture to mpg2 among other choices, & allows the highest bit rate mpg2 capture that I'm aware of. Recently they also use a version of a Cyberlink program, & I believe offer it for sale pretty cheap -- it has poor tuner support, but does mpg2. Either way they're the most economical mpg2 capture programs I'm aware of -- the next choice would be from MainConcept. Another popular option is to use Microsoft's MCE if your card is supported. Again, check out the rage3d forum.

There are several capture apps available at places like videohelp & doom9.org. but you'll want/need to choose a capture codec. These sites have plenty of info too, including more details on capture & final frame sizes. Extremely oversimplified, a std. SD TV shows you a 640 x 480 video frame. Analog capture involves sampling each line *X* number of times - like dividing it up with a ruler. You can take a lot of samples and get 720 pixel wide video, or fewer and get 640, 480, 352, or 320. Because every field only uses half the 480 lines, you can also cut the height in half.

Using the more well known example of audio files, you can have a sampling rate of roughly 44, 48, or 96 and so on... CDs use 44, DVDs 48, studios 96. The studio tracks at 96 hold the same audio as the CD version they release -- it's just more accurate. With video it's possible to fill a SD TV screen with 320 x 240 -- it's just not as accurate as D1 720 x 480/576. And since there's less data per frame, you're files going to be smaller.

Putting this into perspective, when they came up with the 480 x 480 frame size originally for SVCD, their research showed that the average TV only received roughly 480 samples worth of video for each line - the rest of the data was lost because of things like cabling. An SVCD if you've ever watched one, comes close to DVD quality. Will 352 quality be the same? No... Will the files be smaller? Yes. Does video compression hurt quality? Yes -- the more compression, the more data loss, the poorer the quality. At some point the loss from increasing compression exceeds the loss from reducing the frame size.

Again apologies because this is WAY oversimplified. You will find lots of info at the 2 sites mentioned -- it would help you to research. Generally when trying to fit a lot of video in a small space it's common to reduce frame sizes, use more efficient compression abandoning DVD format, or increase the amount of space, i.e. go DL.

"am unclear how to capture 352x480."
In your choice of capture app, just set the capture frame size.

"relevant settings in vegas to tweak."
Select render to mpg2, select DVDA template, click the custom settings button, fill in your frame size, set your field order [probably top 1st if analog capture], set your average & max bit rates, click OK, click OK.