Advice for converting Analog VHS tapes with Quality in Mind – Part II

[mguitonxlt]

With a lot of thanks to those giving advice on this site, I feel like I am ready to start making high quality digitized copies of my family VHS videos (~200 to 300 hours). These are full sized recordings from the 80s and 90s by a Montgomery Ward Signature 2000 VHS Camcorder Model 10687 (low end camcorder). Almost all of the tapes are in very good condition, probably in SP. The video signal is EIA Standard (525 lines, 60 fields) NTSC color signal, recorded using a helical scanning system with 4 rotary heads. Audio used 1 track. Pick-up system is sequential color difference, field reverse system. My work flow is:
1) Play tapes on a JVC HM-DH40000U VCR
2) S-Video connects VCR to AVT-8710
3) 2nd S-Video connects the AVT to a ATI TV Wonder 600 USB
4) Audio left & right cables go from the VCR to the ATI 600 USB
5) ATI 600 USB connects to my computer (AMD quad-core processors/Vista/4GB Ram/400GB free on HD).
6) VirtualDub 1.9.8 with Huffyuv codec installed

The good news: VirtualDub (VD) with Huffyuv creates great quality digitized AVI files with 0 frames dropped. When I play the captured files in VirtualDub or Windows Media Player, I am pleased with the quality, which for me, justifies all this extra effort.

The bad news: if I veer off this narrow workflow path, I either run into errors or do something that doesn’t make sense to me. Also, VD has so many features that I don’t understand. I am worried about not having the right knob turn on or off. I feel like I am flying a jet without any flight training. So I was hoping someone could help me with the following issues/questions:

A) I capture my video using the workflow above to a file, but without using Huffyuv. I exit VD. I can now play this AVI file in Windows Media Player with no problems. However, when I go back into VD and open the video file, I get the following error: “Error decompressing video frame 0: Cannot decompress video frame: the video data is too short (688320 bytes, should be 691200). If I capture the same video, but use Huffyuv compression, I get no error when replaying VD. Any idea on why VD cannot read uncompressed AVI?

B) I capture video using VD with Huffyuv. I exit VD, then return. I open the video file to delete a segment. When I save as AVI with Huffyuv, the resulting file seems to be almost twice as large. I thought cutting out a piece would make it smaller. What did I do wrong?

C) I was capturing my video using 720*480. Would it be better for me to use 352*480 in order to save space? Or will I lose quality? In my test case, the 352*480 seemed distorted – too tall, not wide enough. Is there something additional I can do to make the 352*480 video look better on my computer if that’s the right way to go?

D) I have my audio at 44.1 kHz, mono. Should it be at 16 bits or 8 bits?

E) When capturing my video with VD, should I use Overlay? Crossbar (and if yes, do I want to link related streams)? What levels are best? Should I enable noise reduction? Threshold? Vertical Reduction? BT8X8 Tweaker? Under Capture, what timing options should I have on or off?

There are so many VD functions that I don’t really know what to have on and what to have off. If anyone can help me on this, I would appreciate it.

[admin]

I want to give a quick/partial answer...

A -- Something isn't right. Consider capturing with Huffyuv instead of fully compressed. Quality-wise, the video should be identical. Huffyuv is lossless compression -- it doesn't lose quality (in theory). The theory is way over your head, and some of it is even above mine, but it's data compression, not image compression like MPEG-2, H.264, etc. As crazy as it sounds, sometimes Huffyuv works better than fully uncompressed video. I ran into this recently doing a favor/project on-site, using freeware only, and Avidemux would only accept Huffyuv.

B -- Something may have changed with the compression settings. Are you sure you're using the same compression (or lack thereof) for both the source and output? Resolution changes would also affect this. Lossless and uncompressed video uses a set data per pixel. A 352x480 source saved as a new 720x480 video would be huge when compared.

C -- 720x480 is "too wide" viewed as square pixels. 352x480 is "too tall". However, video uses rectangular pixels. A 640x480 is true 4:3, and appears proper. To change the view of aspect ratio in VirtualDub, right click on the preview image, and change it to 4:3. The x352/x720 number is the same axis as the VHS "240 lines" (optimal). In other words, even with theory applied, and considering a proper analog>digital geometry conversions, VHS has an ultimate max theoretical resolution of around 360x480 or so. In practice, it's generally far less -- especially on tapes longer than SP mode, and on inferior standard grade tapes. 352x480 almost always captures all the actual image detail/data as found on the VHS tape. 720x480 is generally just wasting disc space for DVDs made from VHS tapes, as it should have twice the bitrate as 352x. I'd rather use 352x480 and over-saturate MPEG-2 bitrate.

D -- It really needs to be 48kHz, 16-bit, stereo.

E -- Overlay is just a preview mode. Use it if it works with your card. Crossbar is (going from memory, as that computer is shut off) where you control some of the input settings: PAL vs NTSC, resolution, s-video vs composite, etc. Leave the other options alone -- NR, levels, etc, is best done after capture if the method is software based. The BT8x8 tweaks are for BT8x8 based capture cards only.

Some of this may require an expansion of our current VirtualDub capturing guide. That guide is here: How to Capture AVI Video in VirtualDub
I'll look it over, and propose some updates.

[mguitonxlt]

I continue to play around with my workflow, and your comments have been helpful. In response to your answers:
A) My workflow works when I use Huffyuv when VD capturing. However, when I capture to my computer without any compression, and then I play the file, VD gives me the error. So I don't believe my file is fully compressed.
B) I may have found my mistake. I was using VD to open a file. I would highlight a segment. Then, with full processing mode (the default) selected, I would save the segment. For some reason, that creates a much larger file even if I do it with Huffyuv turned on. My mistake seems to go away if I simply select VD's direct stream copy tab.
C) This helps. I find that 640x480 looks great on my computer, but slightly wide on my TV. While not perfect, it looks better than 720x480 (way too wide on my digital TV) or 352x480 (way too wide on my computer screen).
D) Ok. I believe you. I just thought, after readying your site's source guide, that VHS source audio was 44kHz and that's what I should be using. But I may be mixing apples with oranges.

[mguitonxlt]

I noticed I didn't ask any questions. Most importantly, is it correct to use the direct stream copy tab in VirtualDub when you want to cut out segments of an Huffyuv compressed AVI video file?

It seems right but I wanted to ask you guys for confirmation.

[lordsmurf]

Questions? We don't need no questions! (We read minds here. )

Use 'direct stream copy' any time you don't want to re-encode the video. That only works on AVI input, of course, such as Huffyuv.
(While you can open MPEG files and other formats with input filters, the 'direct stream copy' option should be unavailable.)

If you open a video, make changes, and 'save' it, then you'll be re-encoding to whatever codec was selected for the project.
Or the uncompressed option, if no codec was selected.