Quality loss VHS > CVD, DVD

[rglmrj]

I am following the developments of capturing and encoding for a few years on this site.
I have a lot of family VHS tapes (PAL), some 14 years old with noise, which I want to secure for the future.
I have already done a lot of captures (vdub + huffyuv + filters) and encodings (tmpgenc + filters) for test to VCD, SVCD, CVD and DVD.
The results were always not so good as the original,
also when I capture in 720x576 and encode to DVD with highest bitrates.
I have read that many people said that the results were as good as the original.
Do I something wrong or must I live with a quality loss?

[spectroelectro]

Your report of quality losss seems puzzling to me, since I have seen a lot of credible folks report that the video quality actually _improves_ when they do a competent transfer from VHS to DVD. In fact that was my initial experience with transferring VHS to DVD. It still is. To my eyes, a VHS tapes competently transferred to DVD looks somewhat better than the original VHS tape in a number of ways.
Let me be specific:
[1] VHS tapes I have transferred to DVD lack the "edge crawl" due to the way composite signals are recorded on analog tape. On DVD, edges look sharper and cleaner than HQ playback on VHS.
[2] Background noise on VHS tapes reduces somewhat because the process of MPEG-2 encoding seems to smooth it out in an averaging process that works like a slight low-pass video contrast filter.
[3] Colors seem more vibrant on a DVD transfer, perhaps because of the way colors are stored and reproduced on DVD.
That said, there are some slight degradations when transferring VHS to DVD, or even when cpaturing from off-the-air broadcasts.
[1] One disadvantage is that the gamma is never right. I see a compressed gamma range with VHS transfers put on DVD -- shadows and dark regions tend to disappear into monochrome blotches. This can often be elminated by chaning the gamma settings in TMPGEnc, but it must be done on a per-program basis, and takes time and effort.
[2] There can be some digital artifacts during very rapidly-changing scenes even with bitrates as high as 9000. Momentary, but still visible if you look closely. That's just the nature of MPEG-2 encoding -- no way to fix it.
---
Having said this much let me suggest some reasons why you might not be getting good quality VHS to DVD trnasfers:
You must use a high quality video capture unit. The typical $60 PCI card in your computer will produce poor-quality captures as best. To get high-quality video, your capture device must be a high quality unit. That measn you must spend at least $200. Excellent quality capture devices include: Canpus ADVC-50 and ADVC-100, Datavideo DAC-100, Pinaccle DV-500+, or any digital camcorder with analog A/V inputs.
Poor quality video capture cards include any ATI Radeon card no matter how expensive, any of the Leadtek models (viz., Winfast TV200 or PV-100), any of the PV250/PV350 real-time MPEG-2 converters. Mediocre middling-quality capture devices include the Dazzle series. I would avoid all these units. I installed a Leadtek card in my video capture just to see how bad the video capture woudl be, and boy! It was BAD. The difference twixt that and my Canopus ADVC-100 is like night and day. Almost as gross as the difference twixt VCD quality video and DVD quality video.
You must use the very finest MPEG-2 software encodeer you possibly can. That's TMPGEnc, hands down. Fortuantely, it's cheap. (AVOID all rael-time or hardware MPEG-2 ecnoders, they cannot compete with non-real-time software multiple pass encoders for high video quality.)
You must use a high quality VCR to play back the video which you wish to capture. Using a VCR with digital noise reduction and pro head technology with pre-amps built into the heads makes a world of difference. Gargage in = garbage out. Sony makes a model of VCR with a type of digital video noise reduciton that claims to restore video tape pictures and does a very good job. Toshiba makes VCRs with a digital video noise reduction feature they call DNR perfect and it works very well. JVC used to make a high end VCR with digital noise reduciton using either a 2-meg or 4-meg video buffer, and it also works very well. Remember that you cannot expect high quality video out from a $79 K-Mart VCR, especially if you plan to feed the video into a $1200 computer using $300 worth of video capture peripherals. That is like trying to make an audiophile CD recording by placing a microphone in front of a $29.95 boom box.
Perhaps most important of all, you M*U*S*T use a digital Time Base Corrector in between the video output from your VCR and your comptuer digital video input. This is crucial for avoiding dropped frames, which can drasticaly degrade encoded MPEG-2, and may interrupt the entire video capture process.
---
Point is, superb quality DVDs can be made from VHS, even very old VHS tapes. I recently made several DVDs from program material recorded in 1978 (yes, 1978). While not as good quality as DVD, it looks surprisingly good.
---
One last issue: you may find that you can drastically improve the video quality by doing some processing on the video. LordSmurf has a superb guide where he discusses digital video processing of older VHS tapes. This is often the crucial difference twixt a poor-looking DVD trnasfer and a great-looking one.
Avery Smith, the author of VirtualDub, has pointed out that noise riding ont he chroma channel can drastically muchk up MPEG-2 encoding. It is invisible in a TV picture but is there nonetheless, lurking out of sight. Problems like that can be almost eliminated by running digital filters on the video.
VritualDub is invaluable for doing such digital video processing. You may either frameserve using VDub (search in this forum to find out more about frameserving) or you can open the AVI file with VirtualDubMod or one of its variants and then save to a lossless AVI format like Huffyuv and then simply run TMPGEnc on the huffyuv AVI file.
The temporal noise filter in VirtualDub works wonders!
The chroma filter is also magnificent if used prior to MPEG-2 encoding.
AviSynth also does similar tricks with video, using even more sophisticated video processing filters. The Smart Smoother and Smart Sharpener can also work wonders. You'll many far more complex and subtle video filters on the web written by third parties: FairyDust, PixieDust, and many others. Try 'em and see which ones produce the best results for you!
The final and most subtle method of improving video quality in VHS captures involve square root noise reduction. Make multiple captures of the same VHS tape, then add 'em all together using VirtualDub video procesing. This reduces the noise as the square root of the number of additive signals. What is going on is that you have a video signal masked by noise. The video signal remains relatively constant, while the analog noise changes since it is white noise generated by thermal fluctuations. The average amplitude of the thermal noise tends to cancel out to zero if you add enough captures together, so the relatively constant signal emerges form the noise. The price you pay is some distortion in chroma and luminance but in my experience it's not visible.

[rglmrj]

spectroelectro,

Thanks for your explanation.
So I must invest in better hardware. Better capturecard, better VCR and TBC. How many will this cost?
Maybe its better to look for somebody with such a equipment and to pay for it?
Which of these hardware will do the most in quality?

[mic_]

Poor quality video capture cards include any ATI Radeon card no matter how expensive

Umm....what? Have you actually used one of the AIW cards? And if so, which one?

[lordsmurf]

Poor quality video capture cards include any ATI Radeon card no matter how expensive

Umm....what? Have you actually used one of the AIW cards? And if so, which one?

ATI AIW Radeon cards poor? Well, I beg to differ. The oldest AIW Radeon cards originally cost $200 or more, and many of them still do. The 9800 sells for $400 at the moment. The ATI AIW cards are currently the top-rated cards on the market, in several consumer reports. In fact, the Canopus device you mention may be nice, but it's only DV, and it's reviews seem to vary.

I also see vast improvements in converting VHS to DVD, though much of it relies on playback equipment moreso than the computer. You need a good VCR, a TBC, and knowledge of your source and how video works.

On fact, the beauty of an ATI AIW Radeon card is the new MMC 8.x capture software, which has noise removal to eliminate graininess and other artifacts from the source. Works great on old movies.

Much of my site, www.lordsmurf.com, is dedicated to ATI capture and video and audio restoration. It's not completed yet, but I should have a major update here within a day or so. Most of the capturing, samples, and restoration pages should be up by the weekend.

[yg1968]

I used the ATI AIW 128 for capturing and did not like it at all. All I got is dropped frames and out of sync issues. So I bought a camcorder with a passthrough feature. I have been using that ever since.

[indolikaa]

@rglmrj,

I am in the process of converting hundreds of VHS and Beta recordings to optical media. Most of the recordings were made between 1979-1985. All of my transfers are as good as the original VHS tape. Many of these tapes played poorly in VCRs but have been ressurected through the transfer process.

I pass the VCR signal through my Digital8 camcorder and into the computer. This method has work flawlessly for many weeks now, whereas other capture cards were freezing up randomly. I have since learned that the reason the D8 process works is because the camcorder has a Time Base Corrector built-in. After discussing the matter with 'Those Who Know' the TBC seems to be the critical link in the process.

I author VCDs as well as DVDs with this method, and I'm very happy with the results.

.indolikaa.

[rglmrj]

@yg1968

I have never drop frames. Your cam is that a DV and what is a passthrough feature? Do you use your AIW now with your cam?
Which cam do you have?

@indolikaa

Do you record the VHS on your cam or is it only passthrough? Do you use also a capturecard or something like firewire?
Which cam do you have?

I have a 8mm Sony, not suitable for this.

[indolikaa]

@indolikaa

Do you record the VHS on your cam or is it only passthrough? Do you use also a capturecard or something like firewire?
Which cam do you have?

I have a 8mm Sony, not suitable for this.

I use the camcorder (DCR-TRV120) only as a passthru device. The signal passes through the camcorder to my firewire card, where WinDV captures the stream as a DV.avi file at about 13GB per hour's worth of video.

It is my understanding that Sony put both Passthru and TBC in most of their early-model DV and D8 camcorders, but now it can only be found in their upper tier of their camcorder models.

I also saw a thread this week that refers to a TBC device you can purchase for US$ 179.00 that was spoken of quite highly. I'll see if I can find it and link it to you.

.indolikaa.

[yg1968]

The passthrough feature means that the camcorder has a video in (also called analog in) and that you can pasthrough the video to your computer through your "DV out" port (without having to use a DV tape).

I have the Sony TRV-25. Sony has come out with newer models such as the Sony TRV-22 which also has a passthrough feature (the TRV-19 does not have it). These models don't have TBC included. For DV recorded tapes, a TBC is not needed given that DV has time base correction by default.

However, I don't think that this also applies to video that is captured from VHS through the passthrough feature of the digital camcorder. The passthrough feature does seem to improve the quality of poor VHS but probably not to the extent of a TBC.
On a related note, the Sony TRV-350 (a digital 8 camcorder) seems to have a TBC and DNR. I am not sure but I get the feeling that this feature only gets used when playing back analog Hi-8 tapes on your camcorder (and not with the passthrough feature). I believe that most digital 8 camcorders (with a passthrough feature) have a TBC. However, the TBC is probably not very useful if you don't have analog Hi 8 tapes. The Sony Canadian website gives more details (than the U.S. website) on the features of the Sony TRV-350:
http://www.sonystyle.ca

Here is a useful link that explains the basics:

https://www.videohelp.com/dvanalog

P.S. I use the firewire port. I don't use my AIW 128 for capturing.

[lordsmurf]

I used the ATI AIW

Yeah, that's an old card with an old chipset, so I'm not surprised. The 128 Pro is not a Radeon-class card. I believe it operates off a BT chipset.

[next]

I used the ATI AIW

Yeah, that's an old card with an old chipset, so I'm not surprised. The 128 Pro is not a Radeon-class card. I believe it operates off a BT chipset.

Actually the AIW 128 'Pro' used the Theatre chipset for capturing. The same that is in the Radeon class cards. You may be referring to the AIW 128 (sans 'pro') that did not.

I've made it a point to hang on to a few '128 Pro" cards because they used an analog tuner. As ATI moved the AIW along in its life cycle into the Radeon class they attempted to incorporate a digital tuner. These were not as good from a capture quality perspective.

I was pleased to see that on its latest high end AIW's ATI has moved back to an analog tuner for its AIW product.

Big bucks though because this (re)incorporation is new on the life cycle.

Sorry did not mean to highjack the thread - was just a comment.

[hipster]

rglmrj, I too have been at this somewhat for a couple years and can not get an acceptable looking DVD accept and only when I use Sonic MYDVD which I think seems to be somewhat lacking in the creativity department. I have tried many software suites to no avail.

[davideck]

spectroelectro - can you be more specific regarding the differences you have noticed between hardware and software MPEG2 encoding? I have been transferring SVHS to DVD using an older JVC VCR with excellent low noise playback, a FOR-A professional S-Video TBC, and a PVR-250. My DVDs do look better than the original tapes in the ways you mentioned, and also show the disadvantages that you mentioned. So what improvements would I see if I took the time to go through software MPEG2 encoding as opposed to real time hardware encoding?

[FulciLives]

spectroelectro - can you be more specific regarding the differences you have noticed between hardware and software MPEG2 encoding? I have been transferring SVHS to DVD using an older JVC VCR with excellent low noise playback, a FOR-A professional S-Video TBC, and a PVR-250. My DVDs do look better than the original tapes in the ways you mentioned, and also show the disadvantages that you mentioned. So what improvements would I see if I took the time to go through software MPEG2 encoding as opposed to real time hardware encoding?

The main benefit of software MPEG-2 encoding is when you can do a 2-pass (CCE SP even allows 3 or more pass) VBR encode. Here the encoder can adjust the bitrate so some scenes get less and some get more depending on what that scene needs to look good.

Having said that you cannot get better than a CBR of 8000kbps when using full D1 resolution (720x480/576) or a CBR of 5000kbps when using half D1 resolution (352x480/576) since those are roughly the bitrate limits of the two formats.

Also I have noticed with noisy VHS sources that often times half D1 at 5000kbps can look better than full D1 at 8000kbps though I'm not sure why that is but there does seem to be less macroblocks that way especially with a difficult source.

I recently captured an old VHS movie from the original pre-record and this was a very low budget movie with BLAIR WITCH style hand held camera movements. I saw some macroblocking at 720x480 when I used 6000kbps (which was the highest I could go and get the movie to fit on a single DVD disc). I did Half D1 at a CBR of 6000kbs which was probably overkill since I think 5000kbps is really the upper limit of half D1 but the the movie was short and even at 6000kbps it left plenty of room for authoring overhead. Anyways at half D1 CBR 6000kbps there were no macroblocks at all. I admittingly didn't try full D1 at a CBR of 8000kbps because then my movie would not have fit on a single DVD disc.

In short ... anyone who captures real time to MPEG-2 will probably get better results doing a high bitrate half D1 than a full D1 recording UNLESS you can do 8000kbps but then that only allows for roughly 60 minutes on a DVD which is not much.

- John "FulciLives" Coleman

P.S.
I forgot to add that the other benefit of using a software MPEG-2 encoding is that you can use either VirtualDub and/or AviSynth filters to help clean up the video. This can help A LOT with image quality! I personally am very fond of the Convolution3D filter for AviSynth when doing VHS to DVD transfers.

[sanlyn]

Also I have noticed with noisy VHS sources that often times half D1 at 5000kbps can look better than full D1 at 8000kbps though I'm not sure why that is but there does seem to be less macroblocks that way especially with a difficult source.

I've had the same experience with half-D1. I have ancient VHS tapes either poorly recorded or poorly stored, and the noise problems are horrific. I capture playing on a high-end VCR thru a Toshiba XS34 used as a pass-thru noise filter and line-level TBC. Sometimes I add a full-frame TBC if Macrovision is a problem. Then I go into my Ati AIW 9600XT card using VirtualDub for capture to AVI at 640x480. After a little filtering in VDub, I can go from 640x480 to either full D1 or half-D1. The Half-D1's require a (very slight) smidgin of sharpening or contrast in TMPGenc or CCE Basic. In the end, the half-D1 looks better. Biggest improvement is that most of the remaining VHS and damage noise just don't show in half-D1.

I also found that if my target is half-D1, capturing at 352x480 isn't nearly as clean or snappy as capturing at 640x480 and then encoding to 352x480. Only problem: a half-D1 shown on those mammoth 42-inch tv's isn't as solid as 720x480. If you're tied to sitting 4 feet away from a room-sized tv, you'll just have to put up with the highly visible remnants of VHS noise on full D1.