When to manipulate video: Before or After capture?

[colokate]

Hello,

As many others, I am very late to the digitizing party, but I thought I would give it a try.

The first set of tapes I want to convert are Hi8 tapes.

First the setup:
Windows 7 x64 computer (desktop)
ASUS MyCinema-EHC3150 Combo PCI card
TI 1394 OHCI Compliant Host Controller (Firewire)
Samsung SLC810 Hi8 Camcorder

I am using S-Video fed to the ASUS card from the SLC810
I am using VirtualDub 1.9.11 to capture/transfer AVI video

Now my question:
After reading many, many forum posts, I cannot determine what parameters should be affected during the capture. In other words, it is better to execute a "chocolate and vanilla" capture and save manipulation and processing with VirtualDub or Avisynth after capture? Or would be it be more efficient to tweak settings during capture? Which settings?

Thank you very much for any assistance!
colokate

[sanlyn]

Post processing is always more precise and flexible. The only filtering usually done during capture is essenntial conrol of the input signal to avoid illegal video levels, which can't be easily corrected later. You can denoise if you want, but remember that whatever side effects you get from your card'[s denoising algorithm will stay with the capture forever and can't be undone. Blurring, ghosting, and motion smearing are common effects.

Your ASUS card captures to lossy MPEG2, which is not an editable format and can't be modified without quality loss. It's not designed for post-processing cleanup. Once Hi8 defects are embedded as encoded artifacts, lower your expectations on how much restoration work you can do, which won't be very extensive. MPEG was not designed for it. It's an interframe final delivery format. If you want to use VirtualDub or Avisynth, note that they work with decoded lossless media, so the results will have to be lossy encoded again to a final format.

If your ASUS card will allow recording to a high bitrate broadcast-level MPEG at 15MBPS or so, you'll be ahead of the game slightly and can use that capture as an archive. It will have to be decoded to lossless media for any cleanup, color correction, etc., and re-encoded later.

[colokate]

Thank you sanlyn for your reply. Please forgive my noobness, but I am using virtualdub to capture a file in an AVI format as I understand that is a preferred format for editing video. With the ASUS card, does this mean that ultimately I will be unable to truly capture decoded lossless media using virtualdub and/or avisynth? I have been able to capture AVI, but I do not know the attributes of the capture (lossless/decoded).

I realize all of this is very basic stuff, and I have much to learn. Like a friend always said "It's easy once you know how."

colokate

[sanlyn]

The ASUS spec says the card is a hardware MPEG encoder,. Cards of that type encode to MPEG before output. If you can pick up that card's output with VirtualDub, it's likely already been encoded to MPEG but you can capture it to lossless decoded AVI using lossless huffyuv or Lagarith compression. Note that "AVI" is a container, not a format or a codec. You don't specify how you've set Virtualdub to save the captured signal. By default VDub saves to uncompressed RGB unless you specify otherwise, which isn't what you want.

It's usually a good idea to know at least what format you capture to.

[colokate]

I absolutely agree! Knowing the capture format is probably the most basic piece of information needed. I did not realize that AVI is a container, and does not specify the format of the capture in the container.

After reading your reply, I found two applications that may be of help in determining exactly what was captured: Gspot and MediaInfo.

Could I impose on you to take a look at the MediaInfo output below and let me know if I am indeed capturing mpeg output? Or perhaps another application may be better for this purpose?

With much appreciation,
colokate

This is a portion of the report from MediaInfo:

General
Complete name: \Clip8ASUS_1.avi
Format: AVI
Format/Info: Audio Video Interleave
Format profile: OpenDML

Video
Format: HuffYUV
Format version: Version 2
Codec ID: HFYU
Bit rate: 42.7 Mb/s
Width: 720 pixels
Height: 480 pixels
Display aspect ratio: 3:2
Frame rate: 29.970 (30000/1001) FPS
Standard: NTSC
Color space: YUV
Chroma subsampling: 4:2:2
Bit depth: 8 bits
Scan type: Interlaced

[sanlyn]

Your capture is 720x480 frame sized, losslessly compressed with huffYUV in a YUY2 colorspace. The video is NTSC @29.97fps and is interlaced. So it's a proper capture format for VHS. However, on the way to Virtualdub your video was apparently lossy encoded to YV12 MPEG by the capture device. You have lossless media in the AVI, but you have a quality loss concerning the earlier stage of MPEG encoding. That 20% or so of the original tape signal lost through encoding can't be recovered, and it will be more work to remove the initial stage of compression artifacts. But you now have lossless AVI media to use with Virtualdub and Avisynth.

You would have had the same capture if you had allowed the ASUS capture card to store the lossy MPEG recording into your computer as MPG files, then opened the MPEG recordings in Avisynth or Virtualdub and filtered and saved them as lossless YV12 for intermediate working files. That sounds like a roundabout way of doing it and isn't a lossless process -- unless there is some way you allowed VirtualDub to hook into that ASUS card's capture drivers before the unit encoded to lossy MPEG, which I doubt VirtualDub could do.

There is no mention of a line-level or frame-sync tbc in your earlier posts, so I assume you have the typical unstable,wiggly vertical edges and vertical and horizontal jitter that accompanies captures without line or frame time base correction. Those artifacts can't be repaired after capture.

[colokate]

There is a wealth of information in your post. Thank you. It is apparent that I need a different hardware solution. After reading many posts on the site, I understand folks who are well versed in capture techniques have an XP box that can handle capture hardware that was more appropriate for the task, such as ATI AIW cards. As I have only about 20 Hi8 tapes, it will be more efficient to procure a hardware solution that works with Win 7 x64 and deal with the compromises that solution entails.

My main goal is true lossless capture, from both hardware and software. Would you recommend a workflow applicable to my O/S or point me to a thread where this is discussed, as I know this is a frequent discussion in the post-XP world? You also mentioned TBC in your post. Can that be done for Hi8 tapes with a capture device, or does that entail a different piece of hardware?

Again, many thanks for your response.
Colokate

[sanlyn]

True, getting up an XP PC, home built or used (they are still around), with a legacy ATI All In Wonder and AGP or PCIe slot is still the best solution if not the easiest to come by. There are later alternatives that work with Windows 7, but many are difficult to obtain. The Hauppauge 610 USB device has Windows 7 drivers. Note that devices I mention here are 32-bit and will work with 64-bit systems. You can't use 64-bit capture drivers with 32-bit VirtualDub, so stay with 32-bit devices, VirtualDub, and 32-bit codecs and filters all the way. The Hauppauge, I see, is currently in short supply but sources such as Amazon have them on back order. The Ati 600 USB is often found used, and drivers are available from digitalfaq (the drivers are Vista and are reported to work in Windows 7.). Another alternate is the Diamond Multimedia VC500 USB, which is inexpensive, popular in many circles, and has updated drivers.

TBC: there two types of time base corrector. The most essential is a line-level tbc which corrects scanline sync within frames and which is of utmost importance if your VCR has no built-in line tbc. The other type is an external frame-sync tbc, which corrects frame-based signal timing for av sync and other problems (note that a frame sync tbc has no effect on scanline timing within individual frames, just as a line tbc has no effect on overall frame rate timing). It's possible to get away without a frame tbs, but that's an iffy proposition: most legacy VCR's don't output a constant frame rate and cause all sorts of sync problems along with dropped and inserted frames.

Line tbc's are built-in with high-end SVHS VCRs mentioned on this site. Finding one of these players that still works is an arduous and expensive task. Also, the two most recommended frame-sync tbc (AVT-8710 and TBC-1000) are problems. Current AVT's have serious quality control issues, while the TBC-1000 has been out of manufacture for quite some time. The solution for most of us mere mortals with limited budgets is a legacy DVD recorder used as a pass-thru device -- connect the VCR to the recorder, then connect the DVD's output directly to the capture device. The DVD isn't used to record, but is used to "play thru" the video and take advantage of its built-in line and frame sync ability. Not many of these machines can be used for pass-thru. The recorders recommended are the Panasonic DMR-ES10 and DMR-ES15 which still show up with great frequency on auction sites. I have a couple of non-tbc VCRs and have been using those Panasonics with them for years. They are surprisingly effective. They also have effective y/c comb filters to prevent dot crawl effects with composite cable output from VCRs. Their only drawback for some users is that their frame-sync circuits don't defeat Macrovision. But they are effective enough to prevent frame timing defects that sometimes "look like" Macrovision to capture cards. The Pannies also have excellent composite-to-s-video electronic converters inside, which get a cleaner image than with composite alone.

[colokate]

Wow! A wonderful reply that helps considerably. I am now on the hunt for a Panasonic DMR-ES10 or DMR-ES15. As you wrote, there are a few that come complete with remote on EBay. I also found a post with settings you use capturing VHS using VirtualDub. That has been very helpful to understand parameters that can be set in VD.

For grins and giggles, I captured a short clip on one of the worst parts of the Hi8 tape. It has almost everything: blurry zooms, yellow-ish lighting, shaking, sawtooth edges. I know that many of these "features" cannot be removed in post-processing. But, with my current setup (which I will change), it is about the best output I can get with my very limited level of video capture knowledge.

Thank you again sanlyn for taking the time to respond and provide very useful information to a beginner.

Workflow: Samsung SLC810 Hi8 Camcorder --> ASUS MyCinema-EHC3150 Combo PCI card --> (S-Video cable) --> Windows 7 x64 computer (desktop). I used the Lagarith encoder.

***UPDATE***
Well, I WAS going to upload a small AVI file, but the connection keeps getting reset when I am attempting to upload the file. I can upload a JPG successfully, but not the AVI. It is a rather small file, so limits are not the problem. I will keep retrying.

[sanlyn]

Thanks for the follow-up. I think you'll find image improvements once you start using the pass-thru and replace the ASUS.

If you're capturing to huffyuv, you can create and save your sample in VirtualDub using the "dircet stram copy" mode to preserve huffyuv compfression and avoid another RGB conversion. For some reason, uploads are slower than downloads. Uploads go through a network virus check on the way in so I guess that's the reason for upload slowdowns.

[colokate]

Many, many thanks for pointing me in the right direction to kickoff this project. I am now on the hunt for better equipment for the job. I will post back when it is in place and I have spent some time with the new set-up. I sincerely appreciate your guidance.

Colokate

[lordsmurf]

I'm late to the party. Lots of replies.

Some errors can ONLY be fixed in hardware.
Some errors can ONLY be fixed in software.
Some errors compound (not-great either way) and you must choose the lesser of evils.

The trick is knowing which is which.
And knowing where some errors can be adjusted either way.
This is a workflow issue.

You have no TBC. That's already a problem. You're not correcting signal errors AND introducing new ones. Hi8 loves to drop frames and have other issues. Yes, even when the camera has TBC. Not the same sort of TBC.

Corrections really depends on what's being corrected. Color can be best in hardware (proc amp), NLE/edtiro (Premiere, even VirtualDub or Avisynth), or both.

I'm not anti-MPEG, and never have been. The quality of the MPEG, however, matter. This is why I use certain ATI cards, specific DVD recorders, etc. You can still adjust many things in MPEG, just not all, and some to a lesser degree. So, again, it depends. The trick with MPEG capture is 'the higher the bitrate the better'. So avoid sub-10mbps DVD-Video specs, and go for BD or broadcast 15mbps+.

AVI is a container, yes. AVI can hold Huffyuv, Laragarith, DivX, Xvid, etc. Even MPEG, though it's dumb!

Sometimes an MPEG hardware card will send max compression to VirtualDub. I don't know the specs of the card, but MPEG itself has 4:2:2, no GOP (I only), up to 50mbps+, which approaches transparency with lossless.

The main difference in MPG and AVI, in this exact situation, is that editing programs will like the AVI better. True, the AVI may have lossless versions of an MPEG hardware compressed, but not being MPEG alone removes some decompress lag.

Hardware MPEG compression is generally better than software MPEG compression. ATI had hybrid, Hauppauge had hardare, Matrox had hardware. < This has nothing to do with lossless.

Right now, those sub-$100 ES10 units are undervalued. Last year, they were double that, if you could find them. Buy one while the buying is good.

@sanlyn: Nice replies.

[colokate]

Lordsmurf,

Thank you very much for your reply. "Grant me the wisdom to know what issues I can address in software and which I need hardware to accomplish." I know this knowledge can only be attained through guidance from experienced video folks and a lot of hands on time.

From reading many of the informative posts on digitalfaq and videohelp, I understand for post-processing of Hi8 and VHS that I need to capture in lossless AVI container using huffyuv. I knew I did not want to capture MPEG because I want the opportunity to restore some of the video. Thankfully, sanlyn informed me that AVI is only a container, and with the ASUS capture card, I was most likely capturing hardware-encoded MPEG in the AVI container. Plus, I was confusing lossless with uncompressed. Again, I thought I was getting both lossless and uncompressed using VirtualDub to capture hufyuv from the ASUS capture card. Definitely not where I wanted to be, and I am grateful I understood this before spending hours capturing video in a less than optimum format for what I am trying to accomplish.

From sanlyn's posts, I recently picked up the ATI HD600 USB dongle and a Panasonic DMR-ES15. I wish now I kept that tough-as-nails XP box I had for years. I am now on the hunt for a good S-VHS unit based on your suggestions from your VHS buying guide. The DRM-ES15 was relatively easy to come by. A JVC HR-7xxxx or JVC HR-9xxxx (within my price range) has been harder. But, I will continue to look because I realize how much that hardware can contribute to a successful capture outcome. On my wish list is a AV Toolbox AVT-8710 TBC. In the end, I need to convert VHS, Digital8, Hi8, and DV Video (Yes, family members had to try every new format when they became available). Would a AVT-8710 be useful in capturing the other formats?

I appreciate your reply very much.

With thanks,
colokate

[sanlyn]

New copies of the AVt-8710 are known to have quality control issues. Finding an original green-and-black issue will be more difficult. In the meantime the Es-15 has frame-level sync ability as well as line-level correction, so you can let the Es-15 due the work of the AVT. The only caveat is that the frame-sync circuit in pass-thru units will not defeat macrovision the way a full-fledged frame unit like the AVT. But it will suffice for capture.

Digital8 and DV video are not captured. To "capture" means to make a new and different digital or other recording of an original source. That's not what you want for Digital source. Digital8/DV are copied 1:1 using a Firewire connection and specialized DV-transfer software like WinDV (for which you need XP) or Scenalyzer. Drivers for the latter have been posted elsewhere at digitalfaq. XP drivers for WinDV are on WinDV's old website, just in case you find an old XP box somewhere.

Owning an ES-15 means that you have built-in line and frame sync tbc thru the same pass-thru unit, so you aren't married to a single VCR that may or may not have its own line tbc built-in. For VHS, if you have 6-hour or 8-hour tapes be aware that JVC never catered to slower recording speeds and are best used only for SP tapes. For slower VHS tapes you'll get much better performance from 6-hour tapes with Panasonic or Mitsubishi.

[colokate]

It is been awhile since I heard the term "Macrovision", and that helps me to understand that I do not need the capability to overcome it since the tapes I need to capture are all personal. That is great news, because I can target those funds toward acquiring a JVC S-VHS unit. It is interesting to know that the Panasonic units are better at capturing slower recording speeds. I now understand why folks serious about video capture/restoration have an arsenal of hardware and software tools at their disposal. One size does not fit all.

So much to learn (just about capturing!) and I am grateful to experts like you that help me learn.

Colokate

[sanlyn]

Yes, iI does take some looking about here and there to pick up the seemingly endless details when you first get involved with these projects. But things do settle down more quickly than you'd think. As for slow-speed tapes, I started with more than 300 hours of those pesky creatures. There was a time when I was the last of the big spenders on quality VCRs in the 1990's. Once I got started at recording, I'd set the cable channels for 2 or 3 movies on a day or two each week while at work, almost all of it on 6-hour and 8-hour tapes. That was a mistake -- despite the quality of the recorders, it was awful quality by today's standards and much of that had to be discarded when DVD came along. When I began transferring those tapes a few years ago, the consequences of recording in that fashion became painfully apparent.

In working with them I went through a dozen used VCRs, settling on a handful of JVC and Panasonics for the SP tapes. But the slow speeds, well....Panasonics saved the day with those slow tapes that remained, most of them classic events (the 50th anniversary ceremonies of Normandy, etc.) and movies that will never see print in DVD. Finding players during that time wasn't easy, as VCRs were passing into history. I began by having my 1990 SONY SLV-585HF rebuilt -- a classic, and sad to say that SONY has never been so capable ever since and went downhill fast. That was an expensive project, rebuilt by a pro suggested by lordsmurf. I still use that unit for originals that I recorded on it, but my mainstay became a couple of Panasonics. As it was, once my JVCs died, Panasonic was all that was left.

There are some exotic players still out there, most of them rebuilds that cost almost as much as they did when new. My used/refurb Panny AG-1980 was a major purchase. Beautiful player, but a maintenance nightmare. It has been rebuilt twice since then. But I lived in mortal fear of what would happen if a capacitor or color board went amiss, and the rebuilds were a real drag. With some searching and waiting, I acquired two copies of refurbed 1996 Panasonics, a PV-S4670 SVHS and a PV-4664, No tbc (so that led me to pass-thru units like my ES10 and ES15), but sturdy builds and anamorphous metal heads that just won't die, similar to the video heads in the "AG" high end series. Anything Panasonic built after 1996 is junk -- over-sharpening, pumped up contrast, chroma bleed you won't believe, and really cheap transports and electronics. By 2000 Panasonic was trying to make hay on the reputation of its old "46xx" series by recycling the model numbers with useless new junk. The giveway was the model number itself: the newer PV-V4664 with the extra _V" in the number wasn't anything as able as the old PV-4664, and the newer SVHS machines came out as PV-V46xxS instead of PV-S46xx. The "PV-V" models were preceded by the 7800, 8600, and 9600 series, all of which were horribly noisy affairs with tin-can transports that soon gave out. Meanwhile the 1996 models still appear on auction sites, with the good ones that have been cleaned up selling at not-cheap prices.

So there is hope for players, but the supply is dwindling. You are correct: many of us have learned that depending on a single super-machine can lead to hassles you wouldn't believe.

The tbcs and denoising in high-end machines are great to have. But wait until you get a tape that refuses to track properly in one of them. Yes, it happens. And the denoisers can be overly aggressive with really untidy tapes. So I've had my rounds with mainstream quality and with prosumer silver bullets alike, and I can testify that sometimes the lesser animal outplays the super model. So go for the high end JVC, they are worth the effort. But keep the experience of other members in mind: sometimes that lowly spare machine can save the day.

[lordsmurf]

Quote:

Originally Posted by sanlyn

In the meantime the Es-15 has frame-level sync ability as well as line-level correction, so you can let the Es-15 due the work of the AVT. The only caveat is that the frame-sync circuit in pass-thru units will not defeat macrovision the way a full-fledged frame unit like the AVT. But it will suffice for capture.

The ES10-25 is an odd duck. The TBC is strong, but it does not blank out and rewrite the portion of the signal containing MV. So, for that reason, it's a not a true TBC, as I've said for more than a decade now. But, at the same time, it is an incredibly strong filter, but the side effect is NR/posteratization artifacts.

You cannot win sometimes.

The main problem? The definition of "TBC" is too loose.
Sort of like "cloud" when referring to modern internet storage and hosting.

I still want to find the schmuck that first referred to DV transfer as "capturing" and hit him upside the head with a shovel. That guy has caused confusion for 20 years now! It's as bad as "VHS ripping" by low-knowledge video hobbyists (because you cannot rip non-digital content). And ironically, ripping DV is what you're actually doing when you take unformed digital data and put it into a new file format.

[colokate]

Lordsmurf,

Ah, the curse of the overloaded term. There are a plethora of them in the IT world, too. The important thing to me is to understand the advantages and limitations of the items that are named the same. It is very beneficial to understand the ES-15 does not provide the same functionality as AV, yet both are advertised as TBC. That is very misleading for us amateurs, but is the reality of the situation. I appreciate knowing this because now I can delve into additional research understanding the items are different. Thank you both, again, for steering me in a better direction.

[lordsmurf]

Quote:

Originally Posted by colokate

That is very misleading for us amateurs

And an aggravation to serious hobbyists and pros, too! It's infuriating to buy something called "TBC", yet it doesn't do what you wanted. I've railed against this problem for at least 10 years now. I've pissed off a few companies in the process (good!). I used to go in circles with Davideck on VH, because I thought he was too accepting of the loose term. I wanted to pin down something more concrete. I've somewhat done this, but not yet written up my long piece on it.

You have companies like Canopus and Panasonic that call it TBC because ... um ... it simply makes a good sticker on the box?

Then you have companies like Leitch and Kramer that make TBCs that do everything, though with broadcasting quirks.

And, for the masses, the tape>digital hobbyists especially, all the DataVideo and Cypress/AVT gear -- plus all the VCRs with line TBCs.

When it comes to TBCs, you have to talk about exact models, and hope the person you're talking to has in-depth knowledge on it. Which is why you're asking here, because we do.

[colokate]

There is so much "noise" on the web about video capture/transfer it is hard to determine good advice from "fluff".

For example, I need to clean the head on an old VCR for a family member. So, like most, I head over to the web to find resources that address materials and techniques for cleaning VHS heads. Almost *every" resource I looked at suggested using isopropyl alcohol and a Qtip. Wait! Don't Qtips have the tendency to leave fibers when used? (I learned this in different project.) If that is the case, why would I want to risk leaving material behind in an attempt to clean a tape head? Seems rather counter-productive. Thankfully (and this is how I found digitalfaq) in a previous post, you suggested NOT to use Qtips, but to invest in a chamois-based cleaning product, and provided the name of the product you use. So, yes, I definitely rely on this site for in-depth knowledge on tools and techniques.

BTW: My beginner's tool set is now a Panasonic DMR-ES15, Diamond ATI HD600 USB, and a JVC HR-S9800 S-VHS player. I am hoping to pick up a Panasonic S-VHS in the near future.

Colokate