Need advice on what vcr to use for capturing older video tapes

[Deter]

sanlyn,

Were do you get your info from, half d1 is acceptable for VHS.


http://www.digitalfaq.com/forum/video-capture/1507-sp-mode-best.html




**** Been dealing with this stuff for a while and sometimes have to deal with people's, VHS to DVD copies. You get everything. Personally would rather deal with half d1 recordings than 2 hour SP recorded videos, and anything over that (like 4 hour mode in full D1) for sure......

[sanlyn]

No, deter. Never. Not for me. I get my information from watching VHS converted to half D1. VHS starts with low vertical resolution. Half D1 cuts horizontal resolution as well.

[sanlyn]

I had some time to work with the 2nd video, but couldn't get to the first that appears to be on EP. But I'd do something similar to the work posted below on both tapes. The Avisynth script shown is somewhat hurried and arbitrary, could be improved. The script doesn't reflect RGB work in VirtualDub, which was fairly simple with gradation curves and ColorMill. Couldn't fix all the bluish chroma noise in some shadows (well, it's tape, after all). But it looks as if similar corrections would work on almost all of video #2. Some tapes are nightmares, different work needed on every camera shot. This one still needs a bit of tweaking.

2012032300633703_PC.mpg is IVTC'd, pure progressive for PC playback (5.6-MB). Not for standard DVD/TV display. Plays smoother than the original.
http://dc531.4shared.com/download/A4m9mZ7F/2012032300633703_PC.mpg

2012032300633703_TC.mpg is standard MPEg2/DVD 3:2 pulldown for NTSC DVD 29.97fps (5-MB). On a PC display you'll see blending in 2 of 5 frames. This won't be seen on TV. Still looks smoother than the original, even when telecined.
http://dc385.4shared.com/download/c2mCvJdI/2012032300633703_TV.mpg

The originals are pretty OK captures, videeo, not difficult to spruce up as shown in the mpg's above. Now, what happens when we get to the more difficult home-made material?

[Mini-Me]

sanlyn,

Were do you get your info from, half d1 is acceptable for VHS.


http://www.digitalfaq.com/forum/video-capture/1507-sp-mode-best.html




**** Been dealing with this stuff for a while and sometimes have to deal with people's, VHS to DVD copies. You get everything. Personally would rather deal with half d1 recordings than 2 hour SP recorded videos, and anything over that (like 4 hour mode in full D1) for sure......

I've seen a couple picture comparisons on this forum that show significant improvement using full 720x480 or 704x480 capture resolution instead of 360x480 or 352x480, but I just spent fifteen minutes searching, and I can't find the one I'm thinking about most. I wanted to post a link to visually back up this post, but all I remember is that it showed a grayscale "runway" kind of image from VHS and compared full-D1 and half-D1 captures. Each capture was also resized to the size of the other capture, creating a four-way image comparison. If anyone can remember that thread, please post a link.

UPDATE: HERE IT IS!

I also noticed the same from my own tests a couple years back. The difference isn't glaring on casual viewing, since all of the low-frequency content is there, but it's apparent from a direct comparison...and considering we're upscaling video to view on 65" LCD's nowadays, every little bit helps. Half-D1 may be a fine compromise if you're trying to keep the size down on a capture from a second-generation (or more) SLP tape recorded by a mediocre deck though. If you're making a distribution/viewing copy on DVD, decreased resolution may also be worth potentially fewer compression artifacts.

[videeo]

Sanlyn, thanks so much for processing the clips. They are looking good, I am wondering if you look at the rocks at the end is that the plastic effect that people talk about from sharpening? It reminds me of some old Star Trek footage when they are on rocky planets.

I should have mentioned the VCR is reporting playback in SP mode. I made a longer clip on another laptop that I was able to get VirtualDub to work with. When making a couple more clips it was reporting inserted frames. Is that a capture card/computer/software issue or an issue with the VCR/tape? I made a longer clip here to show some vertical jitter (?) and the inserted frame. Looks like the wide and close shots would need different settings for cleanup since they look different. Here is a longer clip, the download is ~250MB from my DropBox.

http://dl.dropbox.com/u/14605816/test4.avi

I have a combo VCR/DVD unit, pretty old only has S-Video/composite for the DVD part, I could try that too as an experiment.

[jagabo]

Full and half D1 from an old consumer VHS deck, recorded with a Hauppauge PVR-250. And from a DVD recorded with the PVR-250 (so you can see the limitations aren't in the PVR-250).

[slats7]

In terms of a TBC to correct jiggling, is it advisable to go with an external TBC device or a high-end VCR like this one?

http://washingtondc.craigslist.org/nva/ele/2911993751.html

[videeo]

Here is the same clip from my first capture post from an old RCA DVD/VHS cobo deck. I'm not sure if I can tell what the differences are yet.

test5.avi

[Mini-Me]

In terms of a TBC to correct jiggling, is it advisable to go with an external TBC device or a high-end VCR like this one?

http://washingtondc.craigslist.org/nva/ele/2911993751.html

It depends on whether you'd prefer living with the quirks of those particular VCR's or the quirks of the external equipment. Here's a summary of your options:

First, I should note that many full-pro decks have built-in TBC's, but I'm not familiar enough with them to do any talking.

JVC S-VHS, JVC DVHS, and Mitsubishi DVHS decks have line TBC's that are excellent at correcting wiggle. The AG1980's full-field TBC is even better than the one in the JVC S-VHS decks and likely better than the one in the DVHS decks as well. However, there's a catch (there's always a catch): The AG1980 has noise reduction that you can't turn off, and the other decks have noise reduction that you can't turn off if you're using the TBC. The DNR causes motion smearing and other weird effects that remove and distort detail along with noise. On the JVC decks, the DNR conspicuously leaves more noise in moving areas than still areas, and on the AG1980, the DNR sucks the color out of moving areas. The AG1980 introduces sharp lines of chroma noise in certain areas of the frame, and it's notorious for being sold with bad capacitors that need to be replaced (mine had this problem as well). The JVC S-VHS decks are notorious for having a soft picture and poor tracking...and sometimes eating or silently erasing tapes when they break.

Dedicated full-frame external TBC's like the AVT-8710 and TBC-1000 are good at ensuring continuous sync and eliminating Macrovision, but they're pretty poor at correcting wiggle. Examples:
https://forum.videohelp.com/threads/195954-Problem-With-DataVideo-TBC-1000
https://forum.videohelp.com/threads/193619-TBC-buying-guide?p=1467477&viewfull=1#post1467477

There are a number of DVD recorders that do a better job at geometric correction than full-frame TBC's, and the Panasonic ES-10/ES-15 and Toshiba DR and RD-XS series are popular for this. There are probably other hidden gems that nobody knows about too. I've found that my Toshiba RD-XS34 is only slightly better for my tapes than my AVT-8710 though, whereas my AG1980's built-in TBC is far superior. My Philips DVDR3475 passthrough may be even better - it's the best I've seen yet - but it has a horrible AGC problem that reacts to high input levels and causes flickering, unstable levels, and clipping, so you can't get good captures with it unless you invest in a proc amp too (to reduce the gain before the signal reaches the DVDR3475). The DVDR3575/3576 have the same behavior.

Finally, certain DV cameras have very good TBC's built in, but some have their own quirky issues, and you'd have to submit to dubbing in DV (lossy compression) before making DVD copies (or any other reencodes).

My suggestion: If you only have a couple tapes, first capture without a TBC to see if you even need one. If so, you can buy a Philips DVDR3475 and a Vidicraft proc amp for less than $100 total, if you find reasonable deals. I bought my DVDR3475 for $51 shipped I think, and I just saw one with a broken DVD drive that was delisted after going unsold (I think) at $30 shipped. (Actually, it was just relisted...but it doesn't have the remote, and you need the remote to switch inputs.) There was even one listed on Craigslist for $10 a little while back. I also recently saw a Vidicraft proc amp sell on ebay for around $40 or so shipped.

Going with the above recommendation will net you a high-quality TBC solution that will work with ANY VCR, so you can pick your VCR based on merits other than TBC performance. Just feed a composite signal from your VCR into the proc amp, reduce the gain, feed the output to the DVDR3475, and output S-Video to your capture card (the DVDR3475 has a pretty decent comb filter). Bonus: The DVDR3475 seems to output continuous sync as well, so you shouldn't lose A/V sync, and your capture card will never drop frames from being unable to read garbage signals (unless it's Macrovision, which the DVDR3475 may sometimes reintroduce on purpose). Drawback: The DVDR3475 has limited memory, so it has no choice but to drop/dupe a frame every now and then if the input timing differs too much from the clock it's synchronizing the signal to. All external TBC's share this problem, but the severity depends on the memory capacity and particulars.

If you have a VCR with S-Video and you insist on maintaining a 100% S-Video workflow, you will need a more expensive proc amp like the Studio 1 Productions/Signvideo PA-100 to feed the DVDR3475...but those are very hard to find used, and the retail price is very high ($450).

[robjv1]

Excellent summary there. I wasn't aware that the Phillips DVDR3475 corrected as well as the Panasonic ES-10 though in terms of geometric distortions. I use the ES-10 quite a bit for that and own the Phillips deck as well but have never tested it for that, I had no idea! I'll have to play around with it today -- the color posterization (especially on dark footage in shadows) on the ES-10 is pretty bad.

[videeo]

Sanlyn, I'm trying to replicate your workflow to see how you created that video. I installed all the plugins and got the script to work. I have a couple questions since I am new to VirtualDub and AviSynth. Did you apply the colormill settings in VirtualDub what were they? How did you deinterlace? And finally, how did you convert to mpeg?

Another general question, is after I try save an avi with huffy compression, the file size increases dramatically.

Thanks!

[sanlyn]

videeo, I had to rush thru the rework of the clip I posted earlier today, the mrs and I had to attend to in-laws. Been gone all day, but I'll try some brief answers now. Will be back tomorrow:

Sanlyn, thanks so much for processing the clips. They are looking good. . .

Thank you. Frankly I'm not satisfied, as I say I really rushed thru it and had to leave. Not much time for tweaking, IMO it's over filtered and too red. I'll tweak some tomorrow. Note the space ship shot at the end, it's brighter than the earlier shot and too green. This happens with tape. You get many consecutive scenes that act well the same filters, then something completely different pops up.

I am wondering if you look at the rocks at the end is that the plastic effect that people talk about from sharpening? It reminds me of some old Star Trek footage when they are on rocky planets.

No. Detail in that shot is pretty soft to begin with, but the plastic look is from over-smoothing.

I should have mentioned the VCR is reporting playback in SP mode. I made a longer clip on another laptop that I was able to get VirtualDub to work with. When making a couple more clips it was reporting inserted frames. Is that a capture card/computer/software issue or an issue with the VCR/tape? I made a longer clip here to show some vertical jitter (?) and the inserted frame. Looks like the wide and close shots would need different settings for cleanup since they look different. Here is a longer clip, the download is ~250MB from my DropBox.

I downloaded, but will have a look later. I've never seen that inserted frame message, but can check later. You might try Googling that phrase and include the word VirtualDub in the search string.

Here is the same clip from my first capture post from an old RCA DVD/VHS cobo deck. I'm not sure if I can tell what the differences are yet.


[Attachment 11575 - Click to enlarge]

The RCA might be old, but it's the better player - at least, with SP tape. Much better. The difference is immediately apparent. It has slightly more grain and a bit more of the fine-mesh dot crawl effect (it can be cleaned), definitely better contrast and more accurate color, cleaner and better shadow detail, less macro blocking, less chroma noise, more detail, better saturation, and better control over highlights, some of which look burned-up on the earlier AVI.

Sanlyn, I'm trying to replicate your workflow to see how you created that video. I installed all the plugins and got the script to work. I have a couple questions since I am new to VirtualDub and AviSynth. Did you apply the colormill settings in VirtualDub what were they? How did you deinterlace? And finally, how did you convert to mpeg?

ColorMill and the gradation curve are RGB filters. They can be loaded in Avisynth (if you convert to RGB first), but listing their parameters in script is a bigger headache than its worth. Anyway, I'll be changing some settings in that script tomorrow. I didn't deinterlace: the videos are progressive and telecined; I used TFM() and TDecimate(), which come in a package with the TIVTC plugin. When encoding to MPEG I applied 3:2 pulldown instead of hard telecine (the latter is the fuzzy frame blending you see on PC playback) and encoded to standard MPEG2 with TMPGenc Plus 2.5 - - two-pass encoding using 6000/8000max VBR. You can do the same pulldown and encode with HCenc. That bitrate might have been slight overkill, as I observed that no part of the video played at a bitrate higher than 5.7 and stayed around 4.8 most of the time.

Another general question, is after I try save an avi with huffy compression, the file size increases dramatically.

Thanks!

If you open a file in VirtualDub and click "save AS Avi..." in "full processing mode", the file has been converted to RGB, which would result in a bigger file. If you forgot to specify huffyuv for compression (the default is uncompressed), it would be 300% bigger or more. If you open an AVisynth script in VirtualDub and want to re-save it in its original colorspace, look in VirtualDub's "File" menu and make these settings: color depth=YUY2 (or YV12, or whatever), compression=huffyuv, then farther down in the File menu select "Fast recompress". Remember, you can't run VurtualDub filtering with fast recompress.

The clips originate as YUY2, which is a good idea for capture. The Avisynth plugins (most of them) require YV12. Users make that conversion all the time, but once you go into any one colorspace, stay there until you're done with it. Don't go back and forth. If you want to re-save the file after running Avisynmth, use "fast recompress" for the color depth and compressor you want. Hint: I don't think huffyuv will compress YV12. Use lagarith; it works with RGB, YUY2, YV12. Lossless, almost as fast as huffyuv, but smaller files -- but avoid lagarith for capture; it's a tad too slow for real-time capture, huffyuv is much faster for that.

Sorry, the wife and I were busy all day and just got home. Will post more documentation soon.

[sanlyn]

Sanlyn, I'm trying to replicate your workflow to see how you created that video. I installed all the plugins and got the script to work. I have a couple questions since I am new to VirtualDub and AviSynth. Did you apply the colormill settings in VirtualDub what were they? How did you deinterlace? And finally, how did you convert to mpeg?

I'm fairly new to Avisynth, too. I've run some simple utilities now and then, but I worked mostly in RGB/VDub. That is, until I ran into an arduous restoration that required Avisynth. I'll agree it looks intimidating at first, and documentation in Russian/bad links/obsolete plugins, etc., can be infuriating. But the thing is free, and can work wonders if you stick with it. There's also help in this forum, and endless discussion threads in doom9.

I've attached a new script 2012032300633703_C.avs. I tweaked the earlier mpg's for PC and TV:
2012032300633703_PC_C.mpg
2012032300633703_TV_C.mpg

MCTemporalDenoise is a slow runner, as you know by now (Try the heavier settings and see what happens. Go shopping while you wait). By now you know how to run an avs in VirtualDub. You can load the VDub filters and run the RGB plugins in the same processing run. It will be slow navigating manually. The other way to do it is to just run the avs without the VDub filters and save the clip as YV12 with Fast Recompress as described earlier. Then reload the saved avs results later in VirtualDub and work with RGB. That's the way many people do it. Trying to fiddle with slow Avisynth scripts and RGB filters at the same time gets tiresome. Easier to break it into two steps, even if it does require two files. The only file you need to keep eventually will be the final. But save your old scripts -- just in case.

If you don't know this yet: For Avisynth's documentation, go to your main Avisynth folder. Navigate to the "docs" subfolder, then down into "English". In the English folder you'll find "index.htm". Right-click on index.htm and choose "create shortcut". Copy or move that shortcut to your desktop. Double-click the shortcut and Voila! Avisynth's documentation appears.

I have attached a VirtualDub ".vcf" file below. This file saves the VirtualDub filters and some of the other settings for this project. To use the .vcf: Load an avs script or the clip itself into VirtualDub, then use "File. . .Load processing settings. . ." and navigate to the location where you saved the .vcf .

Sometimes when loading a borrowed vcf you get a friendly error message about something not being found, etc. Never fear. VirtualDub won't explode, it will simply load what it can. Go into the filters and look around: you might have to adjust something. For example, you might borrow a vcf that uses the NeatVideo plugin that requires NV's .dnp and .nfp settings files. VDub doesn't always know where these are kept, but it will load NeatVideo anyway without the settings. Go into the filters section, open NeatVideo, and tell it where you're hiding those two files. Works OK. There are older plugins that weren't programmed for this vcf import feature. The filters will load, but you might have to tweak some values.

For this project, if you have ColorMill and gradation curves in your VDub plugins folder, they should load OK. NOTE: one of the settings saved is the Lagarith compressor settings. If you don't have lagarith, here's a link to their easy installer:
http://lags.leetcode.net/codec.html

After loading the filters, you can go to the filters and see their settings. There are two gradation curves; each curve is a tweak on the other. The vcf loads them in the proper order. If you don't know how to use gradation curves: they are standard tools in high-end pro-level apps like Premiere Pro, After Effects Pro, Vegas Pro, and others. Their curves have more features, but you'd never use most of them.

Open one of the curve filters. This is what you'll see:


[Attachment 11590 - Click to enlarge]

It isn't isn't difficult to learn to use these. There are hundreds of free websites with free tutorials. Most of those sites involve Photoshop and still images, but with video the principles of correction are the same. Basically, the "curve" tells you what the filter does. When the curve bends to the right, it darkens a color. When it bends to the left, it lightens. The top part of the grid is bright colors, the lower part is dark co0lors.

In the top of the window you see two selection panels. The upper-left selects the colorspace. Here, we're using RGB. The filter can also create a YUV colorspace, and others. In the upper right corner you'll see that the "Channel" selected is RGB: this adjusts all 3 colors, which is a similar effect to adjusting "luma" in YUV. Click the down area in that selection and you can choose the 3 other curves in this filter, one each for Red, Green, and Blue. You can adjust all four of these curves separately, adjusting each without affecting the others (something you can't do in YUV).

The Gradation Curve plugin works pretty much like the big boys in $1000 apps. It's popular and was introduced in this forum here: https://forum.videohelp.com/threads/287113-Gradation-Curves-VirtualDub-filter-v1-45. The home page is http://members.chello.at/nagiller/vdub/index.html. There's a free tutorial: http://members.chello.at/nagiller/vdub/tutorial/tutorial.html. It comes with an html help file.

I've also included another invaluable (especially since it's free) tool called a pixel sampler. Everyone wonders why VirtualDub doesn't have such a control. It sits outside VirtualDub and lets you sample the RGB value of any pixel on your screen. 3x3 pixel sampling, just like the big apps. The .zip is attached. Unzip, find the small ".exe" file and copy it to your desktop. No installer, just a simple standalone. Don't leave home without it. Its small main web page is here: http://www.netreach.net/~gavin/gavsfreeware/csamp.htm

Encoding is a subject all its own. Later today I'll set up something basic and post here. I looked at your big AVI download and found the problem; will post later today. Will also post some samples to show why I think your old RCA is a better player.

[videeo]

Thanks sanlyn, I am digesting the great info you have posted and look forward to your next post. I didn't see the attached .vcf file...

For the latest mpegs, what were the tweaks you made?

[sanlyn]

Thanks sanlyn, I am digesting the great info you have posted and look forward to your next post. I didn't see the attached .vcf file...

For the latest mpegs, what were the tweaks you made?

Phooey, that's because I think I hit "close window" on that post before the file finished uploading!.
I edited, and made sure it was there. Or here's the same link, it's a small 2012032300633703_c.zip, which contains 2012032300633703_c.vcf.

[sanlyn]

Sorry it's taking so long, videeo. Weekends, family around, etc., etc. Will have more a little later. Every weekend is like this. These people just have the wrong priorities ! ! !

[videeo]

Haha, yeah, no problem, I'm just cruiz'n trying to get familiar with this stuff. I tried a avi to mpeg conversion with your AviSynth and VirtualDub settings. The car scene kind of looks odd to me here. I think there might be too much smearing going on that kind of makes it look like a painting. It's more apparent when I play it on my PS3, which is very convenient to test with since I can play off a USB drive.

test9d-try2.mpg

[sanlyn]

Will be posting something in a bit about encoding (working away on it now), but neither Avisynth nor VirtualDub encode to MPEG. Well, not easily...there are ways to hook into ffdshow for encoding, but I'm not crazy about ffdshow's encoders. Besides, they seem to go through an annual "now-it's-obsolete-and-incompatible" phase too often for me. Besides, there's still a learning curve.

BTW, I must mention here: your test4.avi (the big AVI) has been cropped to 740x470. This is a non-standard frame size that most MPEG\DVD encoders will not accept for DVD. More on that later.

The car scene, or you mean the flying saucer? That's a studio composite anyway, it looks like quick-shot work. The other shot looks better. I'd treat that shot separately, then patch it in later. Just one of many problems you encounter with video, especially with old-fashioned special effects.

[videeo]

BTW, I must mention here: your test4.avi (the big AVI) has been cropped to 740x470. This is a non-standard frame size that most MPEG\DVD encoders will not accept for DVD. More on that later.

Yes, after I did that to remove the head noise, I read that it's not the best to leave it that way. I see in your script you crop and then add the borders. I also find it interesting that if I convert a "raw" avi to mpeg and play it on the PS3, it doesn't show the head noise. Speaking of borders, some of the files have while pixels that show up on the very right and appear sporadically.

Another interesting thing is you found the original source is from film. I was working under the assumption it was shot in video.

The car scene, or you mean the flying saucer?

The car scene. The background looks painting-like. That shot might be hard to enhance- it might just be a poor transfer or not shot properly.

[sanlyn]

Now that everyone's gone to sleep (i wish that included me, but it will soon): I'll clean up all the typos tomorrow.

OK, time for encoding the AVI to MPEG2. This will be a basic setup guide for the AVI encoded as progressive video with 3:2 pulldown. Because this is the Capture forum, not the encodijng forum, I've made a quickie Rich Text File (.rtf) that can be opened in any word processing app, including MS Word and the free Windows Wordpad. I had to make a fast rewrite of this tutorial, which was given to me a while back, but now updated as of this late evening.

So far, the the AVI has been processed as progressed (non-interlaced) with a 720x480 frame size at 23.97 fps. 3:2 pulldown will concoct an algorithm that tells your player to insert 2 extra frames for every 5, converting this 23.97 fps movie film to NTSC 29.97 fps DVD playback. 3:2 pulldown is a term meaning that for every 5 frames of video, 3 are progressive and 2 will be duplicated/blended in some fashion to bring the playing speed up to standard for NTSC DVD. Your players are "trained" to recognize this convention. It's just one of several playback standards, and is the way most commercial DVD's are encoded today.

There are two pro-quality encoders that I'm familiar with: TMPGEnc Plus 2.5 and HC Encoder (HCenc). Let's not argue about which one is "best". No one ever wins that debate. Each has pros and cons. Both are used by pros and amateurs alike. There are a few basic differences:

TMPGEnc isn't free, though it offers a free trial. Still sells at its cheap 2004 price. Very popular world wide for some years now. Their newer versions are not as high-rated. So it's a big favorite, still sold today, still cheap. It used to sell in a blue box on Amazon for $80 -- it costs less than half that today via online download. It does require that you have a license key, but not for the free trial. It can encode for 4:3, 16:9, and other display ratios. It has extra features, excellent advanced color correction, a great ghost filter and other goodies. Many of its settings are automated, depending on what you feed it. It has built-in help (with lots of pictures for the textually-challenged. But you have to get used to oddities like the word "frame" spelled as "flame" now and then) and a Windows XP interface. It has not been verified for 64-bit , but it works in Win2000, XP, and 32-bit Vista. One inconvenience: its audio is either uncompressed PCM or MPEg1-Layer II. There are other ways of adding AC3 Dolby or other audio formats.

HCenc is the newer boy on the block and is also a favorite. It's up to you. It has a more basic interface, a help file that isn't that much help for people who are new to video (although I can give you a pdf with lots of info). Two caveats: you can't "open" a file into HCenc directly, you have to use an Avisynth script. Second, if your input video is not in the output frame size (for instance, if your AVi is 640x480), HCenc expects you to resize it to 720x480 or other standard encoding frame size -- even though its tv display size might differ, as with Cinemascope. It has no auto resizer, no color controls, no fancy graphical help file. its preview function is somewhat less convenient and it doesn't appear to hook into monitor calibation profiles the way TMPGenc or Photoshop do -- but, then, it's only a preview. It encodes faster than TMPGEnc, but it is more difficult for newbies. But it's free. One other inconvenience: adding audio isn't as convenient as it should be, but there are ways to get Dolby audio into your video stream.

So the encoder's up to you. For a newbie, I'd suggest TMPGenc Plus. I wouldn't recommend TMPGEnc's newer HD/super-featured encoders. They have great features, but the newer encoding engine is only "OK".

TMPGEnc
Here's a link about TMPGenc Plus 2.5: https://www.videohelp.com/tools/TMPGEnc-Plus
A more advanced guide: http://www.afterdawn.com/guides/archive/advanced_tmpgenc_guide.cfm
An older but still OK basic user guide: http://www.digitalfaq.com/guides/video/convert-tmpg.htm
TMPgenc Plus product page: http://tmpgenc.pegasys-inc.com/en/product/tp.html

HCenc
HCenc listing in Videohelp Tools: https://www.videohelp.com/tools/HC
An older, limited HCEnc guide: http://www.afterdawn.com/guides/archive/hc_encoder_settings.cfm
The latest HCenc pdf: HCenc PDF
HCenc web page: http://hank315.nl/

I made Rich Text File basic guide as easy and basic as I could, updating it from the older one I was given. You're only going to encode one method this time (3:2 pulldown for TV), with no audio. So this should be enough to get you addicted -- er, I mean get you started right off. The attached guide covers some basic HCenc. I can make up a quickie for TMPGenc but the afterdawn link above is pretty good.

Sorry for the delays.

[jagabo]

I also find it interesting that if I convert a "raw" avi to mpeg and play it on the PS3, it doesn't show the head noise.

The TV is hiding the edges of the frame for you. It's called "overscan". You never noticed it before because you had no external reference to compare to. In the old days CRTs couldn't keep the frame centered, the right size, linear, etc. All those problems would vary with the temperature and age of the TV. To make these problems less noticeable they would draw the picture about 10 percent larger than the face of the CRT. Ie, about 5 percent of the frame at each of the four sides isn't visible. Since broadcasters and VCR manufacturers know you can't see the far edges of the frame they don't worry about junk out there. Modern HDTVs don't have the same problems regulating the screen but they continue to simulate overscan because there's often junk at the edges of the frame and they don't want buyers complaining about it. Some of the better HDTVs have a non-overscan mode that lets you see the entire frame.

[sanlyn]

Another interesting thing is you found the original source is from film. I was working under the assumption it was shot in video.

Actually some of it looks like film, some like video. That's not unusual, but video can be shot at the same progressive/frame rates as film. Sometimes it's difficult to tell.

The car scene, or you mean the flying saucer?

The car scene. The background looks painting-like. That shot might be hard to enhance- it might just be a poor transfer or not shot properly.[/QUOTE]
I see, you refer to the "test9d" video. Yes, that scene doesn't look like the rest of the video. In editing, it happens.

I made a longer clip on another laptop that I was able to get VirtualDub to work with. When making a couple more clips it was reporting inserted frames. Is that a capture card/computer/software issue or an issue with the VCR/tape? I made a longer clip here to show some vertical jitter (?) and the inserted frame. Looks like the wide and close shots would need different settings for cleanup since they look different. Here is a longer clip, the download is ~250MB from my DropBox.

http://dl.dropbox.com/u/14605816/test4.avi

I have a combo VCR/DVD unit, pretty old only has S-Video/composite for the DVD part, I could try that too as an experiment.

I took a long look at test4.avi. First, laptops are much slower than desktops. They have slower drives and processors. That can cause capture problems, such as dropped frames, bad audio sync, etc. But that jitter looks like a stretch of damaged (worn) tape not streaming evenly through the tape path. It's likely not the player's fault; damaged tape will look damaged on almost any player, some will play it better than others.

There's an old trick that helps stabilize tape motion across the heads. If you look at a tape that's been played/stopped/restarted many times, you'll see that it doesn't rewind smoothly onto the reels. The windings have humps and uneven packing, with sections of tape edges sticking up higher or lower. This causes tape to enter the path with vertical rises and dips; often the tape guides can't stabilize it that well.

This can be fixed (mostly or partially, depending) by repacking the tape. This refers to fast-forwarding the tape all the way to the end -- not playing, just fast forward with no pause. Then rewind it to the beginning without stopping. You might have to repeat this a time or two. When the tape looks more evenly wound, let it sit for a day or two. Mylar's innate flexibility will often "mend" itself. Then re-capture the bad sequence.

You can look at this clip in VirtualDub in three different ways: (1) the original test4.avi, unprocessed, (2) a bobbed version, (3) an ivtc'd version. (telecine removed). Let's handle the unprocessed version first:

Load the unprocessed tes4.avi in VirtualDub. Don't play it. Use the frame-forward button (>) and advance frame-by-frame. Frame count starts at 0. There's not much motion at first, but you can see blended frames at 8, 78, 148, and so on. These often occur at scene changes. Move ahead for some motion shots. You can see telecined frame blending when the girl moves her head in frames 168,169,178,179. Check the guy in the background throwing the frizbee at 238 thru 262. Motion sequences show the telecine effect, which occurs every 2 frames in 5.

You can also see the jitter, but some "hopping" as well, and a broken frame. Look at frames 271, 272, 273; there's a hop on frame 272. Some jitter in 386-402. We'd call frame 523 a broken or bad frame. I don't know if 523 is the frame referred to in the "inserted" message. Instead, I think the insertion came earlier. But frame 523 looks out of sequence, blended from two other frames: 522 or the "original" 523, blended with another frame (I'm using 494 as comparison, but I'm not sure of the exact frame. Or it could be a later frame). Compare the three images below -- I resized them to 4:3 display ratio instead of "stretched" frames:


[Attachment 11602 - Click to enlarge]

Look at these frames another way. Create a "deinterlaced" avi clip with the bob() utility. This will treat the video as having a top and bottom field in each frame (it was likely played and recorded that way, even if it wasn't encoded that way). The Avisynth Bob() utility splits frames into top and bottom fields and resizes the fields, doubling the frame count. You wouldn't usually "save" a clip this way, but we're saving it for demo only. To bob() test4.avi, use the script below. Save the file as "test4Bob.avi". For now, don't worry about colorspace but use huff or lagarith compression. If you try to play while the script is "bobbing", you'll have slow or jerky playback.

AviSource("place path to file here\test4.avi")
AssumeTFF().Bob()

Open test4Bob.avi in VirtualDub and view the clip frame-by-frame. Remember, the frame numbers are now doubled, because you're really looking at fields. The even fields are numbered 0,2,4,6,8..., the odd fields are 1,3,5,7,9.... In the bobbed clip, the first thing you see is frame 0. It doesn't look like the original frame 0. Look at frame 1. If you blended frame 0 and 1, the blend would look like the jumbled first frame in your capture. The "normal" frame is the top field 0, the damaged frame is odd field #1.

Go to bobbed frame 612. This is the top (even) field of the original frame 306. Now, move slowly frame by frame and notice that some images appear twice, some appear three times. I'm labeling the bobbed images as a, b, c, etc. These are the images you see:

a a b b b c c d d d e e f f g g h h h i i j j j k k l l l

Beginning with "f f", the 2-3-2-3-2-3 sequence is broken. It should be:

2-3-2-3-2-3-2-3-2-3-2-3

Instead, it's:

2-3-2-3-2-2-2-3-2-3-2-3

When you inverse-telecine a clip, you remove duplicated fields and the video becomes progressive. If you bob() the IVTC'd clip or use SeparateFields(), the unbroken portions play as:

a a b b c c d d e e f f g g h h i i j j k k l l

or: 2-2-2-2-2-2-2-2-2-2-2-2

When a clip is encoded as soft-telecined (which means "software telecined"), it doesn't encode the 2-3-2-3-2-3 sequence, it encodes 2-2-2-2-2 but has signals that tell the DVD player to play it as 2-3-2-3-2-3. When you capture, it will be captured the latter way because that's what your player outputs and that's what your capture device "sees".

When this clip is inverse-telecined, you'll see where the sequence is broken. the IVTC algorithm will trip up and insert an extra frame at that "break" in the original sequence. The video will play back with an apparent jerk or freeze-frame at that point.

To demonstrate, you should inverse-telecine the original test4.avi using this script, and save it as "test4ivtc.avi":

AviSource("place path to file here\test4.avi")
AssumeTFF().TFM().TDecimate()

Open the saved "test4ivtc.avi" in VirtualDub.The original avi had 989 frames. With 3:2 pulldown removed, it now has 791 frames. You'll see that frame 0 now looks "normal". Go to frame 253. Look at the next few frames one at a time. The images play one image at a time, no duplicates....until you get to frames 257 and 258, which are duplicated. Now, using the frame-forward button in Avisynth, press the button to play the frames more rapidly. There's uneven motion across frames 256 to 258. One frame is missing (it seems to me), and two of them are duplicates.

Go to frame 396. This begins the sequence that contained the broken, blended frame. Play this sequence and you'll see that it plays normally. The broken frame isn't there. When 3:2 pulldown was decimated, that blend was removed.

The broken sequence might be due to bad tape movement. Try repacking and capture again. If that sequence is still a problem, the frames can be repaired in Avisynth before using IVTC. But that's a problem for the Restoration forum, not capture. I'd start a new thread in Restoration and submit any problems you have with that clip.

Even with repacking, it's often necessary to use a full-frame TBC during playback. It doesn't always solve the problem, but it often does. This is not a line timing problem that would be addressed by a line TBC. I don't see any geometric wiggles or bent verticals here, so I still assume your capture device has a line TBC (?). Your DVD/VHS combo would have one, since most such units do.

[sanlyn]

Here is the same clip from my first capture post from an old RCA DVD/VHS cobo deck. I'm not sure if I can tell what the differences are yet.

Test5.avi

This is all personal preference, and it might apply only to this specific video rather than all videos you have. But that's why many people have more than one VCR around.

I'd prefer to work with the RCA capture. In the image below, the left-hand portion is from frame 149 of 20120323000633703.avi, the right-hand side is from frame 121 in test5.avi (made on the old RCA). Both are unprocessed.

left=20120323000633703.avi. Right-test5.avi (RCA)

[Attachment 11606 - Click to enlarge]

Neither has perfect color: the right-hand side looks warm and could use a tad of blue, but the left-hand side has too much blue especially in shadows. The RCA image has better skin tones and smoother gradations in shadows. The left-hand image has rougher looking gradations in shadows and some cyan chroma noise there that I found difficult to clean; the shadows just don't look as normal, and the excess bright blue in highlights makes skin tones look unnatural. The highlight on the bridge of the left-hand girl's nose has a hint of green. The RCA image isn't just a little darker (RGB values in the bright areas are nearly the same in both pics), it has a more natural contrast range, and cleaner contrast at that. Shadows in the left-hand image look anemic and murky in the trees, and skin shadows are unnaturally grayish; there's too much cyan in there. The RCA's better contrast range leads to an impression of better delineation, even though neither image is any "sharper" than the other.

Other frames in both images reveal that the top-end magenta rainbows are the same in both videos. In the scene with the rocks and trees, the RCA image looks more well-defined and highlights on the space ship don't seem so "hot". When it comes to cleanup and correction, I'd rather work with the RCA capture. The left-hand image is more of a pain in the neck.

[videeo]

Nice tutorials sanlyn! Things are starting to make more sense.

I don't think the Diamond ATI 750HD USB dongle includes an in-line TBC. I am guessing the quality you have observed so far shows it might not be needed?

There is some lag in the capture, usually there is at least one dropped frame reported in the beginning of the capture. Sometimes jitter is also reported. I think when I simultaneously capturing sound, there was a large lag where a whole bunch of inserted frames got dumped in the beginning and the sound was way off. It could have been because I did not change the audio capture settings, and by default the sampling rate seemed super high. If it is a problem to capture both, I can resort to using a PC and/or using a separate device for audio capture. I have a nice digital audio recorder and I can capture it at something like 24/44 or 16/24 and then add the audio later.

For video capture in general, does it make more sense to capture in smaller chunks vs just one continuous file? If I am going to do extensive restoration where there are different sections that are each treated separately, then it seems like it would be better to break-up the capture. Is there an easy way to rejoin the segments so that it will not suffer quality degradation?

Test9d was a 4th subsequent capture of the same segment of the tape. I recaptured to see if I could (I think I did) get rid of some jitter, which I think is mechanical in nature.

[jagabo]

The ATI 600/650/750 capture devices do no have a line TBC. Neither do most DVD recorders. They make no attempt to correct horizontal jitter. Yes, line jitter is caused by variations of the drum speed while recording and playing back.

A few dropped frames at the beginning of a cap is usually due to the program opening the file for writing. Windows may not return from that call for a while. If you know how big your cap is going to be (or just have lots of free disk space) you can use VirtualDub's File -> Allocate Disk Space option.

[sanlyn]

That jitter could be damaged tape, which occurs often with old tape. Sometimes it was just manufactured that way, and I've seen that as well. Capturing on a laptop doesn;'t help, as I said; laptops are often just too slow.

How much you capture at a shot is personal preference. I just make a few settings for dark and bright levels in VirtualDub capture (nothing major, it's going to be corrected extensively anyway), and capture the whole video. Then I break it into smaller files, each to fit on a DVD. Now and then I've had to ecapture a sequence that needed stronger correction during capture to prevent problems later (ah, good ole NTSC!), but I have a PA-100 proc amp for those problems.

I'm not familiar enough with your Diamond unit to know all its details. Don't know why it won't work with VirtualDub on one PC but not on another. Maybe other readers can advise on that. My very first captures were the lazy way, direct to MPEG. Brother, what headaches! But capturing audio separately often doesn't work. I never tried it, but I've seen problems posted. Most capture devices and VirtualDub make adjustments to maintain sync. Doing it separately would likely undo those adjustments.

[sanlyn]

The ATI 600/650/750 capture devices do no have a line TBC. Neither do most DVD recorders. They make no attempt to correct horizontal jitter. Yes, line jitter is caused by variations of the drum speed while recording and playing back.

A few dropped frames at the beginning of a cap is usually due to the program opening the file for writing. Windows may not return from that call for a while. If you know how big your cap is going to be (or just have lots of free disk space) you can use VirtualDub's File -> Allocate Disk Space option.

Hm, well thanks for clarifying the tbc issue. I do notice the videos have a smooth flow, no ripples around. But I've heard (i.e, read) that most DVD/VHS combos have a basic line tbc, or something to maintain audio sync (that's what I saw posted here recently). The current Magnavox top of the line (if you can call it that) DVD/VHS combo sold in the U.S. has one, as verified in an earlier post -- which, of course, I can't find now! Anyway, I can only vouch for my own Toshibas and Panasonics.

I posted some caps recently using a Panasonic PV-4662 VCR with no tbc in circuit. Darn if it wasn't smooth as glass, no bends or wiggles anywhere. Now, I know from 5 years ago that the tape has wiggles -- that's the tape that had me using tbc pass-thru in the first place for all VHS captures. How the Panasonic managed it, I don't know. I intended the test to show the difference between a tbc and no-tbc. It didn't. In that respect, aside from Macrovision effects, both samples look alike. Played on a later VCR, geometry problems reappeared (I should have posted those caps, too, but didn't). Beats me. The post compared two tbc pass-thru units and their effect on a played-to-death, beat-up old VHS. For anyone who cares, it's in the forum here.

[sanlyn]

There are Avisynth filters that can help with jitter (sometimes), and they usually work well on frame hops. The bad frames can be fixed, too, but that can get complicated and belongs in the Restoration forum.

[jagabo]

But I've heard (i.e, read) that most DVD/VHS combos have a basic line tbc, or something to maintain audio sync (that's what I saw posted here recently).

They have a full frame TBC. The capture hardware saves frames into a frame buffer and the MPEG encoder reads them from there at a fixed rate. That is separate from a line TBC which restores the length and phase of individual scan lines.

[sanlyn]

Ah, OK. That explains what I read about most or many of those products, and makes sense.