Questions: converting VHS/VHS-C and 8mm/Hi8 tapes
I'm in the process of bringing all of my old family tapes and videos back to life in the digital age, and making an online hard drive for our family with all the videos, dated and listed. I sort of understand how to go from the individual devices to the computer but I'm no expert in doing so. I bought the Elgato Video Capture and tried it out for a day...wasn't satisfied. So now I have a few questions on what to do next and how.
* I have the option tomorrow to purchase a Canopus ADVC-100 tomorrow off Craigslist, the asking price is $95, with FireWire/power cable/RCA cables included. Is this device great for me to get for what I'm doing? And is the asking price for it a bit high in today's world and should I look elsewhere? * So I have 3 devices I'll be using to physically play these: JVC S-VHS Camcorder - GR-SXM260U, Sony Hi8 Handycam CCD‑TRV318, and a LITE-ON LVC-9016G DVD/VCR Combo (but the DVD tray is stuck). What exactly is the best way to get the best quality possible from each of these devices and with the Canopus in play too and how? Here are pictures of all the ports on all of them: https://imgur.com/a/h7Ymt The 1st picture is of the S-VHS-C camera's ports, the 2nd is of the Hi8 camera's ports, the 3rd and 4th are of the VCR player's ports, and the last 2 pictures are of the computers FireWire ports (one in the front and one in the back). With all these options, what would be the BEST method for EACH type of tapes: the VHS-C tapes on the JVC camera, the Hi8 tapes on the Sony camera, and the VHS cassettes on the VCR? Based on all the ports that I showed you and that they have. S-Video over composite if that option is available? Use the A/V out thing on the Sony camera instead of the S-Video port..? Things like that please, this world is all new to me... * Should something else be used for the VHS-C tapes other than this JVC camera? Like an adaptor to plug it into the VCR and let me capture it from there? Or another VHS-C camera with more port options? When I was using the Elgato and recording it, noise would only be present out of one side and one speaker. This is because the JVC camera has only one audio jack...right? Or was that an Elgato problem and it shouldn't happen with the ADVC-100 if I get it? There was literally no noise coming out of one side...so what actually happened there and should I just go the VHS adapter method? Unless that makes it worse quality? * Some of these VHS-C tapes go blue for a second when playing and then goes back to the video, plays for a bit and then goes back to a blue screen for a second, it only does this in some parts. I'm guessing this is just because the tape is old? It fell on the ground once and a bit is damaged? It's dirty? I don't know, is this fixable or only able to be done in post? * I see that the DVD/VCR combo has a "DV Link" or FireWire port in the front? Would it be possible and better to go straight from that into my computer with a cable and capture it like that (if that's possible)? Or would it be better to go through it with the Canopus like everything else? * I also heard about some method called the "pass-through" where you connect one camera to another and go from there, over and over? Can you explain this and would it be better if one (or all) of my captures went through a specific thing, like the Hi8 camera maybe? Or the JVC one? Speaking of, would me passing through the JVC to the Hi8 fix that audio out of one side problem? I don't really understand it so if you could explain that and let me know if any of these tapes would get better quality going through this method I'd appreciate it. * Does the ADVC-100 do the whole interlacing/decombing bit I always hear about or is that done manually after capture in your software (and is it supposed to be done with all of these tapes: VHS, VHS-C, 8mm)? * What software do you recommend for capture or is it all really the same because the source will always be what you're giving it? Or are some really better than others and there is a difference? Would Premiere Pro CC 7 be okay? If you read all that I appreciate it and I await your answers to all my questions, thanks for reading. |
Helloo, and welcome to digitalfaq! :salute:
Quote:
Quote:
Quote:
Quote:
Quote:
Quote:
Quote:
Quote:
Quote:
Quote:
Do not deinterlace interlaced source during capture. Deinterlacing itself never improves anything and is not used for DVD or many BluRay formats. If you don't want to degrade your capture, don't deinterlace. If it has to be done, do it in post with the proper tool, which is usually Avisynth. Premiere Pro's deinterlacing is visibly inferior to several other methods and apps. Many cleanup operations can be performed without the need for deinterlacing. Deinterlacing is for internet streaming (although I hear that UTube now supports interlaced video). There is no need to deinterlace for PC or TV play. Media players, set top players, and TVs are pretty good at deinterlacing on their own. http://www.digitalfaq.com/guides/video.htm Some of the hardware mentioned in the capture and restoration guide has changed over time, but the methods and principles remain the same. The guides are pointers to the "best" results that you mentioned several times. Anything else will not be the "best". Analog capture is a world of its own that requires proper tools. Many blame analog tape itself for poor results. There's some truth to it, because analog can be pretty ugly. But that's no reason for not processing those sources with better tools and methods. |
Adding to sanlyn's comments:
That's a fancy S-VHS-C with digital (line) TBC and chroma noise reduction. If it tracks the tapes well, it should be a good way to play those cassettes. The blue screen is concerning, though. If it was generated by the camcorder's playback itself and not by the capture device, perhaps there is a way to disable it like there is for many VCRs. You'd have to look through the menu system and/or manual. The photo of this S-VHS-C camcorder shows only a white RCA jack for audio. This single wire can only carry one audio channel, and the camcorder only recorded monaural in the first place. In summary: you're not missing anything with your captures that only have one functional audio channel. After capture, you can select to keep the audio channel that actually contains audio and drop the empty garbage channel from the stereo pair. The DVD recorder's DV iLink/Firewire port is only for input, not output. You can't use it for what you were thinking, even if you wanted to (and you shouldn't). |
Thanks for msgihan's notes.
Mono audio, hm? Like my sister's old Panasonic camcorder. After capture you can use the free Audacity audio app to copy one channel into the other to improve overall listening quality. It won't be stereo but it will have a fuller sound to it. Audio for this kind of project is usually captured to uncompressed PCM at 16-bit/48KHz. Uncompressed audio is preferred for post processing to avoid degradation thru recompression during the work. You can encode the audio to a final format when you encode for final output (usually Dolby AC3 for DVD or BluRay). |
sanlyn and msgohan, thanks for your replies, I appreciate it. So what would you recommened I get as a capture device other than keeping the Elgato or getting the ADVC-100? Some people point to cheaper USB ones that do a better job, some people point to TV source capture, some people point to the super expensive ones with hundreds/thousands of dollars machines...
In my situation if I don't want to spend hundreds of dollars on a TBC and device, what should I do and get? |
TV capture, no. The best that we've always recommended is the old ATI All In Wonder AGP line, which in their heyday sold for $200 - $300 MSRP but which can occasionally be found on the used market -- not that easy, though, and motherboards with AGP mounts haven't been made for years. The AIW line was competitive with pro devices selling for much more.
The forum posted a list of more current alternatives to those highly valued AIW's: Hauppauge 610 USB2 capture stick ~$50 ATI TV Wonder HD 600 USB2 capture stick ~$50-100 ATI TV Wonder HD 600 PCI capture card (aka Diamond ATI TV Wonder HD 600 PCI capture card) ~$50-100 ATI TV Wonder HD 650 PCI capture card aka Diamond ATI TV Wonder HD 650 PCI capture card ~$50-100 (not the USB versions of those two!) Those are recommended because we know that they give optimal results. Cheap knockoffs of other products just don't cut the mustard for analog capture. If you're using Windows 8 or 10, you're limited to the Hauppauge and Diamond products. Despite what you think of new operating systems, the mainstay for capture and for much of the restoration work done by advanced hobbyists and even pros these days is still Windows XP, with Windows 7 running a close second. Newer versions of Windows have more limited choices. There are two types of tbc devices: line tbc's and frame tbc's. Line tbc's correct scanline timing errors within frames, a problem common to all analog tape playback devices except some cameras and high end VCR's that have built-in line tbc's. Frame tbc's correct the stream of frame timing errors that result in dropped or inserted frames or poor audio sync. A frame tbc might or might not be necessary depending on the tape and the player, but most members here use a frame sync device of some kind or other. If you don't have a VCR with built-in line tbc and an external frame tbc, you can get pretty much the same effect by using a recommended pass-thru device that can supply both functions. A line tbc is essential. A recent thread from a few days ago has links to samples of disturbances that result from scanline errors and a sample of how a preliminary test with a pass-thru device fixed the problems. I get better results today from more able pass-thru units. The most recommended DVD recorder for pass-thru is the Panasonic DMR-ES10, with the DMR-ES15 in second place. They are all over the auction sites at decent prices, about $100 and often less. Few DVDR's can be used for pass-thru, and many that can will do a crappy job. The post with line error samples: http://www.digitalfaq.com/forum/vide...html#post43209. There's no way to avoid the distortions and noise that result from scanline errors, and no way to fix them in post processing. It would be great if getting decent results from analog tape were the easy 1-2-3/click-an-icon process that almost all consumer blogs preach. But it isn't, and it never was. The "usual tools" give the usual results, which are really not pleasant to watch and more often than not play worse than the the original tapes. As you've learned already, analog to digital capture is a unique animal that goes haywire when not done with the tools and methods designed for it. The noise and distortion of analog sources are mortal enemies of digital work. Those analog glitches, if not processed properly, will be forever embedded as ugly digital boo-boo's that won't go away. |
How is the Hauppauge 610 USB2 capture stick any different than what I'm doing now, just the hardware inside is better than the Elgato? And all those other links have a TV cable port in the back of them instead of a USB port...so for those does that mean I'd have to hook it up to an actual TV too? Slightly confused as to what to do with those other links if they don't have USB ports/composite cable adapter on the other end. I'm on Windows 7 btw.
Quote:
And what do you recommend for my capture method, VirtualDub and the Huffyuv tool? I did this and the results are better with the Elgato now...but I was left with 7.78gb with 6:39 of capturing time...this still seems crazy high if you want to capture an hour. Is this normal or did I do something wrong? I mean I'm all about quality but that seems insane if I'm trying to do an online shared family hard drive with these memories, no one there is going to want 90gb files. I was able to export it to 217mb with Premiere, and dropping down the target bitrate on VBR 1 pass from 6 to 4 saved me about 80mb, so this makes more sense I guess... speaking of, what exactly does that target bitrate do and am I doing it right? If I drop it even lower does it do anything or no since these are all old tapes and don't need high bitrate? |
Hi again! Wow, more questions. Good, that's what folks are here for. Much of what follows is in the capture guides and restoration areas, but here goes:
Quote:
Quote:
Quote:
Quote:
Quote:
Quote:
Quote:
The usual workflow is (a) capture to lossless media using the original colorspace, at either 720x480 or 640x480. (b) Then denoising and defect repair, usually fairly simple most of the time with a good VCR as player. This isn't something you can do in Premiere, which isn't good at it. (c) Lossless editing in an NLE of your choice for cuts, joins, added features like transitions, title overlays, etc. Premiere can do this, although premiere has a lot of trouble with lossless codecs like huffyuv. But you can always use Lagarith. (d) Encode to final delivery format: DVD, SD BluRay, AVCHD, mp4, web posting, whatever. Final delivery formats are not designed for further modification. That's why they're called "final". (e) If you want to, you can author for optical disc. The end of this process is the time to delete your work files and, if you wish, archive the capture to a high bitrate format like MPEG or h264, archival bitrate no lower than 10MBps for h264 or 12-15MBps for MPEG. Lower bitrate is used for final output. Quote:
On the other hand, if you don't cotton to cleanup, denoising, defect repair, color correction, etc., you won't need the majority of information in restoration guides and posts, which are there to help make analog tape look better than the original. The lossless/restoration route isn't for everyone's taste or patience. You can continue with what you use now, although the results won't look better than those you have so far. You could likely get somewhat better results with no post processing by getting a better VCR, line tbc capability, and a good DVD recorder (many of which have some level of built-in tbc during record, and some of which have the excellent LSI encoder chip). The lossless process has been used and recommended for decades by restoration and transfer professionals and by advanced users who want to get the best they can from their taped memories. It's not something invented by digitalfaq. |
Thanks for taking the time to respond again, appreciate it.
Quote:
Quote:
The first file is from a VHS-C tape played by the JVC camcorder captured by Elgato's own software, in the high quality setting, you are told to download on their website. The second file is the same clip in VirtualDub with Huffyuv. The third file is from a Hi8 tape played by the Sony camcorder captured by Elgato's own software. The fourth file is the same clip in VirtualDub with Huffyuv. edit: The forum won't let me upload them for some reason, every time it tries and then takes me to an error page. I uploaded on Dropbox instead, download them all here please: https://www.dropbox.com/sh/fq0kd6rst9mbg05/AAA4gjYLTo8fRINtiBLWf4bSa?dl=0 Now I'm no expert in this, but the same device did way better in vdub than in Elgato's own thing, you can notice it immensly if you look at the text comparison in the first 2 files. But you know more about what to look for than me. So seeing these files and what can be done in vdub would you say this is great and to stick with it..? Quote:
b) What do you mean by denoising and defect repair, how do you do this and how would you do this with a VCR? I thought you meant that's what you do on your PC, you know in post, doing alterations to the videos. c) From what I've done in Premiere it's worked fine with the videos that are in Huffyuv, no problems from what I can see, and it automatically scales the video to 4:3 as it should be when exporting. d) For final output for the final file (not archival lossless stuff, but the .mp4 file for the family online drive) the default when exporting in Premiere is 6, but I drop it down to 4 and don't see any difference other than file size. Is this good or normal? Quote:
2-pass VBR is the way all the time for me? No more 1-pass? What's really the difference here? When I select it the estimated file size doesn't even change... Quote:
But if that's also the case for 8mm/Hi8/VHS-C tapes, the restoration process, can you give me an example and do it with clips 2 and 4 (VirtualDub Huffyuv version) that I uploaded? Unless it takes too long, then that's okay. For what its worth my process right now would be JVC Camcorder and Sony Hi8 Camcorder > S-video and 1 audio jack > Elgato USB capture device > VirtualDub with Huffyuv compressor > Premiere Pro export to .mp4 (with target bitrate of 4 from the default 6, and either 1 pass or 2 pass VBR). As you can see, I haven't gotten to the VHS tapes and VCR part, that's the last thing I'm gonna get to maybe. A lot of tapes are dusty and I have to figure out how to clean them... :/ |
4 Attachment(s)
Quote:
Quote:
Quote:
Quote:
Quote:
Before I can get to your other questions I confess I took the liberty of posting your samples in the forum. Most members won't go to sites that display ads (especially SignUp! ads) to download samples. Besides, offsite samples disappear and readers are left scratching their heads and wondering what the discussion's about. I removed about 30 of the trailing frames from each .avi sample, so in the attachments I added a "c" to the title to distinguish them from your originals. The mp4 samples are loaded exactly as downloaded. The original avi's were just a tad over the 99MB limit, likely due to differences in segmented storage blocks or some other spooky tech mystery. Except for the removed frames, the .avi's were saved in VirtualDub using Direct Stream Copy in the original huffyuv YUY2 so that no re-processing or colorspace conversions would occur. Quote:
1) test 1a.mp4 (4.89 MB) 2) test 1b_c.avi (90.92 MB) 3) test 2a.mp4 (3.37 MB) 4) test 2b_c.avi (91.31 MB) Tomorrow I'll prepare some notes and images to post. I can see why you said you're unsatisfied with the captures. The mp4's were field blended. This is permanent damage that can't be repaired. Because of ghosting and blurred double images the mp4's are of no further use. The avi's would be more workable with a better player and capture card. Still, even though there's no way to retrieve details destroyed by clipping during capture, the avi's might be good enough for improvement -- up to a point -- so I'll post some ideas tomorrow. |
Quote:
Quote:
Quote:
Quote:
And if you can please elaborate on the whole YUY2 thing, if its recommended for capture of all 3 types I'm doing: VHS, VHS-C, and 8mm/Hi8 tapes, and what to do next since VirtualDub says it's not compatible. Thanks again. edit: Did a little bit of research and saw this: http://forum.videohelp.com/threads/2...s-UYVY-Capture https://answers.yahoo.com/question/i...7151804AAuQZKD Basically, my VirtualDub defaults to UYVY, and this says they're identical. So...I should be good right? |
7 Attachment(s)
Been a long day so far.
Quote:
Quote:
Quote:
Quote:
YV12 = v:1:1 or 4:2:0. This means that for every 4 pixels of luma data, there is 1 pixel for U chroma and 1 pixel for V chroma. 4:1:1 and 4:2:0 simply means a different arrangement of the chroma pixels. The Y, U, and V are the same in both. 4:2:0 is used for DVD, BluRay, AVCHD, and some others. YUY2 & YVUY (and some others): 4:2:2. For every 4 luma pixels, there are two U pixels and two V pixels. Different 4:2:2 systems store the same data in different pixel arrangements. 4:2:2 is similar to the way analog tape stores the same information, in which case it's called YPbPr. 4.2.2 NTSC video captured to DV becomes 4:1:1 or 4.2.0, thus losing one each U and V pixel -- i.e, the conversion loses 50% of the original color resolution, and does a pretty sloppy job of that and a few other factors. RGB = 4:4:4. For every data bit of luma there are matching bits of chroma. Luma and chroma values are stored together. If you want compatibility in your situation across the line, you should be using a compatible capture device and drivers. Quote:
Quote:
The advantage of 2-pass encoding is that the first pass does some motion estimation and other work that makes the 2nd pass more efficient. Often it can make for a smaller file. Again, depends on the video. Quote:
I spent a few hours working with these nightmare captures -- something I'd normally avoid, but you keep asking. Some of the problems are coming from your capture gear, others are due to inappropriate levels during capture. The image below, resized to 4:3 to save a little forum space, is an otherwise untouched frame 42 from test 1b.avi. http://www.digitalfaq.com/forum/atta...1&d=1460662191 Color is washed out. That's not the only problem. There's some color corruption going on, a loss of color density and bad skin tones many of which look grayed out. There are no clean or realistic darks in the image, so it lacks a perception of depth. Clipping is easy to see. Look at the elephants and the rocks. Bright detail is clipped, or destroyed, because the luma and color values in your capture exceed RGB 255 at the bright end. Digital video can't display properly with values beyond RGB 0-255, and in factv the broadcast vstand ard is RGB 16-235 -- 16-235 gets expanded when converted to RGB for display. IF values exceed the accepted range, those values are simply clipped (i.e, destroyed). There's severe clipping in the bright rocks and elephant area, and the elephants look like cartoon ghosts. Also look at the man's shirt on the right: the sleeve is so bright white it has a neon effect (called blooming) with no detail. It's easy enough to graph this with YUV histograms and vectorscopes based in this frame: http://www.digitalfaq.com/forum/atta...1&d=1460662255 This simple YUV levels histogram in Avisynth has three horizontal bands. The top white band is luma, the middle band is U chroma, the bottom band is V chroma. On each side of the histogram are dark shaded areas -- values overflowing into those shaded areas represent clipping, or unacceptable values. Yellow arrows point to two major problems in this shot: the right-arrow shows bright data hard-clipped against the right border. The downward arrow shows that black levels don't go lower than about RGB 60, which is just a dark grayish shade. A waveform monitor (below) is another way of representing the same data. In the waveform below, you see little or no data in the darks (the left side), and data overflowing and crushing against the right unsafe-zone border (brights). http://www.digitalfaq.com/forum/atta...1&d=1460662395 Actually the clipping is so obvious you don't need histograms to picture it. But below is yet another version of the same thing, this time with values expanded as in RGB display. The charts are from VirtualDub. The usual RGB histogram on the left shows brightness (top band) and all 3 color channels. You can see clipping (peaks and climbing up the right-hand wall) in that histogram. The waveform monitor on the right shows the same thing, with clipping at the top of each band. Also, not much data at the dark end, which is odd for a daylight scene. http://www.digitalfaq.com/forum/atta...1&d=1460662474 Adobe has the same or similar tools, doesn't it? They sure charge enough loot for them. I used Avisynth and Virtualdub to clean up some problems. Worked with levels first and applied a negative offset to luma about 20 points lower than the original. This brought the hard brights down a bit and recovered just a tiny bit of bright data, but not much. The rest of the brights, said to say, are destroyed. In the histogram below note the bright white peak on the the right-hand side. No data can exist beyond that point. So the bright parts and part of the upper midtones in this image will lack a snappy dynamic range because of lost data. In VirtualDub I also did some color tweaking. QTGMC deinterlacing was used to try to calm some of the line twitter and excessive interlace combing, general noise, and buzzing edges. The video was then reinterlaced in Avisynth. But overall 75% of that work should not have been necessary, and some of it can't be fixed. Color is still a mess. http://www.digitalfaq.com/forum/atta...1&d=1460662556 As you watch the attached reworked encode (test1b_Reworked.mp4) note some bad line twitter as the camera moves, notably on the ground at the elephants' feet. With most consumer webcams there's always a certain amount of that twitter and broken lines, but here we have interlace artifacts. Test 2b.avi has similar problems. But in that case the line twitter and obvious interlace artifacts, shimmer, and some bad chroma noise from poor done color treatment is a nightmare. The only way to make this video look smooth is to correct levels first in YUV, apply some heavy duty denoisers and smoothers, and finally leave it progressive at 60fps. Experienced users would say the video looks ruined, and I'd have to agree. Restore attached: test2b_reworked_60p.mp4. In the post below I'll attach some clips from problem tapes as samples of restoration, repair, and capture using recommended methods and gear. |
1 Attachment(s)
Restoration and repair samples:
The first sample is from a home video shot by sister with her cheapo Panasonic VHS camera. She was roaming around a house party all day playing kamikaze with camera motion. But the big problem is that late in the day the camera battery was dying. The result was bad underexposure. Bad exposure is common with home video, but this took some work in Avisynth and Virtualdub. An edited mp4 unrestored version of the original is in the forum at http://www.digitalfaq.com/forum/atta...pe_originalmp4. You don't need histograms to see how dark this thing is. The restored version (in the second shot ignore the idiot clowning in front of the camera): http://www.digitalfaq.com/forum/atta...tape_reworkmpg. If you think I used brightness and contrast filters for this work, not so. Those two controls are too primitive for this kind of work. This video was captured with an AG-1980p VCR and ATI AIW 9600XT card. The second sample demo is from one of the ugliest film to retail VHS transfers I've seen in quite a spell. It's full of noise, blotches, huge projector punch holes, etc., and on top of that the color is sickening. This tape was captured with a now-gone Panasonic PV-4664 circa 1996 (no tbc), a DMR-ES10 for tbc pass-thru, and an AIW 9600XT. The repaired version is a first-test sample -- I stopped this project years ago when the film was released on remastered DVD. The movie is telecined, which had to be removed for cleanup and was reinstated for the final encode. Sample frame, with ugly blotch (white arrow). There are several of these -- really nice for a retail VHS, huh? They actually charged money for this. Plus other defects, spots, miscellaneous junk, etc., and horrible color. http://www.digitalfaq.com/forum/atta...1&d=1460663109 original: http://www.digitalfaq.com/forum/atta...es_originalmpg test repair: http://www.digitalfaq.com/forum/atta...mples_aftermpg The third sample is reworked from a 1991 tape recorded to a cheap VCR from a really really really bad cable signal, on really really cheap Radio Shack tape at lousy EP speed (So what did I know in 1991? Those were CRT days). Serious analog cable RF noise and color problems, with color bleed not to be believed. Biggest problem was trying to add some accutance to the blurry image and denoise without ruining everything -- not to mention emulating the original 1954 movie color. The horrible old tape was played with a Panasonic PV-S4670 SVHS VCR (no tbc), a DMR-ES10 for pass-thru, and an ATI AIW 7500 Radeon card. repaired: http://www.digitalfaq.com/forum/atta...back-samplemp4. I no longer have the original capture except for one unrestored snippet of an ad from the same broadcast and cable line. This shows bad that signal could be. The original was run with a very mild degrain filter (a waste of time), but otherwise looks like it did when it was recorded. Original snippet: http://www.digitalfaq.com/forum/atta...0_mdegrain2mp4 So it goes with restoration. Many tapes are easier. I have others, but I guess this is enough bad tape for now. |
Wow! Thanks for all the insight and examples, I appreciate it a lot. I'm comparing how the camcorders play on the TV versus how they are captured with the Elgato in VirtualDub and its so...so much better on the TV when comparing it. So as of now, I don't care what the cost is, in terms of Hi8/8mm tapes being played out of my camcorder and the VHS-C tapes being played out of my other camcorder, what's THE BEST or what few are the best at capturing device/card/tool/USB, anything that I can buy? I'm done with being frugal after seeing how much better, vibrant and smoother it looks like on the TV and on the camcorders' little screens. Forget about the VHS part and the TBC's and all that, I'm strictly referring to Hi8 camcorder and VHS-C camcorder capturing now (but I guess it's all the same thing for VHS as I'll be capturing those with the same card too), if I want the EXACT same thing and source footage what do I buy?
Anything from $100-200 that does the job? And then if I can get some links on the basic and typical restoration things people do in VirtualDub and avisynth, the actual effects, where to go in the options, etc. like what you did, I'd appreciate it. Thanks for taking so much time on me. I didn't expect it to take you hours to work with those 2 little clips, sorry! And that only scares me a little bit about how much I'll have to work on my home tapes, but as you said a good capture card will make it so that a lot that isn't necessary. So please fill me in on what to buy as well, or lay out the best options. I want it to look as good as it does when plugged into the TV, as good as it does on the camcorder's little screen, I'm done with Elgato. |
It took longer making the still images and finding old forum links than it did to create the videos. As you said, problems from improper capture are more troublesome. For better captures it still takes time and learning, but you soon find that there are common glitches and many commonly used tricks for cleanup. Things get faster and easier after the human break-in. You learn to identify defects and how to apply proper tools. There was nothing in the cleanup videos that other users and I haven't used many times over. The first time I tried to repair a simple bathroom tile was a disaster, but with some research and the right tools it went like 1-2-3.
The best cards out there are pro shop gear that cost plenty in dollars and experience. That's beyond the salary grades of most of us, but the cards recommended in the forum's guides (the list I posted earlier) are known to work the way analog capture to digital media is supposed to. At first glance the way tapes play directly into a TV can be deceptive. TV's do make some correction, but the more you learn about video the more faults you see, even on TV. Most viewers aren't very discriminating and those with keener eyesight just say "Yeah, well, it looks screwy because it's old tech". But many, like my wife, give a bad video transfer a frown and ask "Does it have to look so ugly?". The idea behind analog cleanup isn't perfection -- and good luck getting perfect, because it ain't possible with consumer sources. But the results with a little cleanup will be better than the tape. Digital sources have faults, too: we see a lot of DV sources that need work, and plenty of DVD and BluRay "edits" that are totally botched. If you think stuff like UTube video looks nice, try downloading one and blowing it up on a PC monitor full screen instead of watching on that tiny UTUbe web player-- you'll see visual horrors galore. Some of the web stuff is from people who know how to do it well, but most of it is trash. Replace the Elgato with one of the recommended devices. Make short captures to get the hang of things. Some tapes will have problems, some will look better. Hi8 usually looks pretty good. None of those capture devices costs more than $100. That's a far cry from the old AGP line of All In Wonders at $300 or more MSRP, and that was 12 years ago. Still, the new guys hold a credible second place, cheaper than the old AIW's on one count because the new ones aren't full-scale graphics cards. Avisynth and VirtualDub are mainstay cleanup apps. They both take getting accustomed to, but there are plenty of klutzes out there (including me) who soon get the idea. If you're used to NLE's from Adobe, Vegas, etc., it's a different experience and a different modus operandi. NLE's are nice (mine is After Effects, with lossless media). It's difficult to learn from NLE's about the innards of video problems. But that's not what NLE's are designed for. They make many decisions for you that are very often inappropriate. Avisynth and VitualDub aim for precision rather than automation. One example would be the noise in your "Test 2b.avi" sample. Elgato's problems aside, the video is dominated by fine, bristling grain, low frequency flicker (most visible in the reds), and chroma noise (stains that are often called rainbows). Likely that grain and chroma junk are on the tape -- it's camera CMOS noise from shooting in low light, a problem even for some pro cameras. To clean it up I used some Avisynth and VirtualDub filters that specifically address those glitches. The filters included a denoise feature in QTGMC plus SmoothUV and the heavy duty TemporalDegrain, then VirtualDub's CamcorderColorDenoise and a touch-up with a mild NeatVideo. You'll see those filters mentioned often. With those two apps you have at least 400 filters at hand, maybe more. Likely you won't use more than your choice from a set of a dozen or so. How to learn all this? The apps have documentation, but most of the tricks come from browsing in restoration and capture forums. I submitted many samples of misbegotten video projects and took the advice of those who've already gone down the same road. It's learned hands-on. Learning the full potential of Premiere or After Effects is no different. And believe me, learning After Effects is no milk run! Even Photoshop Pro, used the way pros do, was a hassle. I found Avisynth and VDub easier to learn. |
Quote:
Quote:
|
Did you notice the connections on that Aja Kona? They're BNC. Tricky to set up, too. I'd advise one of the USB devices on lordsmurf's list, since your Gateway doesn't seem to have legacy PCI slots. The links (underlining) on the device list in post#2 leads to sites that have them in stock. All the USB devices are pretty much alike in form, connections, and operation. All can be used with VirtualDub.
Your Gateway brings back memories. I've been through 5 of them since Windows 3.1. Eventually I had to build my own -- can't install 3 or 4 hard drives in those retail boxes. OEM's like to pack big drives in their product. Looks impressive ("Gee whiz, I'll never fill up that big drive!". Fat chance.). People who know better use a smaller drive for the OS and program installs, but keep everything else on separate drives. Saves wear and tear on the main drive. If you ever have to reinstall (heaven forbid, but it happens), everything on that drive gets zonked with reformatting unless you backup beforehand. I suggest you get a second hard drive for capture and working storage. Win7 does all kinds of silly things in the background that compete with the i/o controller on that main drive during capture and other work. I have several external drives for archives and edited HD captures. The built-in graphics should suffice for capture and post processing. A new PCIe would be better and faster, but not absolutely essential. I stick with ATI/AMD for better color performance. Built-in audio isn't designed for quality work; they're more like utility devices where quality doesn't matter so much, and they're noisy. SoundBlaster doesn't make the best cards on the market, but they're decent and reliable, and the budget models perform well. On one PC I'm still using an ancient PCI SoundBlaster I took from one of my old Gateway Win2000 machines! For Avisynth and VDub, install 32-bit versions. 32's run pretty much like 64's, the big difference being memory access rather than speed. But you won't be using that much memory. The other reason is that almost all of those 400 filters I mentioned are 32-bit. No one's in a rush to develop 64-bit versions. You can't mix 32's with 64's. Personally I installed Avisynth and VirtualDub on my "D" drive, not my "C" drive. You don't have to take the default C drive. I keep those apps and all its plugins together on a separate drive, which I can copy in toto to other machines and which won't get killed if my C drive goes ballistic. Some very popular Avisynth plugins come with a dozen support files, many of which are standalone filters in their own right. So having a couple of the big boys gets you a whole bunch of bonus plugins. For example, QTGMC uses the LSFMod sharpener (which means Limited Sharpen Faster, Modified). The filter's name doesn't always tell you that much. You don't know what GradFun2DBMod could mean (it helps smooth rough-edged gradients, many of which were in your Test 2b.avi) or even "QTGMC" or "FFT3D" by name. No, FFT3D doesn't refer to 3D video. But RemoveSpotsMC -- well, that tells you something at least. I'm running an HD edit right now that slows things down, so I'll get up a filter list later. The best way to get the hang of them is for you to submit a capture, and someone here will pounce on it and explain what's happening. I can also look up some old restoration threads you might take a look at. I'll try to get up some detail on what I did with your avi samples, but that will likely take until tomorrow. |
Quote:
Quote:
Quote:
Quote:
Quote:
The only thing that's bugging me now is how everyone here always says that a capture card is the best thing you can get and now you're pointing me back towards the USBs...:( I don't want to go that route anymore after seeing what it looks like on the TVs I play the camcorder through. I want that quality, not a what-if and chance, I want the best capture possible and that's direct RCA/S-Video into the computer...not with a USB device as a middle ground... Speaking of, I also found something like this. But it seems its only video and not even S-Video so I don't know how I'd make audio capture work with this thing. |
Quote:
Quote:
Quote:
Quote:
Don't underestimate those USB cards. There are reasons why they're recommended and others aren't. With proper handling they'll look better than tv playback now. I've plugged high-end VCRs from several purchases directly into my LED and plasma TV's, and I never thought they looked so hot. Cleanup made vast improvements later. Quote:
Or build or find an XP machine with an AGP motherboard and look for All In Wonders, from 7500 Radeon to the 9000/9600 series. The X800's (correct numbers?) were PCie cards. |
2 Attachment(s)
Quote:
Quote:
Quote:
And I know this may have nothing to do with any of this but I'm wondering if you know anything about borders being added after Premiere exports. I got videos off of my uncle's Sony Handycam DCR‑SR68 as well for this and that camera's method of transferring video was a tad bit closer to our time here, was in the process of combining clips of days into their own videos in Premiere, exporting them to be placed on the list. And as I'm nearing the end I notice that the exported videos have a small border on the top and bottom that's not present in the originals. And I've worked on these for about 4 days now for no reason now. I can't for the life of me figure out what's wrong, I've exported to preset match source (720x480) which should be exact and the resolution in the description says it is (the source and preview pictures don't even change), I've done it in widescreen, I've done NTSC DV widescreen, all of them have that border. I attached two files showing what happens, the first is the original and the second is an export. You might know what's going on here since you seem to be proficient in everything, but if you don't that's okay. I've been Googling it for answers and have no fix so far. |
Quote:
Quote:
Quote:
Quote:
Some of my threads: http://forum.videohelp.com/threads/3...on-screenshots http://forum.videohelp.com/threads/3...on-screenshots (previous thread; the source file used is different despite looking similar -- they can't be directly compared with the newer thread) Quote:
|
Quote:
|
5 Attachment(s)
Quoted here from the other thread:
Quote:
Below is a capture of original frame 111 from the interlaced test2b.avi. I believe you already know that levels and chroma are out of spec here, exceeding RGB 255. Levels should be corrected during capture as well as you can within the preferred range RGB16-235 to prevent clipped brights and crushed darks. One can always tweak later. Once data gets clipped, it can't be recovered. The frame below shows clipping in facial highlights and in the open book pages, podium trim, and other areas. There's a lot of chroma noise, rainbows, discoloration, and bad red flicker on the front red apron of the podium (or whatever it's called. Pulpit?). http://www.digitalfaq.com/forum/atta...1&d=1461873986 Below, a YUV histogram of this frame. The arrow points to clipped brights at the right edge of the histogram. http://www.digitalfaq.com/forum/atta...1&d=1461874075 Below, and RGB display histogram of the same frame. Arrows point to both brightness and chroma exceeding RGB255 and crashing against or climbing up the right side of the histogram. http://www.digitalfaq.com/forum/atta...1&d=1461874110 Below: 2X blowups of three sections taken from the original frame 111. In still images this noise doesn't move -- but during play this noise jumps around everywhere and looks much worse. This kind of noise and discoloration (rainbows and blotches) eats up encoded data bits that waste more time on noise than on content. Top image: in the lower left corner of the frame, there is discoloration, excessive grain, tiny dropouts ("pinholes"), streaks. Lower left: from the front of the podium, severe red flutter and flicker, with more pinholes, streaks, etc. Lower right: from the upper corner of the frame -- the same noise, different color. http://www.digitalfaq.com/forum/atta...1&d=1461874178 f The attached video test2b_old_new_compare.mp4 compares two videos. I combined the left-hand half of frame from the original test2b and the left-hand half of the same frames from the denoised video I posted earlier as test2b_reworked_60p.mp4 in my post #12 on 4/14/2016. In the comparison video, frames from test2b are on the left, frames from the denoised video are on the right. My emphasis was on denoising and attaining a more natural look and balance for interior lighting, avoiding oversaturation and flared colors. The targets you set for your final output determine the choices made for processing. The script I used is similar to the one from jagabo, with just a few differences. The output from my script added three VirtualDub filters to Avisynth's output: Camcorder Color Denoise, some mild color correction with ColorMill, and a very low-power NeatVideo tweak. Compare the noisy half with the denoised half and you'll begin to learn how to recognize some problems. test2b_old_new_compare.mp4 From this capture it seems to me the tape isn't playing smoothly. Any line tbc you might have in your gear is fairly weak, as there are notched verticals and horizontal shimmer during the play. Those issued can't be completely repaired without severe over filtering. |
2 Attachment(s)
Quote:
Download Avisynth and VirtualDub plugins into separate folders. Plugin zip's have other files you don't want in your plugin folders. I keep a subfolder for each filter, where I unzip the package and keep other articles and web links about the filter. Below are filters I use most frequently. There are many others, but these get a lot of use. If you've downloaded plugins from scripts posted in the other forum, you might already have some of them: Avisynth: --------- ChromaShift27 http://avisynth.nl/index.php/ChromaShift FixChromaBleeding http://avisynth.nl/index.php/FixChromaBleeding Dither Tools + dfttest 16-bit http://forum.doom9.org/showpost.php?...59&postcount=3 DeBlock_QED http://avisynth.nl/index.php/Deblock_QED QTGMC and plugins http://avisynth.nl/index.php/QTGMC SmoothUV http://avisynth.nl/index.php/SmoothUV TemporalDeGrain http://avisynth.nl/index.php/Temporal_Degrain MCTemporalDenoise and plugins http://avisynth.nl/index.php/MCTemporalDenoise SmoothAdjust http://forum.doom9.org/showthread.php?t=154971 The plugins below are sometimes hard to find or have multiple versions: - plugins2.zip (posted in earlier thread. Download link: http://www.digitalfaq.com/forum/atta...ve-plugins2zip): RemoveDirtMC, LimitedSharpenFaster, and DeHalo_ALpha_MT2 - The attached Deen_FixVHSoversharp.zip has Deen v1.b1, and FixVHSoversharp Some filter packages like QTGMC use support plugins that are filters in their own right, such as dfttest.dll. VirtualDub: ----------- TemporalSmoother (VirtualDub built-in) The attached VirtualDub_5.zip has 5 Virtuadub fileters, docs, and some web links: - Camcorder Color Denoise ("CCD") - Color Mill - Gradation curves ("curves") - Hue-Saturation-Intensity - SmartSmoother1.1 I sometimes use NeatVideo v2 or v3 for touch up as a final step. Sometimes, not often. Not free. Worth it, but only for special and very specific conditions. Cannot be used as an all-in-one filter. Care is required or you'll ruin your video. The new version 4 is overpriced and over-hyped. General workflow for analog capture: ----------------------------------------------- VCR -> frame or pass-thru tbc-> capture -> Avisynth/VirtualDub cleanup -> Encode -> Author Playback, tbc, capture cards, etc. ----------------------------------------- - Panasonic AG-1980P (rebuilt) - Panasonic PV-S4670 (SVHS, no tbc, uses pass-thru tbc) - Panasonic PV-4664 (VHS, no tbc, uses pass-thru tbc) - tbc pass-thru: Panasonic DMR_ES10 - external frame tbc: AVT-8710 - ATI All In Wonder 7500 Radeon AGP & ATI All In Wonder 9600XT Radeon AGP - NLE: AfterEffects Pro CS3 for color correction, special effects, slide shows - Capture PC: home built XP Everyone uses different gear and software. |
Thanks for the 2 posts, it means a lot.
Quote:
Quote:
* First off, if I'm capturing with Vdub and a compressor like Huffyuv or UT Video, since I'm capturing it in YUVY/YV12 do the levels I change still get changed in the "RGB" format? * How exactly do I pull up those bars and histograms in Vdub before capturing while getting a live preview of what I'm changing as well? In AvsP (Avisynth) I can pull it up by typing histogram, but that's after the capturing is done. * Lastly, how do I read the data and know what I'm supposed to alter before capture (and then after)? I see your graphs and your arrows but I still don't understand what they mean. I see 3 different sections in one, then 4 in another, strange wavelines. But I don't exactly get how to read any of it...and then what to alter or do after reading it. Anywhere that describes it *all* in full in the simplest way? (if you don't want to write an essay on it yourself) Quote:
Here's what I don't get, do you go from Vdub capture > Avisynth > back to Vdub? Is this how you usually do it? And then how do you get what you did in Vdub lastly to be part of the export when you encode the .avs file (I don't see an option to save what you did in Vdub last to the .avs file)? Quote:
Quote:
Thank you for the filters and download links BTW! Once I get the capturing part down and learn how to do those adjustments beforehand and get back to it I'll return here and download everything you linked to/uploaded when I get to filtering for my Hi8 tapes (and maybe also my VHS/VHS-C tapes if I'm left with having to do it myself and ordering an AG-1980). |
Quote:
Quote:
Quote:
http://www.digitalfaq.com/forum/atta...ubhistogrampng Quote:
The VDUb capture histrogram is a simple YUV luma levels graph. But it works like any horizontal histogram, with dark data at the left side and brights at the right. For reading histograms in general, there are 2 good intros with good images on how to understand what are called "RGB parade" histograms that show three color channels. The links are Photoshop pages for still cameras and Photoshop histograms -- but remember that the graphics principles are the same for motion video. Video is really a stream of still images. Understanding histograms Part 1 and Part 2 http://www.cambridgeincolour.com/tut...istograms1.htm http://www.cambridgeincolour.com/tut...istograms2.htm A YUV histogram usually has 3 horizontal bands, one for luma at the top, and one band each for the U and V data. VDub's capture histogram, like many digital camera histograms, shows lume (Y channel) only. YUV stores chroma data separately from luma. RGB stores brightness and color together in the same group of data bits. How to know what alterations to make? Most are rather simple. If you see in YUV that your levels at the bright end are smashing against the right-hand wall of the graph, then obviously you have to tone down the brights. The ColorYUV filter has plenty of functions for that. In your test2b sample it was obvious that the entire luma spectrum started in the lighter grays at the left end (black levels too high) and overran the histogram at the right. So you could say that all of the luma data in that "white" horizontal band was shifted toward the right during capture. COLORYUV(off_y=-21) in the script I used is an offset with a negative value that reduces all brightness values both dark and bright by 21 points, or about (roughly) 20RGB values. So the entire luma spectrum gets shifted toward the left or darker side, like taking a straight white stick and moving it to the left from one position to another. During capture, control darks and brights with proc amp controls. Brightness raises or lowers black levels. Contrast raises or lowers brights. The two controls interact somewhat, so you have to fiddle a bit to keep things inside the desired range. It doesn't take long to get the hang of it. I believe jagabo covered that partially in the other forum. During capture it won't be perfect, just do the best you can to keep darks and brights away from the "red" zone. Tweak later. When you first start with Avisynth, VirtualDub, color filters, histograms, etc., and even color in Premiere, you do need some initial readup on what the controls do and how to use them. Color correction itself is a big country. But almost anyone can develop an eye and the knowledge to fix color problems. There are tons of free color correction tutorials on the internet for Premiere, AfterEffects, Vegas Pro, and even Photoshop. Because correction principles are effectively the same in all those products, almost any tutorial can be informative. But if you run into a tutor who depends on automated controls for setting color balance, go elsewhere. Those autowhites create disasters. Their authors aren't teaching you anything, they're just pulling off a cheap cop-out. Except for adjusting basic levels, almost all touchy or very precise color work is impossible during capture. It's a subject more suited to post processing. Quote:
The captured video is opened in Virtualdub by opening an avs script written for that capture. VirtualDub runs that script and displays the results. The output from Avisynth, including the effects of any Virtualdub filters that might also be applied, is saved to a new, lossless huffyuv or Lagarith avi as a separate avi (VirtualDub top menu: "File..." -> "Save as AVI..."). Sometimes there might be other things I'd want to do with that new avi, usually because doing everything at once can really be slow-running. Sometimes I open one of those new avi's or partial edit into AfterEffects for some special processing, composite work, titling, or whatever, which I also save as a new lossless avi. When I have the final avi results I want, I open the avi in an encoder. After encoding, I either use the encoded video as-is or, more often, I open the encode in an authoring program. Then I burn it to disc or other media for playback. After the project is complete I delete the intermediate avi workfiles (but I save the scripts!) and archive the initial capture to other media. Because most most analog sources are usually inconsistent in noise, color balance or levels, different parts of the captured video need different processing, color work, or other cleanup. These scenes are cut with an Avisynth script and processed as separate lossless avi's. When all of the segments I want are complete as lossless media, I join them in the encoder. Quote:
Quote:
Quote:
Quote:
I should mention that "x264" isn't a codec. The codec is "h.264". X264 is one of many h.264 encoders. |
Quote:
Thank you for all the help! If I get stuck on converting/restoring I'll come back to this topic and ask away, you're a great teacher. |
I've used all of them. Huffyuv and Lagarith have been around for a long time. Most media players either have them bult-in or can use them if they're installed. The other two are newer, some media players can't use them. I use huffyuv for capture, Lagarith for post-processing (very slightly smaller files, and it works with YV12).
|
Quote:
|
If you capture something using the UT video codec, it can't be UT and h.264 at the same time. In order for a program or encoding app to encode a video to h.264, the video must be decompressed or decoded , then encoded to h.264. A video that is compressed with UT codec and then encoded to h.264 is no longer a UT-compressed video. Video can't be two codecs at the same time.
Some media players can play a UT compressed video, some can't. I think MPC-BE will play the audio but not the video. I've never had a problem with huffyuv or Lagarith. |
Aha, I see, if I go from Ut Video Vdub capture -> encoding to x264(H.264) I should have no playback problems whatsoever. I should stick with Huffyuv unless Ut proves to be significantly better, but it is good to know.
|
3 Attachment(s)
Quote:
mpc-BE and the SMPlayer play UT video streams with picture and sound from See Screens |
Thanks for correcting. I don't remember specifically which players had the problem. I have 6 players, 2 only played the video. Really doesn't matter, if you have at least 1 player that works it's OK, but it's inconvenient.
|
Site design, images and content © 2002-2024 The Digital FAQ, www.digitalFAQ.com
Forum Software by vBulletin · Copyright © 2024 Jelsoft Enterprises Ltd.