A Capture Test of Equipment
 

I was contacted via PM by another forum member who asked me to do a comparison test of capturing with the hardware I've previously talked about on this forum. He was mainly interested in a comparison test of the different VCRs, and he provided a copy of Men In Black 2 as the test tape.

In this first post I've gone a little beyond that request; I decided to do a comparison with and without time base correction. My normal capture chain uses a TBC-4000 (obtained from this forum's Marketplace), but I also have a Snell & Wilcox CVR-600 standards converter which some people (elsewhere) have said can also function as a TBC. And, finally, I did one capture with no frame TBC at all, other than the line TBC and DNR on my Mitsubishi VCR.

The capture chain for each of these clips was Mitsubishi HS-HD2000U (#1) >> TBC (or not) >> ATI AIW X1800 on a homebuilt PC running XP with a quad-core processor (ASUS motherboard). All connections were S-Video. Captures were performed in VirtualDub 1.9.11. I did check histograms before each capture and adjusted the VirtualDub proc amp settings to minimize clipping; each capture chain required a slightly different setting. Video processor amp settings on the TBC-4000 and the CVR-600 were left at default values. Clips were extracted on my Windows 10 machine using VirtualDub 1.9.11 and direct stream copy.

For some fun, I randomized the order of the clips. The first poster to reply with the correct answer as to which clip was captured with which TBC wins bragging rights. I'd also appreciate an analysis from our resident experts as to the TBC performance on each clip.

Coming next will be captures with my four good VCRs; the two Mitsubishis, a Panasonic AG-1980, and a JVC SR-VS30.

:D I don't want to spoil the fun, so I will PM my guesses.

All of the samples show some wavy distortion/interference in chroma. I've also seen this with my HS-HD2000U. Same cables, same workflow with other VCRs = no such issue.

The S-Video cable or cables you're using for the TBC in-chain samples seem(s) to be faulty. 2 of the 3 samples show these Y/C crosstalk artifacts. including checkerboarding in luma. This is caused by chroma leaking into luma and therefore being decoded as luma. https://bavc.github.io/avaa/artifact..._decoding.html

The frame I'm showing here also demonstrates one of the side-effects of the built-in DNR. This is at a shot change, and the chroma is averaged with previous frames that were showing Will Smith's face.

Because of width restrictions, I had to crop off part of the frame and all of my scopes in the top PNG. But the original is attached within a ZIP file.

Attachment 11283
Attachment 11281

The most neutral setting for the Picture Mode of the HS-HD2000U is Play PerfectTape = Normal. Did you use this setting? It can only be changed during playback, by hitting Video multiple times on the remote until the option comes up. Then it's adjusted using the free-turning wheel inside the remote's Jog/Shuttle dial, I think. Maybe using the outside dial.

The frame is 720x480 but there are no black Hblank borders. That means the ATI AIW X1800 is sampling the centre of the image and stretching it horizontally, compared to standard sampling. Myself and others have noticed this with certain capture cards from ATI and others. (WARNING: do not click that link if you have a data cap. It will auto-load many MB of PNG files.)

Okay, here's the side-by-side VCR comparison. A note on the machines:

• Mitsubishi #1 is an HS-HD2000U which was purchased new (floor sample) from Fry's in ~2007. It's had little use as a VCR until I started capturing, and has never needed service.
• Mitsubishi #2 is an HS-HD2000U which I purchased as an eBay crapshoot, intending to use it as a parts donor if necessary. It turned out to be fully functional and I use it often.
• Panasonic is an AG-1980 which was purchased from Lordsmurf. He had it recapped; performs very well.
• JVC is a SR-VS30 which was also acquired from Lordsmurf in 2018. Hasn't needed service since I bought it.

Source tape was again MIB2, the clip this time was taken from a Ghostbusters trailer in the previews. Capture chain was VCR >> TBC-4000 >> ATI All-In-Wonder X1800 on my PCI-Express XP tower. Audio was captured by an ASUS Xonar DGX sound card. For my next post I intend to shift to my other capture chain, which is based around an ATI All-In-Wonder 8500DV on an AGP motherboard Dell tower with a Turtle Beach Santa Cruz sound card.

Hmm. The AG-1980 & SR-VS30U samples have the same S-Video cable problem as above.

HS-HD2000U #1 has the "waves" in chroma mentioned above, but #2 does not!

Sample B looks like the one without any external TBC, Sample C with TBC and darker, Sample A with TBC and as bright as sample B - see attached, I don't like the checkerboard pattern on the TBC's, If those TBC's were mine I would not use them, I just filed a return for one I bought recently, The Pixie, It has a similar pattern but worse.

Curious to know if you can pass the tape on to other members for more samples, I will be glad to participate.

http://www.digitalfaq.com/forum/atta...1&d=1581235396

http://www.digitalfaq.com/forum/atta...1&d=1581235396

http://www.digitalfaq.com/forum/atta...1&d=1581235396

You're right. I've got three suspect S-Video cables (RCA cheapies purchased off-the-shelf at Fry's); I'm going to retire them and try re-running the captures with my better cables.

Ask Mr. msgohan; it's his tape. I'll be happy to mail it on to you next.

I would love it if you passed it along to latreche34 once you're done with it. :) In theory a comparison of the identical tape should be better than comparing multiple copies of the same release.

For the next comparison I would prefer one of the non-Ghostbusters previews: the version of this one on the DVD is a bit different, so we can't compare to the digital version.

The Spider-Man trailer is nice, but letterboxed (even on the Full Screen DVD set it's 16:9 anamorphic; no direct comparison). The Men in Black DVD/VHS ad is the best for comparison, I think. (The Mask of Zorro SE DVD ad is composite-sourced on the DVD.)

Attachment 11292

Not a problem; I can clip out the MIB ad from the trailers and upload it here. Also, I'm going to make the full 16 minutes which I've captured accessible; PM me for the link(s). Be advised that each capture file runs about 7 to 8 GB.

Right now I'm wrapping up for the night. I'm still new enough at this that I need help correcting my technical issues; they may be due to routing and interference as much as to the cables themselves. Especially for the AG-1980...that was connected with some of my best cables. If I can get suggestions for self-diagnosing the issues I'll see if I can resolve the problem and run the captures again; if we're going to use this as a potential comparison benchmark I want my contribution to be as good a quality as I can achieve.

The following five videos were captured with the ATI All-In-Wonder 8500 DV video card and a Turtle Beach Santa Cruz sound card. Capture chain was VCR >> TBC-4000 >> AIW 8500DV; captures again made with VirtualDub 1.9.11. Computer is a Dell tower; much less horsepower than the homebuilt PCI-Express machine...capturing consistently took 38-42 percent of CPU whereas the quad-core homebuilt loafed along at 13%-15%. Captures still look good, though, at least to my inexperienced eye. The JVC SR-VS30 was captured twice; first time with the old (cheap) cables and the second (B) capture with a (hopefully) better-quality S-Video cable I had in my spares. Let me know what differences you see.

Finally, here are two more captures. I re-captured the tape from the JVC to the X1800 with the better cables. Also, bearing in mind the discussions we had recently about DaVinci Resolve and its acceptable codecs, I ran one capture on the X1800 with ProRes 422 in VirtualDub2. Quality setting was 2; one step below max. File size is in line with the HuffyUV captures; a bit larger but not too bad. Capture chain here was Mitsubishi HS-HD2000U #1 >> TBC-4000 >> ATI AIW X1800/Xonar DGX.

The quad-core machine ran about 40% CPU while capturing to ProRes; as I mentioned above it usually runs 13%-15% when capturing lossless with HuffyUV. I tried a ProRes capture on the Dell with 8500DV but the machine just didn't have enough horsepower to keep up. I also attempted a lossless x265 capture with VDub2 but even the quad-core machine couldn't keep up with those demands; a massive number of dropped frames and I pulled the plug halfway through the first trailer.

I'm ready to go to bed, but...

Here is the clip of the MIB DVD ad which msgohan was referring to. For reference, I started with the first black frame before the video fades in and then selected the next 300 frames. This first set is from the same raw captures as above, taken with the VCR >> TBC-4000 >> ATI AIW X1800/Xonar DGX capture chain.

Mitsubishi #1
Mitsubishi #2
Panasonic AG-1980
JVC SR-VS30

The same clip, the same VCRs, this time with the VCR >> TBC-4000 >> ATI AIW 8500DV/Turtle Beach Santa Cruz capture chain.

Mitsubishi #1
Mitsubishi #2
Panasonic AG-1980
JVC SR-VS30

Note that the Panasonic and JVC had consistently larger file sizes, both for the clip(s) and for the capture overall, than the Mitsubishis when captured with the X1800...but the file sizes were quite comparable for all four machines when captured with the 8500DV. Any idea what might be causing that? No real settings changes between captures, other than tweaking of the proc amp settiings.

Here's the ProRes version of the MIB DVD ad.

And now it really is time for bed. Remember, if you want access to the full capture files just send me a PM and specify which file you want the link to. Mr. latreche, if you'll PM me with your name and mailing address I'll forward the tape to you. I will want to wait until I've ironed out as many of the quality issues as I can; hope that you don't mind.

The larger file sizes are caused by the Y/C crosstalk, which is present in the X1800 caps of the Panny & JVC (even 'B' cable) but not the 8500DV caps of them. This may be due to the 8500DV simply being unable to resolve as much detail as the X1800; i.e. the luma dots may be present in the input signal, but blurred by the chip. Or maybe some other reason.

I then suspect that it is a routing and/or equipment interference issue, as I used the same cables.

I believe that checkerboard pattern is TBC related not cables, I've seen it in a lot of TBC setups, Even the new project VHS decode are having the exact same pattern with their virtual TBC, I've given up on a lot of TBC boxes and I always fall back to my beloved setup which is VCR (Line TBC)-> S-Video -> BE75 (Capture+TBC) -> SDI -> SDI2USB -> USB 3.0 -> VDub2 (or BM Media Express), Haven't had any issues with any tape so far, though I don't do this for a living, just a hobby.

As to longer clips, you can always encode and wrap into mp4, It will still reveal the original quality despite the compression, For short clips it is always good to post the raw lossless footage of 20 to 30 seconds.

I've PM'ed you my mailing address for the tape.

In these clips only the Panasonic and JVC exhibit the checkerboard pattern but not the Mitsubishi, Though the sound of the Panasonic is much richer, wonder if this is a linear audio track?

Not at the moment, but around April, I can put this tape through at least 15-20 different capture cards, at least a dozen different TBCs, using at least a dozen different decks.

That checkboard pattern tends to be workflow conflicts (especially between TBC and capture card), and also random (appears/disappears from session to session, nothing changed in wiring or hardware).

Interesting thread. :)