DV is best to convert VHS? Do you agree with this article?
 

http://www.trevorthurlowproductions.ca/dvmpeg.php

There are a bunch of opinions in that article so I don't think anyone is going to agree with all of them.

The author makes some objective technical errors, for example:
That isn't what "co-siting" means at all, and in fact DV 4:1:1 and DV 4:2:0 both co-site their chroma while MPEG-2 4:2:0 doesn't in the vertical direction. All it means is that the chroma sampling is done at the same location as a luma sample. MPEG-2 is still a YCbCr system while what he's describing would be YC.

Always quote the article here. Fixed it for you.

DV = not for conversion!

DV was never invented for transfer. That's BS aka myth.

DV was a shooting format. Period.

The latter half-cocked conversion idea was completely Canopus. Until the craptastic ADVC boxes came along, the only way to capture as DV was using Matrox or Canopus NLE cards in ways for which they were NOT intended. The intention was (A) working real-time with shot DV, and (B) capturing uncompressed or lossless. You'd then integrate everything in Premiere 6 and export to MPEG or SDI. Due to the somewhat willynilly nature of the devices, you could also capture DV. After the Matrox/Canopus NLE cards had been on the market for a while, Canopus added a DV-only product to their lineup, as a cheaper option for folks that were trying to capture analog on archaic IDE Pentium III systems. In those days, you needed a RAID array for better capturing.

By the time the IDE 7200 (and later SATA) Pentium 4 systems were out, such nonsense should have died off as the inferior method that it was. But for some reason -- mostly Canopus BS marketing ("audio lock", etc) -- it persisted. To this day, it's mostly only stubborn "old timers" (pre-2004, a decade ago) that insist it's the best method. In all honesty, it's mostly them (still!) trying to justify that shitty method as good.

It was indeed an eventual broadcast standard, as stating in the first paragraph -- but ONLY for shooting. Big difference.

For capturing, uncompressed (or lossless) was, is, and will probably continue to be best. The runner up format for capture/transfer is MPEG, not DV, due to the way the color information is recorded, and because of the GOP (short of long) use of variable bitrates.

Mhz? My head hurts...

I get tired of people that keep parroting the 90s era Canopus marketing doublespeak about the Mhz on the image. That doesn't matter. What matters is (A) how the colors are stored, and (B) the bitrate specs. Nobody should give a crap about the Mhz of the analog system. This is digital now, not analog.

Unless you're a capture card engineer, not a capture card user, much of the analog spec info is trivia more than anything else.

MPEG is NOT high compression!

The idea that "MPEG = 10:1 compression" shows a complete lack of understanding about MPEG. For example, I-frame only or IP GOP is/was a preferred studio editing format, especially at Sony. If you actually worked at a studio, you'd know this. These days, quite a few MXF files are delivered at 4:2:2 MPEG-2 MP@HL, and at 25+ Mbps bitrates. To say something as stupid as "MPEG = 10.1 compression" is to have only read the help file from a DVD authoring program. In the world of video, it's a dumb as saying the earth is flat.

DV to MPEG = double loss!

That's horrible. You start with 4:2:2 (assuming VHS/Video8 sources), compress is down to 4:1:1, and then compress that down to 4:2:0. It's the video equivalent of putting your color quality through a shredder first. It's reduces the hell out of the chroma by double converting it needlessly. Pick one! Either use 4:2:2 to 4:2:0, or simply start with 4:2:0 to begin with.

Hardware = good!

To say "software = bad" or "hardware = bad" is really stupid. You have to take each piece of hardware (the WHOLE workflow, actually!) and the software on a case by case basis.

For example, with DVD recorders, Panasonic makes some of the worst encoder chipsets there are. Those nasty chips choke on everything from Half D1 to chroma accuracy. On the flip side, LSI Logic made some of the best chipsets around, and it's a shame that the recession killed them off last decade. They just could not compete with manufacturers that wanted to make crappy in-house chips to save $5. Zoran chips are also very, very nice.

The hybrid software+hardware method used by the ATI All In Wonder cards (Theatre Rage/100 and Theatre 200 chipsets) is one of the best their is for analog conversions.

Conclusion...

When it comes to video, this person doesn't know his ass from his elbow. Yes, I'm being mean here. But after 10+ years of hearing this dribble, I grow tired of it. It's crap. This person is mixing up facts all over the place, and is doing nothing more than justifying his own purchase of the Canopus junk.

That last paragraph makes no sense. Why is he "copying" VOB files? That's the WRONG method for decompiling a project for re-edit. It sounds as if he wants to convert MPEG-2 to DV, and back to MPEG-2. Yikes! That's what happens when amateurs start video services -- they have no idea what the hell they're doing.

FYI: MJPEG or JPEG2000 is like JPEG. DV is not even close. MPEG I-frame would also be similar to JPEG, and be a better choice than DV. Not only is MPEG color better, but the 4:2:0 is cross-platform (NTSC/PAL), which is important in a studio setting.

You should also re-extract an MPEG to lossless or AVI before dumping in an NLE. Yes, trying to live edit an MPEG (or even a DV file, to be honest) is a test in patience! Better yet = only editing the MPEG parts as needed, then seamlessly re-integrating everything in an MPEG editor post re-encode. That way, you only introduce re-encoding where needed, and at the merges -- not wholesale on the entire file. But that's not the lazy way, of course.

Even studios make dumb mistakes. I'm irked by the pathetic chroma work done on the Critereon release of "It's A Mad, Mad, Mad World". It's like they just failed to process the last leg of the project for the uncut edit, as it has horrible rainbows everywhere in the footage. Avisynth could have easily fixed that.

Amen. I don't think the reply is mean. Rather, it's simply to the point. If lordsmurf doesn't mind, next time I see a VHS->DV project I'd like to reference the above post, along with several others that are similar and with even more detail. I've had and seen countless debates with those who say they "can't see a difference" between capturing VHS to DV and capturing lossless. DV fans can't see a difference, perhaps, but many people can.

Thanks, lordsmurf. Again.

One thing I haven't seen that would probably help a lot of people make the right decision is a tape captured in DV and a tape captured uncompressed. There are few if any canopus samples available, so if anyone does have a sample to put up it would be appreciated.

I used to have some, years ago, that was intended for the site. But a computer crash many years ago destroyed them. I had 99% recovery, but those files were in that 1%. Several items for digitalFAQ.com were destroyed, along with some various Smurfs-related files (games, comics, images, etc). The "My Documents" folder was hardest hit in the crash.

Recreating clips would require another ADVC, which was never budgeted for. In addition to be one of the worst devices, it's also one of the most expensive. So it's double punishment! We have the DataVideo DAC-100 still, a Canopus rebadge/clone, but it's finicky, and sometimes will not power on.

It'd have also required a large time investment to tediously create the various clips. And it just never happened.

I would not mind redoing a few, but it would require the donation of an ADVC 50, 100 or 300 series. The last test showed the 300 series, and the horrid abuse the filters do to video, in addition to DV loss.

I have some samples, not sure hot these managed to end up 720*576. They should be 720*480.

In any event this is direct from the DV file transferred from the cam over firewire, this is the original source. It was never analog:

http://www.digitalfaq.com/forum/imag...13/07/dv-1.jpg

This is from the cam over S-video going into a Canopus 110:

http://www.digitalfaq.com/forum/imag.../canopus-1.jpg


My advice has always been DV especially for newbies, if you are using a Canopus it just works each and every time. ;)

Hmm. Perhaps you don't understand the point of this thread ? ? The question deals with analog VHS -> DV vs analog VHS -> lossless. However, the samples do prove the point that re-encoding DV involves a quality loss. No one can doubt it from your samples.

Yes I understand the point, the DV source provides a control. The second sample is not renecoded per se because it's being encoded from an analog source.

Image 1: Straight transfer over firewire.
Image 2: Camcorder >> S-Video >> Canopus110

In a sense image 2 is substitute for your VHS source and since we have image 1 we know exactly what it should look like. ;)

Clarify, if you would. I thought you said the source was never in analog format. I also thought this was a discussion about comparing VHS-to-DV and VHS-to-lossless. Is the source VHS? Is the top image a DV capture from that VHS source? Is the bottom image a lossless capture from the same VHS source? What your source proves is that isn't a good idea to capture lossy DV source to lossy DV. I think most readers here already know better than that.

One might also note that while you can deduce a few things from a still frame capture, a still image isn't the same thing as a video source with motion.

It isn't until output over s-video. I'm using a DV camcorder to output the DV over s-video into the Canpous. When I said it was never in analog format I was referring to the source.

If you are going to do any real comparisons you'll need to do something similar because you need to know what the result should be. If I had another capture card I could for example capture that from S-video too. Now we can compare the Canopus capture, the capture from the other card and compare them both to the source.


It's not ideal but if were to do it again I's use .png. Those are from 8 years ago. That said the original and the capture went through the same editor, .jpg compression settings etc. One other thing to note is that's it's not the exact same frame.

Thanks for clarification. The original request was for this:

a) analog VHS -> capture device -> DV capture

b) the same analog VHS -> analog capture device -> YUY2 lossless media (e.g., uncompressed, or huffyuv or Lagarith lossless, etc).

What you posted was:

a) DV source -> DV copy device (Firewire ?) -> exact DV copy
b) the same DV source -> capture device -> lossy re-encoded DV

Where is the VHS original?
:)

Not really because that would imply it's already been encoded. The DV source is of course digital but is output over s-video, that is analog and a substitute for the VHS. Again this provides a control because now you can compare the capture to the source.

Thank you for posting the pics. While the topic wasn't necessarily on native DV video, e.g. captured on a DV camcorder, the S-Video image comparison does help, because after all, while the difference may not be as stark as VHS to Canopus, it still does provide a good comparison.

Any chance you can try a VHS tape for us to take a look at too?

The issue is what are you going to compare it too? I can't post VHS. ;) You have no clue what the source looks like.

I have all that equipment stored, not digging it out. Besides my VHS deck need work.

There is some sample captures here but be aware the VHS tape was a copy.

http://www.digitalfaq.com/forum/vide...html#post21873

Sorry, incorrect on all counts. DV source is lossy encoded video to begin with. Playing it as analog and saving it again to DV is another lossy encode, and DV under any circumstances is a long way from being "like" VHS source.

The two photos he shared demonstrate an analog capture vs. a digital transfer. While not VHS as preferred, it does show, though only slightly due to the codecs matching, the differences in quality. The canopus can copy DV directly via firewire or capture over analog, so he is not in the wrong.

Enjoy. I've seen his VHS->DV captures, and I've seen yours. No thanks.

I've never captured to DV. What are you talking about?

Hold on. Whose VHS->DV captures were linked earlier? I refer to those, whoever they belong to.

Should we take it, then, that your request for VHS source comparisons is now null and void? Since VHS data is stored as a flavor of YCbCr, uses a different color matrix for translation to other color systems, isn't lossy source, and has several other characteristics that distinguish it from DV (tape surface and head noise being two of VHS's DV-unfriendly problems), I guess this means that comparing genuine VHS source processing is no longer part of the question. Perhaps if someone would play a DVD or standard-def AVCHD thru s-video and capture it to DV, would that serve as an equivalent analog capture source? How about DivX and Xvid? I also take it that capturing to huffyuv lossless media is no longer relevant. So....since all of the source variables have changed, and many of the hardware components that would be used for VHS->lossless capture are also removed from the equation, then the original proposal no longer exists. What remains is a different question. The images demonstrate that it's a bad idea to "capture" DV by decoding it in an analog circuit and re-encoding it thru a Canopus DV encoder. It's always better to just transfer or otherwise make an exact copy of a lossy digital source. One might also cast a tad of skepticism about the posted images: they are not from the same moment in the video.

Oh, well. I must apologize for misinterpreting your original proposal and being a little slow to track all of its modifications. However...Like many people I'm pretty firm in my belief that VHS is not like DV, and DV is not like VHS. They aren't just different breeds. They're different species.

Can you contribute an example?

I last used Firewire to capture VHS about 10 years ago. Maybe longer. Haven't made that mistake since, and never will. Captured several tapes, had to do it all over again when I discarded the Firewire setup and went back to AIW's and lossless media. I've seen a few VHS->DV / VHS->lossless comparisons over the years, and just last year by some gal in another video forum. Especially with the last mentioned, some of the differences were subtle, but others (like banding, posterization, clay-face, and highlight-clipping effects in the DV versions) became more obvious once you started working with them. Edge noise, motion noise and jaggies in the DV versions were impossible; I gave up on them and was thankful I had the YUY2/huff versions to fall back on. Using cheap players and a budget ATI USB capture stick for the huff's was a bit of a setback for those samples, IIR -- they're OK, but a far cry from the old AIW AGP's that were very well optimized for analog sources, considering their price point -- not sky-high, but far from cheap.

I have no DV originals, don't know anyone personally who has a DV webcam or shoots regulation DV with a still camera, my Canopus cards were returned and exchanged for better stuff many years ago. The one person I knew who owned a Canopus card is 6 feet under, never liked it anyway, and used his webcam and its software and attachments for direct Firewire transfers. When he wanted a VHS capture, he loaned me the tape so I could capture it to lossless with my ATI, and he tweaked the captured video himself (he was just too cheap to buy an AIW, the scoundrel). I've met a few pros in video and broadcasting who outright refuse to mix media: VHS is for analog capture, DV is for DV transfer, and never the twain shall meet. When I say pro's I don't mean paid-amateurs, most of whom will go with cheap, quick and dirty every time. I've seen their work (My neice wept when she saw her old graduation VHS->DV "digitation"). I'd be embarrassed to show them, much less to charge for them. But that's just my opinion, I guess. PQ standards everywhere are going down the tubes, some of it's worse than bad VHS, and most consumers frankly can't tell one from the other.

End of rant. Maybe someone can find some VHS samples.

Yes and I didn't say it wasn't, it's irrelevant to the discussion about my samples ither than the quality of the encoded capture.

Using the high quality DV as source is not ideal but if you wanted to make true comparisons you'd want to do something similar. Perhaps add some noise and some shakiness to further simulate a typical home VHS.

You're really missing the point, if I had your capture device we can then capture the same video over s-video using uncompressed. Now we can compare the Canopus capture, the uncompressed capture and most importantly we can compare them to the source. You can't do that if your source is VHS.

In any event someone asked for some samples and those are really the best I had as it pertained to this discussion because you can compare to the source. I can post VHS samples all day long but without something to compare them too it's kind of pointless.

Just so it's clear the transfer of DV to computer that produced image 1 was directly over firewire to the computer. The Canopus does have firewire on the input/outside side but I never used it. Not sure if reencodes DV or just passes it through but if I were to guess it just passes it through.

It is versatile device though. If I recall you can use that to capture analog sources directly to a DV camcorder, you don't even need a computer. You can even simultaneously capture to the camcorder and the computer which is nice because you can make a tape backup and get the footage onto the computer at the same time. Don't quote me on that though.

Another cool thing you can do with it is use it to preview from the timeline in your editor. You can send the video from your editor back to the Canopus and it will output over RCA/S-video. Very nice for previewing edits on a TV without spending a fortune.

This has been something I've always agreed with in two situations. :old:

1. You're using a Mac.

There's very few choices for Mac-based hardware, and most of them require the higher-end Mac systems. Of the non-DV hardware that does exist, many have severe quirks that make the hardware unusable for quality work. As such, the Canopus DV boxes are, sadly, the least-worst options.

Of course, a Mac is simply the wrong tool for capturing analog video -- and for most video tasks entirely!

Note that I said "most" -- not "all" tasks! Keep reading...

Unlike Windows systems, a Mac is not a one-size-fits-all universal tool. Instead, Mac (like Linux) is good at very specific tasks. Although Mac is bad for "most" tasks, it has an excellent DV workflow. In fact, NLE editing (mostly DV) and DVD authoring are all that platform is good at. It's just missing too much software and hardware to be useful. Mac is not, and probably never will be, a video platform.

2. You do NOT care that much about quality.

You can use one, and I've given tips on "capturing" (transferring) DV for more than 10 years now. ;)

But it's not ideal. You're going to have a quality hit. If you can live with that -- and hopefully it's something NOT being distributed to others (ie, commercial/pro video) -- then fine. But don't lie to yourself, or lie to others, when you discuss it.

There's zero advantages to DV, aside from (maybe) the laziness factor -- ie, the Canopus boxes can be dummy-friendly, assuming the computer doesn't have quirks/issues (Firewire ports not working, etc).

It's never been home users that have been the recipients of my scorn* -- again, I've helped people with Canopus items for years now. It's the assclown "professionals" that spout nonsense who I target in these DV topics. Anybody that has experience in any kind of professional video setting, studio or broadcast, would never say some of the stupid things I see online. (Note that I mean an actual studio -- "Hollywood", etc -- not a hokey little local strip mall store.)

* Unless they argue, and parrot the same crapola.

________________

thecoalman shoots some good DV video, and if I recall correctly, he had a nice Canon DV camera that took EOS lenses. :congrats:

FYI: thecoalman, your video is almost done! :)

Again, shot DV is fine. It's the conversion where you start to have problems -- more when you start to double-convert everything (tape > DV > MPEG/DVD).

I primarily use a Mac, and when doing video work (recording shows or commercials for clients) I exclusively use FCP X for editing HDV or H.264 because the workflow is easy and I really like the FCPX DVD encoder's performance.

Analog to uncompressed, I use a blackmagic intensity thunderbolt edition, which is more expensive than a canopus but allows for 10-bit captures so I can color grade if I don't use my BVP. ProRes with VHS mainly degrades saturation from what I've seen, but you can edit in uncompressed too.

GL2, no lenses but actually glad I didn't go all out for the XL because HD was coming on the market. I would have spent a hunk of cash on SD camcorder. The GL2 was more than sufficient for my needs then and now because anything I'm producing is going on the web. It's SD but the quality is fantastic.

http://www.digitalfaq.com/forum/imag...4/06/avi-1.jpg




....and the check is in the mail. ROFL

https://www.youtube.com/watch?v=TC5a10kg158

This entire show was shot on DV.

Shooting native DV is fine. It's using it for conversion that is the problem. :wink2:

The more AviSynth I learn, the more I think a VHS tape shot with a good camera beats out DV

The GL2 is a mid range camcorder, you're not going to get this from a VHS camcorder.

This first one is deinterlaced:


http://www.digitalfaq.com/forum/imag...02/temp3-1.jpg

http://www.digitalfaq.com/forum/imag...13/02/40-1.jpg

VHS has way better quality that most of the compressed videos on the internet.

Also, both DV and MPEG-2 are obsolete, better way would be capturing to uncompressed formats.

I hate to burst your bubble here, but MPEG is still the broadcast standard and isn't going away soon. It will continue to exist, if for no other purpose than that of annoying you and reminding everyone how godawful VHS was, is, and always will be, and how filtered, denuded and plastic the newer codecs look.

Encoding for typical internet viewers is the worst thing that ever happened to video. It's bad enough that so much of it is dumbed-down for increasingly dumber viewers, but is now expected to be really ugly as a norm.

If someone bought $1 tapes and had a shitty deck, then there was no other way to have awful quality, just like with all other analog media, but with proper equipment analog might be just fine. MPEG is lossy in both audio and video, so it isn't the best there is.

That's not accurate. :no2:

I'm betting that you're referring to the very narrow MPEG-2 specs used for DVD-Video. DVD-Video (or "DVD") used the MP@ML profile/level with very/overly aggressive sub-10mbit bitrates. But MPEG is a very versatile format, intended for both capturing and archiving, as well as broadcasting. Along with H.264 (which some also state is "obsolete"), MPEG rules both the airwaves and disc storage. You're overlooking the 422 and High settings, which negate all complaints of compression, be it colorspace or bitrate.

Compared to those high-profile/level MPEG, YUY2/YV12/UYUV lossless AVI compression has no real advantage aside from compatibility -- and that's the reason that lossless AVI is suggested (and often required) for analog SD work. With the right workflow, however, MPEG can be equally decent. It's all about details, not simply "MPEG" or "lossless AVI".

Contrary to popular belief (within hobby communities), lossless AVI is actually a minority for ingest, suitable mainly for analog conversions. When I worked for studios, I dealt with a wide array of sources. MPEG outnumbered DV, and lossless AVI was never submitted. Codecs like DNxHD and ProRes422 long ago usurped others, and those are lossy codecs as well. MXF that contained MPEG or DV was extremely popular for 4x3 sources; mostly MPEG, not DV.

MPEG gets much, much more complicated that I'm describing here, too. I have some broadcasting technical papers that would boggle the average user. For example, we didn't even talk about multicast streams.

It's not just equipment (hardware), but the entire workflow. It's starts with hardware, yes, but it'd doesn't stop there. :wink2:

DV: For conversion, yes. In fact, not just "obsolete" but "never was". For shooting, no.
H.264: No, not if encoded properly.
MPEG: No, not if encoded properly.

Hint: Don't use Youtube as the reference for "compressed videos on the internet". Those videos sucking is well known.

Yes, I was specifically talking about DVD-quality MPEG. I forgot that MPEG was also used on a medium with far superior quality than DVD, that medium being VHS, D-VHS to be more specific.

As for capturing and editing lossless is rather easier, as it won't produce as many lossy compression artifacts. It's similar to editing music in wave and mp3, I guess noone would be that crazy to submit mp3s to pressing plant.

It starts with hardware, but it also requires skill.

Well, a VHS source image can be better than DVDs made from it, but you still can't say "VHS is better than DVD". It's still not entirely accurate. For example, if you take a studio source, the DVD will always look better than VHS, even with compression artifacts.

VHS creates chroma and timing issues that would never be present on a DVD. Conversions of that VHS to DVD is the main issue, as the errors become compounded, and the DVD indeed always looks worse. A middle step is required, and that's why I established this site about 15 years ago. (In fact, I was doling out advice in the 1990s, for creating better analog video.)

You can say "DVD looks worse", but only in certain contexts. In others, it's entirely false.

No disagreements here. :)

Guess again. Sadly. Too common. Ugh. :smack:

I didn't say "VHS is better than DVD" (though some people like it more than digital, the same as some like winyls over CDs), I said D-VHS was better than DVD.

Lossy vs lossless, there is nothing to discuss in that matter. Properly made lossy compression might be good enough, but there is a loss of quality from definition. Capturing to lossy to then edit it and recompress isn't the best idea one might have.

Well, again, it depends. As you said, it's also about skill.

A properly encoded DNxHD or ProRes422 loss will be imperceptible. When I used Avid or FCP, with studio sources, I couldn't tell the difference between it and a massive lossless file. And we're talking about me here, not somebody that doesn't know any better. I looked really hard, and saw nothing. I never met a professional that has. (Disclaimer: I never worked on anything that was viewed larger than 100 inches.)

So again, it's about details. Not just "lossless vs. lossy", but which codecs? That matters.

Granted, this is getting off-topic to the initial question: DV made from VHS. (And of course, that's blah.)

Offtopic indeed.

To go back ontopic, I shall put my opinion on the title question: No, uncompressed is better, then edit, upgrade and clean up, then convert to whatever format you like, even DV.

By the way, I need to find my miniDV camcorder.