canopus v other devices

[marada]

I am currently looking for a capture card that will allow me to capture uncompressed VHS footage with either lossless Huff or Lagarith. I have been doing some research and people seem to be recommending the Canopus devices such as the ADVC110. In peoples opinion do you think there is anything else on the market that produces similar or better capture within that price range? I need something that will do both analogue to digital and digital to analogue. Also I have a Panasonic AGDVC30 camera which has an excellent Analogue to digital converter. They use the same converter on this as many of the high end professional cameras. I would not be willing to invest in a capture card if it did not produce significantly better results than my Panasonic camera. The only thing with the Pansonic is it does convert the video to DV and therefore does not achieve my goal of lossless Huff or Lagarith.

I was also wondering what people’s opinions were on whether they believe external devices such as the Canopus are better than internal PCI cards? To be fair I would prefer an internal card as it would be one less bit of clutter on my desk, although I would not be willing to sacrifice quality for this convenience. What I have noticed from most of the cards I have seen, a lot of them are TV cards. I don’t want to watch TV on the pc, and view this as just unnecessary electronics on the card. I am not after an all singing and dancing act, only something that does one thing very well (video capture)

[redwudz]

I've used a ADVC-100 for several years and never any problems. If you are capping from VHS sources, DV quality is more than enough, IMO. It's DV output should be exactly the same as your camcorder DV output. If you are using a internal capture card, Lagarith or HuffyUV works well, but expect file sizes of about 25GB/hour. DV is about 13GB/hour. Unless you are doing extensive editing/filtering, a lossless codec may be a bit of overkill for VHS sources. DV isn't lossless, but it retains most of it's quality for several generations.

What is your final format? DVD? Why do you need analog output?

All capture codecs will have some compression, including HuffyUV or Lagarith. That's why they are called a codec. Uncompressed raw AVI video will be a really huge file, probably twice the size of a HuffyUV capture. That would be more suited to HD sources. It also usually requires a large, fast RAID setup to handle the HD raw video.

[marada]

I am not bothered in the slightest about file size as I just plan to achieve the footage on a hard drive. I have over 6TB of drive space so could not care less. I don’t know why people are so paranoid about file size in this day and age when large hard drives are so cheap. Also one of my computers has a Raid setup, running a core 2 duo extreme over clocked past 4GHZ. I could even overclock to 4.5GHZ with water cooling if I had to. I just want the best quality capture I can get. Why compress video when I don’t have to? Just out of interest is you Panasonic camera giving better results than say the Canopus?

[edDV]

The Canopus only captures to DV format. If you want uncompressed capture (or Huffyuv or Lagarith), use a conventional BT/Conexant type tuner/capture card or a higher end broadcast card. The old ATI All-in-Wonder cards can also capture uncompressed.

[edDV]

You will need a RAID 0 for reliable fully uncompressed capture. Expect ~ 65-120 GB/hr depending on format used.

Huffyuv or Lagarith compression result in ~ 30-40 GB/hr

DV is ~13 GB/Hr

DVD MPeg2 is ~2-4 GB/hr.

[marada]

So what would people recommend for one of these cards? The old ATI all in wonders seem to be a jack of all trades. I can’t help but believe that a card that just did capture would be better than something that pretends to do all sorts of things. Another thing is I would preferably like a card that would work with windows 7 (and if really lucky Linux).

[edDV]

Depends on your budget and source quality.

The BT/Conexant or Philips chipset cards are cheap and cross platform. Review the Linux driver support list to narrow the field.
http://www.linuxhelp.net/guides/tvtuner/
http://www.wlug.org.nz/TvTunerCards
http://www.mythtv.org/wiki/Video_capture_card

ATI AIW drivers are more limited for Linux or Vista/Win7.

Look to AJA/BlackMagic/Ensemble Designs for broadcast quality cards and devices.

Note that high quality VHS capture requires a TBC.

[marada]

So what about from personal experience does anyone have any recommendations. What card gave you the best quality capture?

[edDV]

So what about from personal experience does anyone have any recommendations. What card gave you the best quality capture?

I'm trying to tell you that 90% of the cards that capture uncompressed use the same chipsets so the quality is essentially equal. Prices are as low as $30. You narrow by drivers for Linux. For better quality you move to the higher end (>$500) or TBC/Frame syncs (>$1000). Not much in between.

[edDV]

If you are looking for close to the "best" capture card you are in this class.
http://www.aja.com/products/xena/xena-lse-ls.php
http://www.bhphotovideo.com/c/product/439543-REG/AJA_XENA_LSE_XENA_LSe_12_Bit_SD_Video.html

One of my neighbors.

Just noticed B&H has the XenaLse on sale for $784.95. I'm tempted.

[marada]

"I'm trying to tell you that 90% of the cards that capture uncompressed use the same chipsets so the quality is essentially equal"

I know what you are trying to say but I know years ago this was not always true. Around 2001 I purchased a pinnacle capture card and the quality really sucked (it was around £200 at the time).

Just on another note somebody told me that there were some VHS camcorders made that had digital outputs. Do you know what the quality was like when using these? I have looked around the net but can't find any VHS camcorders that ever had digital outputs.

[edDV]

There were a couple of very expensive JVC DV/VHS combo units. Digital output was DV format.

There may have been a broadcast S-VHS unit with SDI out but I never encountered one. You can get TBC/Frame Syncs with analog in and uncompressed SDI out.
http://www.ensembledesigns.com/products/brighteye/be03

About $1200

Another of my neighbors.

PS, I'd avoid Pinnacle for poor driver support reasons.

[marada]

So if I did not go for a high end broadcast card, it sounds like I would just be looking at an analogue software encoder card. This way my CPU would be doing all the analogue to digital processing and as I have a core 2 duo 4 GHz over clock and raid drives this should work quite well capturing uncompressed footage. Reading up on this I think you would get much better results using a software encoding card than a hardware encoding card as long as you have a powerful CPU. Is my thinking correct?

I know you say unless you go for a broadcast card there is not much difference in capture quality as they use the same chipsets. This may be correct but what about the quality of the jacks on the card and interference? Some must be better than others? What is your take on PCI v PCI-E capture cards? Just wondered if you had any recommendations?

[edDV]

So if I did not go for a high end broadcast card, it sounds like I would just be looking at an analogue software encoder card. This way my CPU would be doing all the analogue to digital processing and as I have a core 2 duo 4 GHz over clock and raid drives this should work quite well capturing uncompressed footage. Reading up on this I think you would get much better results using a software encoding card than a hardware encoding card as long as you have a powerful CPU. Is my thinking correct?

This is called workflow design. The broad choices are three...

1) Uncompressed 4:2:2 capture (or lossless software compression) followed by software editing and filtering, followed by encoding to the desired end format.

2a) Capture (hardware encode) to 5:1 compressed 4:2:0 DV as an intermediate followed by edit/filter/encode. Main advantages are low CPU load during capture and smaller intermediate file.

2b) Capture to another digital intermediate (e.g. Cineform or AIC).

3) Capture (hardware encode) to more compressed MPeg, followed by decode/edit/filter/encode.

The third option has obvious quality compromise. Options one and two can be argued. It comes down to the software and filtering you want to use.

I know you say unless you go for a broadcast card there is not much difference in capture quality as they use the same chipsets. This may be correct but what about the quality of the jacks on the card and interference? Some must be better than others? What is your take on PCI v PCI-E capture cards? Just wondered if you had any recommendations?

They all follow printed circuit board design references from the chipset makers. There may be small differences. The jacks in and out vary (e.g. Euro cards may vary for composite, S-Video and SCART). They vary for data bus and HD support which we are avoiding.

PCI vs PCIe won't make much difference except for full uncompressed. Both handle uncompressed but PCIe has potential for less bus overload. Depends on motherboard.

[jagabo]

So if I did not go for a high end broadcast card, it sounds like I would just be looking at an analogue software encoder card. This way my CPU would be doing all the analogue to digital processing

No, the capture card always does the analog to digital conversion. With raw YUV 4:2:2 capture cards the CPU does colorspace conversion (if necessary) and compression (if desired).

and as I have a core 2 duo 4 GHz over clock and raid drives this should work quite well capturing uncompressed footage.

I don't know why you're so hung up on "uncompressed". Lossless compression with HuffYUV is fast and shrinks the data by half or more and gives you overhead in case the drive is delayed for some reason. The data coming out of the decompressor is bit-for-bit identical to the what went into the compressor (assuming you don't have HuffYUV set to convert colorspace).

Reading up on this I think you would get much better results using a software encoding card than a hardware encoding card as long as you have a powerful CPU. Is my thinking correct?

You don't even need a powerful CPU unless you're trying for realtime MPEG 4 compression or something like that. The main reason you don't want to use hardware MJPEG, MPEG, or even DV capture cards is because they will introduce compression artifacts. If you don't need to do any further processing that may be fine. But if you plan on extensive filtering you want your caps to be as pristine as possible.

I disagree with edDV's assertion that the cards are the same quality because they use the same chips. They do use the same chips but there is a lot more to analog capture than putting a chip on card. The cards have to have the analog and digital portions well isolated, and the analog section needs to be well shielded, to prevent noise from other devices in the computer entering your caps. There is a very big difference between good and bad cards in this respect.

What is your take on PCI v PCI-E capture cards?

I don't think PCI vs PCIe is an issue. Cards can be well or poorly designed for either. It's just a matter of what you have available in the computer. Although, in my experience, most motherboards have the PCIe slots very close to the graphics card -- meaning more potential for noise.

Unfortunately, not many people are interested in raw YUV 4:2:2 catpure these days and the influx of cheaply made cards makes it difficult for companies that can make well designed cards to compete.

In my experience with these type of cards the BT based cards work a lot better in practice than the Philips based cards. This is more of a driver issue than the capability of the chips. But without decent drivers you can't get good caps. I haven't bought or used them in years so I can't give any specific recommendations.

[edDV]

I disagree with edDV's assertion that the cards are the same quality because they use the same chips. They do use the same chips but there is a lot more to analog capture than putting a chip on card. The cards have to have the analog and digital portions well isolated, and the analog section needs to be well shielded, to prevent noise from other devices in the computer entering your caps. There is a very big difference between good and bad cards in this respect.

I agree if they deviated from the reference design provided by Conexant specs can vary. The only way to know is to research the card web references in depth and confirm tests. The cost to do this is far more than the cost of a XENA-Lse.

That said, the Conexant reference design is for consumer, not pro application.

[lordsmurf]

DV is not an optimal method, no. The colorspace compression can and often does muck up VHS input, although it's mostly a problem for cheap boxes, including the ones from Canopus (yes, that includes the ADVC-300, too). Some of the better DV cards work well, but at that cost, it's a waste. A cheaper card at another lower compression would have been just as good. Remember that DV was made for SHOOTING video, not converting it.

The ATI All In Wonder cards, specifically the AGP Radeon series cards, do quite well at this. (I have a few extra ones that I may sell soon.)

To upgrade from an ATI, I would look at some of the offerings by Matrox, Blackmagic and Aja. The "upgrade" would mostly be for HD, and maybe slightly better (tiny, minute, maybe non-existent) improvements on the SD input filtering. I've been looking into this for a while now. (Even then, I'm keeping several ATI cards!)

Remember that VCR and TBC is almost more important than the capture card!
Read more about that topic at http://www.digitalfaq.com/guides/video/capture-playback-hardware.htm

[jagabo]

edDV has alluded to this a few times but it probably bares repeating: VHS capture quality starts with getting the signal off the tape cleanly. That means you need a line time base corrector to get rid of all the horizontal motion between scan lines. And that pretty much means you need an s-vhs deck. Those decks will also usually include some noise reduction which may, or may not, help. The squiggly scanlines are a killer of MPEG compression. This post contains an animated example of what a line TBC fixes:

https://forum.videohelp.com/topic369003.html#1975151

The next step is a full frame time base corrector. These provide rock steady output even if the signal on the tape completely drops out (the picture may still be junk but the timing signals will be clean).

An analog video processing amp is good to have too. With that you can adjust levels and colors (and maybe sharpness) before conversion to digital.

Then you worry about what capture card to use.

[edDV]

DV is not an optimal method, no. The colorspace compression can and often does muck up VHS input, although it's mostly a problem for cheap boxes, including the ones from Canopus (yes, that includes the ADVC-300, too). Some of the better DV cards work well, but at that cost, it's a waste. A cheaper card at another lower compression would have been just as good. Remember that DV was made for SHOOTING video, not converting it.

First, I don't claim to be an empirical VHS capture expert. I have a pallet load of self recorded VHS/SVHS tapes in cool dry storage that I intend to mine someday.

On theory, I don't understand how DV colorspace has limitations for VHS/SVHS capture. The bandwidth is adequate. The "NTSC" 4:1:1 issues don't apply to 4:2:0 PAL but even so the only issues are for heavy chroma saturation as covered in previous threads. These heavy chroma saturation transition issues are encountered mainly in un-natural animation or music videos.

As for Canopus vs "uncheap DV"? boxes is the Canopus DV codecs followed Sony DV chip designs of the early 2000's. These were considered broadcast worthy then.

I may agree that DV codecs may not be the best for VHS if you can show why.

I'm not tied to my ADVC-100, I'm hoping to score a Xena cheap on eBay someday.

[jagabo]

The 4:1:1 chroma subsampling isn't an issue with NTSC DV because the horizontal chroma resolution of VHS is something like 50 lines across the entire width of the frame. But what can be an issue is how the software handles the upsampling to 4:2:2 or 4:4:4 for filtering. Some decoders simply duplicate the DV subsamples. This can lead to obvious banding of colors. Other software will interpolate (or use some other smooth resizing algorithm) to eliminate this problem.

Ah, found an old example: https://forum.videohelp.com/topic351247.html#1850539

[edDV]

I'm hoping the ideal VHS/SVHS capure formula is verified before I get to that 1976-2003 VHS pallet. I know there are great memories in those boxes. I'm still working through the 90's Hi8 tapes.

[marada]

I have looked up cards using both Brooktree / Conexant as well as Philips Chipsets. The problem is nearly all of these cards also include a TV tuner on the board. I view this as unnecessary as I only want to capture from VHS. Also I can't help thinking with all the added electronics associated with the TV tuner this will just add interference, and certainly won't add anything to the capture quality.

Can anyone recommend cards that do not have a tuner and allow uncompressed capture?

[edDV]

I have looked up cards using both Brooktree / Conexant as well as Philips Chipsets. The problem is nearly all of these cards also include a TV tuner on the board. I view this as unnecessary as I only want to capture from VHS. Also I can't help thinking with all the added electronics associated with the TV tuner this will just add interference, and certainly won't add anything to the capture quality.

Can anyone recommend cards that do not have a tuner and allow uncompressed capture?

It would be an older design or a pro-broadcast targeted card. Low cost chip volume targets TV tuners with hardware compression. There isn't a mass demand for a higher quality uncompressed capture only card until you get up to the pro NLE market. The leaders there are AJA, BlackMagic, Matrox, etc.

[marada]

On another note I don’t see why anybody would want a Canopus device. Surely you are better off with a good VCR attached to a TBC feeding into a capture card that you can do uncompressed capture with. I really don't see the point in introducing compression during capture. Get the video onto your hard drive uncompressed then you can do anything you want with it, including editing and finally deciding on a codec for final production. Why would anyone want to limit themselves at capture? It does not make sense considering how cheap hard drive space is these days.

[jagabo]

On another note I don’t see why anybody would want a Canopus device. Surely you are better off with a good VCR attached to a TBC feeding into a capture card that you can do uncompressed capture with. I really don't see the point in introducing compression during capture.

Because DV compression is close enough to uncompressed or lossless for most people -- especially with the low color resolution of VHS (NTSC VHS has about 350x480 luma resolution and 50x480 chroma resolution -- easily handled by NTSC DV's 720x480 luma and 180x480 chroma resolution). Also, DV capture is much more bulletproof than YUV 4:2:2 capture. And lastly, DV is handled very well by almost all editors.

PAL DV is a little different. It has 720x576 luma resolution and 360x288 chroma resolution (4:2:0 subsampling). That means it can't capture all the vertical color resolution of VHS. But the subsampling matches DVD MPEG 2. So, if you're going to DVD it's not really an issue.

[edDV]

Also true uncompressed requires a RAID for storage and large dedicated drives for a large project. The same can be done on one drive (even a 5400 RPM laptop) with DV compression.

You may say that Huf or Lag eliminates the need for a RAID. True but those formats are not directly supported by NLE software and if used, add a CPU decompression load to timeline operations. Other digital intermediate formats such as Cineform are more tuned for editing and filtering demands. Cineform is optimal for high definition editing where RAID becomes prohibitively expensive.

[lordsmurf]

4:4:4 uncompressed vs 4:2:2 uncompressed, right? What is it that SDI can supposedly capture that YUY2 cannot? Yes, higher bits and bitrate, but what else?

I've long noticed DV cooks the colors palette from VHS sources. Many people like the false saturation, but it can lead to a lot of chrominance issues. I'm sad to say the research I had on this subject was lost in 2005, in a hard drive crash. It was one of the few things not backed up at the time. Re-creating it is just not something I've ever re-visited.

In 2009, where I can buy a 1.5TB drive for $100, and use a fast 2Ghz+ CPU (or dual/quad), DV compression doesn't make any sense. Maybe 10 years ago, when you had teeny tiny hard drives, and computers had CPU/RAM/HDD bottlenecks, sure. But not now, and not for a long time now. DV should have given up the ghost 2-3 years ago already, minimum.

[edDV]

4:4:4 uncompressed vs 4:2:2 uncompressed, right? What is it that SDI can supposedly capture that YUY2 cannot? Yes, higher bits and bitrate, but what else?

4:4:4 ? That would take dual SDI runs. It will also carry Alpha for 4:4:4:4 (RGBA or YCbCrA*).

SDI (SMPTE-259M) is 10 or 8 bit 4:2:2 480i/576i with 8 channels of PCM audio all at 270Mb/s bit rate.
That is roughly the same as four cc YUY2.
All that on a single coax that can go hundreds of feet.
http://en.wikipedia.org/wiki/Serial_digital_interface

SMPTE-344M gets you 852x480p/59.94 or 1024x576p/50 at 540 Mb/s
(yes, square pixels)

SMPTE-292M gets you 1080i/720p at 1485 Mb/s
(also square pixels)

SMPTE-424M gets you 1080p up to 59.94 fps at 2970 Mb/s
(also square pixels)


* That was my unique CCIR-601/SMPTE-259M contribution Twin D1 tape machines to record 4:4:4:4 YCbCrA with Pantone color accuracy. In those days D1 machines cost $250K each but ad agencies wanted a perfect Pantone color match.

[marada]

I have done some further research, and aside from the broadcast quality cards people seem to hold the PDI Deluxe in high regards – as shown here:

http://www.pixelmagicsystems.com/products/pdi/pdi_deluxe.htm

The only problem is it uses an old BT878 chipset which uses an 8bit DAC instead of 10bit DAC. How do people on this forum feel about this card? And what about the 8bit v 10bit question?

Another card I have been researching is the Turtle Beach PCI card, although I have read a negative review on this card.

[edDV]

I've long noticed DV cooks the colors palette from VHS sources. Many people like the false saturation, but it can lead to a lot of chrominance issues.

DV uses same "color palette" as DVD or any other Rec-601 based digital video.

If you see chroma shifts, it is happening during A/D conversion.