Analogue Capture Cards

[Lycaon]

Been reading up on Analogue Capture and I’ve decided to invest in some equipment, and do some C&C. Thanks to all on this board for the great info, especially vhelp, Arachnotron, trevlac, and FredThompson. I’m looking for any suggestions, caveats, etc. I’ll post a review later next week. Thanks.

UPDATE: Just got into Los Angeles yesterday after driving cross country from D.C. What's with all the rain? I should get the equipment on Wednesday and can start capturing and comparing. Still waiting on the Compro; out of stock.

Source: All VHS NTSC Tapes.

Hardware:
JVC S-VHS VCR HR-S5902 (http://www.jvc.com/product.jsp?modelId=MODL027009&pathId=49&page=2) (Refurbished $79 eCost) after Christmas, pocket change is in short supply :rolleyes:
EDIT:V-Stream VS-883DV-PRO (http://www.kworld.com.tw/en/product/DV-AV-Expert/DV-AV-Expert.html) (Conexant 23883)
ASUS TV FM Card – 7135 (http://usa.asus.com/products/vga/tvfm_7135/overview.htm) (Philips SAA7135)
Compro VideoMate TV Ultra (http://www.comprousa.com/products/vmtvultra.htm) (Philips SAA7133/4…um, which one?)

I’m getting the Compro to satisfy my own curiosity about the NEC chip, but doesn’t it stand to reason that an AVS filter can do similar clean up once you capture from S-Video. Shouldn’t I aim to capture as purely and unfiltered as possible?

@Arachnotron, any update from this thread about the quality and/or efficacy of TBC and DNR: TBC Discussion (http://neuron2.net/board/viewtopic.php?t=583). I’d love to get the JVC HR-S9911U (http://www.jvc.com/product.jsp?modelId=MODL026935&pathId=49&page=2), but even at $270 Refurbished, I’d rather save the $200 if I can attain comparable result. And my sources don’t seem to be in that poor of condition.

REFERENCES
Conexant CX2388X (http://www.conexant.com/products/entry.jsp?id=720)
Datasheet (http://www.conexant.com/servlets/DownloadServlet/102206A.pdf?FileId=1339)

Philips SAA7133HL (http://www.semiconductors.philips.com/pip/SAA7133HL_V101.html)
Datasheet (http://www.semiconductors.philips.com/acrobat/datasheets/SAA7133HL_1.pdf)

Philips SAA7134HL (http://www.semiconductors.philips.com/pip/SAA7134HL_V1.html)
Datasheet (http://www.semiconductors.philips.com/acrobat/datasheets/SAA7134HL_2.pdf)

Philips SAA7135 Datasheet (http://www.semiconductors.philips.com/acrobat/literature/9397/75010349.pdf)

NEC D64083 (http://www.necel.com/digital_av/english/3dyc/d64083.html)
Compro 3D Y/C Separation (http://www.comprousa.com/products/vmtvultra-3dyc.htm)

RichardBerg’s Analog Capture Samples (http://richardberg.net/VidCapSamples)
Compro Ultra Discussion (http://episteme.arstechnica.com/eve/ubb.x?q=Y&a=tpc&s=50009562&f=67909965&m=855008076631&p=1)

[Sergei_Esenin]

I’m getting the Compro to satisfy my own curiosity about the NEC chip, but doesn’t it stand to reason that an AVS filter can do the same once you capture from S-Video. Shouldn’t I aim to capture as purely and unfiltered as possible?

The problem is, capture chips have their own internal comb filters for composite sources and as far as I know they're all more or less 2D filters without motion adaptive and other advanced features. So for composite sources a comb filter *will* be used, and the question becomes whether you want it to be the capture chip's own mediocre filter or a true fully 3D comb filter. Any AVS comb artifact filters you'd like to use, such as Guavacomb, will be rendered far less effective by the fact that a comb filter has already removed some potential artifacting.

For a long time now many laserdisc users have been outputting composite--which is laserdisc's native format--and routing it to an external device with a high quality 3D comb filter and then capturing it, since all laserdisc players but the $2500-$5000 HLD-X9 and the $700-$1200 LD-S9 either have older/poor comb filters or lackluster playback. A typical configuration involves using a pro-level S-VHS or a D-VHS deck to do the 3D comb filtering and then feeding the PC S-Video or component [AFAIK only the PDI Deluxe/Sweetspot has affordable SD component input].

This newer Compro card basically provides an advanced NEC 3D comb filter for composite signals on the card, eliminating the need for an external comb filter for best-quality captures of composite sources. S-Video signals are unaffected since by definition their Y/C has already been separated outside the card. The NEC comb filter would be used for composite and tuner input.

But the NEC chip does more than comb filtering. It does the card's ADC, and does digital noise reduction. These would be applicable to all inputs.

[Arachnotron]

@Lycaon

Cool! I am very curious about how this compro card stacks up.

As to post combfiltering: as suggested, no matter what you do, some form of Y/C separation will already have taken place during capturing, and I suspect the only way to reverse the artefacts this has caused would be to have a filter that is custom made for the chip you used. Most cap devices use adaptive 2D filters, so you would also need to take the adaptive logic into account. Even then, most capture chips capture at higher sampling rates than the final catured video will have, so a certain amount of information will be irretrievebly lost.

The big advantage the first combfilter in the chain has is that it still has access to all the raw data, i.e. the digitized composite signal.

any update from this thread about the quality and/or efficacy of TBC and DNR:nope, but I finally have the time for this. Expect something in a day or two.

[edit] the capture devices with 3D combfilters are finally beginning to emerge. Read this (http://forum.doom9.org/showthread.php?s=&threadid=87092) thread about the ATI theatrix chip that has one too.

[leonid_makarovsky]

Originally posted by Sergei_Esenin
The problem is, capture chips have their own internal comb filters for composite sources and as far as I know they're all more or less 2D filters without motion adaptive and other advanced features.

Hold on. How do I know whether the card I'm using has a 3D Comb Filter or 2D Comb Filter?? I am using FlyVideo 3000FM from LifeView.

I was under impression that
http://www.semiconductors.philips.com/pip/SAA7133HL_V101.html
was using 3d comb filter. It says on website:
The analog video is sampled by 9-bit ADCs, decoded by a multi-line adaptive comb filter and scaled horizontally, vertically and by field rate. Multiple video output formats (YUV and RGB) are available, including packed and planar, gamma-compensated or black-stretched.
Adaptive comb filter for NTSC and PAL, also operating for non-standard signals

But I guess it doesn't necessarily mean 3D, now does it?

--Leonid

[Arachnotron]

decoded by a multi-line adaptive comb filter

That's adaptive 2-D. A 3D combfilter is motion adaptive, so it needs a framebuffer to hold several fields to function instead of a FIFO to hold only a few lines as the 2D filters do. That is how you can recognize them: by the memory they have, often in the region of 2-4 Mbyte.

There are some links in chapter 12 of the capture guide (right above the title of chapter 12.1) about the various types of combfilters.

[edit] the first link appears to be dead :(, but the second still works.

[leonid_makarovsky]

Originally posted by Arachnotron
That's adaptive 2-D.

I took a look at the product:
http://www.comprousa.com/products/vmtvultra-feature.htm

It says that Y/C 3D adaptive comb filter applies only to NTSC. Doesn't PAL need one too in case I'm capturing from PAL VHS?

If I buy this card and use it, do you think it will work fine with Fly2000 software in native Philips mode?

Originally posted by Sergei_Esenin
But the NEC chip does more than comb filtering. It does the card's ADC, and does digital noise reduction. These would be applicable to all inputs.

So in general this 3D comb filter gives me a better quality if I capture from TV source and VCR's/LaserDisc composite source, right?

I don't believe NEC is doing ADC. Philips does. NEC just separates Y from C. It's not even used in PAL. And if you think NEC is doing ADC, why would they even put Philips chip there on board?

--Leonid

[Arachnotron]

@leonid_makarovsky
The 3D combfiltering is the function of the NEC chip compro has tacked on to the card to add the ability to the philips chip which does not have a 3D combfilter of it's own.

I don't own one, so I do not know if it works with fly. From this (http://episteme.arstechnica.com/eve/ubb.x?q=Y&a=tpc&s=50009562&f=67909965&m=855008076631&p=3) thread I see it works with dscaler, but you must set the nec chip with a separate util from compro. Perhaps it works the same with fly?

The NTSC limitation is for the NEC chip. If so, PAL is out even though the philips can do it. See here (http://www.necel.com/digital_av/english/3dyc/d64083.html)

So in general this 3D comb filter gives me a better quality if I capture from TV source and VCR's/LaserDisc composite source, right?right.
I don't believe NEC is doing ADC. Philips does. NEC just separates Y from C. It's not even used in PAL. And if you think NEC is doing ADC, why would they even put Philips chip there on board?

It must do ADC. A 3D combfilter works in digital. The only question is: does the NEC return the signal to analog afterwards, or does it pass the data on in digital. The latter can be neccecary if the NEC chip does not have a PCI interface for instance. The nec chip can output both digital and analogue, so it is unclear what happens on the compro. Keeping it digital would be best. There have been other hybrid cards like that in the past; even one with a philips chip that could do component video married to a BT878 to give Dscaler compatibility.

The compro page does say "NTSC, PAL, SECAM1 (Products cannot support NTSC and PAL simultaneously)" which indicates the NTSC card cannot do PAL at all. So I suspect the NEC outputs digital.

[Arachnotron]

@Lycaon
@Arachnotron, any update from this thread about the quality and/or efficacy of TBC and DNR: TBC Discussion. I’d love to get the JVC HR-S9911U, but even at $270 Refurbished, I’d rather save the $200 if I can attain comparable result. And my sources don’t seem to be in that poor of condition.

I finished the capping, but processing will take some time. Just a few facts for you:

- enabling or disabling the TBC did not make any real difference in the number of dropped frames. The reason for that, I suspect, is that my HR-S7950 does not have a full frame TBC but only a line TBC. So any damage that goes beyond a few lines will still result in a dropped frame.

-The digital filters do work, though the soft filter is too strong for my taste. The sharp filter is so subtle you hardly notice the difference with the edit mode.

So in general, the only thing I use from my deck is the s-video output. I prefer to filter in post, and the TBC is to weak to help with really bad tapes while for good tapes you hardly notice the difference.

Unfortunately for you, the HR-S9911U does have a full frame TBC, so you cannot tell from my results if the TBC on that VCR is worth it. :)

[edit] I'll post the results later in this (http://neuron2.net/ipw-web/bulletin/bb/viewtopic.php?t=583) thread.

[leonid_makarovsky]

Originally posted by Arachnotron
It must do ADC. A 3D combfilter works in digital.

Ok, then if it works in digital, can I get the same effect using one of the AVISynth filter as part of postprocessing? And if so, what would the filter be?

And the other thing, what is the role of Philips here?

even one with a philips chip that could do component video married to a BT878 to give Dscaler compatibility.

SAA71xx series with BT878? Interesting.

The compro page does say "NTSC, PAL, SECAM1 (Products cannot support NTSC and PAL simultaneously)" which indicates the NTSC card cannot do PAL at all. So I suspect the NEC outputs digital.

I need all in one.

Actually, I have the multi-system JVC VCR with S-Video output. I use S-Video output to connect to FlyVideo. I hope its comb filter is good one as the deck is only 2-3 years old.

With the laserdiscs the story is different. The S-Video output from LD player gives worse quality than composite one when connected to FlyVideo.

--Leonid

[Arachnotron]

Ok, then if it works in digital, can I get the same effect using one of the AVISynth filter as part of postprocessing? And if so, what would the filter be?

And the other thing, what is the role of Philips here?That has already been answered above: you cannot undo bad combfiltering completely in post processing since some of the information the original combfilter had is lost by then.

The philips’ role may simply be to talk to the computer. By using this chip, you can simply use a reference design card and driver and tack a nec chip onto it. Which seems to be exactly what compro did, as the nec chip is controlled by a separate utility.

Actually, I have the multi-system JVC VCR with S-Video output. I use S-Video output to connect to FlyVideo. I hope its comb filter is good one as the deck is only 2-3 years old.Since Y and C are stored separately on tape, even on VHS, the combfilters of the VCR and of the card are not used when you capture s-video so it does not matter.

Anyway, the SAA713x combfilter is a very good adaptive 2D filter. A 3D combfilter will give some improvement, but for PAL and SECAM tape the difference will be small since there is very little overlap between chroma and luma for VHS tape.

For laserdisc there is more overlap because of the higher luma bandwidth of LD. But even there, I would expect the difference to be less dramatic than for NTSC.

By the way, if you cap NTSC you are probably capping PAL-60, so the chroma encoding is still PAL.

[leonid_makarovsky]

Originally posted by Arachnotron
Since Y and C are stored separately on tape, even on VHS, the combfilters of the VCR and of the card are not used when you capture s-video so it does not matter.

Ok, then I shouldn't worry about my VHS caps at all. That also applies to NTSC VHS, right?

For laserdisc there is more overlap because of the higher luma bandwidth of LD. But even there, I would expect the difference to be less dramatic than for NTSC.

By the way, if you cap NTSC you are probably capping PAL-60, so the chroma encoding is still PAL.

PAL-60? Why is that? I'm in US land. My LD players are NTSC.

I guess the best way is to just go ahead and buy a separate 3D Comb filter for LD caps and plug it between an LD player and a cap card.

Thanks.

--Leonid

[Arachnotron]

Ok, then I shouldn't worry about my VHS caps at all. That also applies to NTSC VHS, right? yes, it does.
PAL-60? Why is that? I'm in US land. My LD players are NTSC.sorry, my mistake. From your multinorm VCR comment I assumed you were in PAL country.

@Lycaon:
Sorry for hijjacking your thread. :D

As to the testing of combfilters:
I was wondering if it would be an idea to convert a clip with high and low motion sequences to pure black and white (Y only?) and put it on DVD as a test source. Any colour that shows up after capping composite should be down to combfiltering artefacts. Maybe something similar could be done for the croma, after suitable downsizing to correct for the low chroma bandwidth TV and VHS sources have.

It should be very illustrative, though I am not sure how representative it would be. What do you think?

[trevlac]

Originally posted by Arachnotron
As to the testing of combfilters:
I was wondering if it would be an idea to convert a clip with high and low motion sequences to pure black and white (Y only?) and put it on DVD as a test source. Any colour that shows up after capping composite should be down to combfiltering artefacts.

Sounds like a good test. I would think you could just use one of those resolution test charts and make it rotate at a few different speeds. I think there is a sample avi script that does this at avs.org.

Nice part is you can look for the rainbows and also just see if there is any chroma in the digital values.

I also believe there are options on the capture cards that allow you to turn off the comb filter. Have to check what that means. This and the source would give a good base.

Obviouslly the DVD player would have to create the composite signal. Not sure what the range and quality of this could be or how much it matters. This is always a weak link in testing for me. I don't have pro signal generators or ways to really know how good/bad my test signal is.

[Arachnotron]

Obviouslly the DVD player would have to create the composite signal. Not sure what the range and quality of this could be or how much it matters. This is always a weak link in testing for me. I don't have pro signal generators or ways to really know how good/bad my test signal is. Yes, that is always the problem with synthetic tests. But you can learn a lot from it.

Using a testcard does illustrate things, but I think you should do 'normal' footage too to get an idea how the 3D part works out in real life.

To get an idea how it works out for VHS, an option would be to take a DVD testpic through a downsize-upsize cycle. downsize to say 312x576 and upsize to 704x576 with lanczos4 to get rid of anything above 3 MHz. You would have to do something similar for the chroma. Or maybe simply encode an old VHS capture.

For TV, the bandwidth difference with DVD is not so large for the luma, but obviously very large for chroma (3.75 MHz v.s. 0.55 MHz(?) )

I never have found exact definitions for VHS chroma anywhere. Is the bandwith the same for U and V for instance?

[leonid_makarovsky]

Originally posted by trevlac
I also believe there are options on the capture cards that allow you to turn off the comb filter. Have to check what that means. This and the source would give a good base.

That was my next question. So far I haven't found how to turn the comb filter off, but I have a workaround. There exist Composite to S-Video adapters. Use this adapter to feed your composite output to S-Video input of your capture card.

Now there're a few more questions.

1. Does Y/C separation happen only in digital domain? Are there any Y/C separation filters in analog domain?

2. Suppose I bypass 2D comb filter of my capture card. Can I use AVISynth filters which would be equivalent to 3D comb filter to restore the original colors? If so, what would that be?

3. If the answer to question #3 is no, are there any external high quality 3D comb filters that can be plugged in between my LD player and capture card?

Thanks in advance?

--Leonid

[Arachnotron]

2. Suppose I bypass 2D comb filter of my capture card. Can I use AVISynth filters which would be equivalent to 3D comb filter to restore the original colors? If so, what would that be? I just realized something: this will always be impossible.

The reason is this: suppose you have a CGI picture of letters with what looks like dotcrawl next to it. How do you know it is dotcrawl and not something the artist intended? You don't. In the case of letters, you can make a good guess the dot-crawl does not belong there and filter it out with some sort of chroma smoother. Which will inevitably also filter out some stuff that does belong. But how do you do that in a shot of, say, a river flowing?

A digital combfilter on the other hand has access to the whole original composite signal. It can use the colorburst as a reference to decide which part of the composite signal is chroma and which is luma. Once the picture is digitized, this colorburst is gone and you cannot decide from the digital pixel if a certain croma value is a combfiltering artefact or if it belongs there.

The only way to write your own combfilter would be to sample the whole composite videoline including all the sync and colorburst signals and go from there. In effect, you would do the complete logic of a capture card in software. You could even write your own full frame TBC while you were at it. :)

You would probably be better of using a ADC card for general voltage measurements for that instead of a TV card. At least those would have the C++ libraries and documemtation needed to access them.
National Instuments (http://www.ni.com/) has a lot of that stuff.

[trevlac]

Originally posted by Arachnotron
I just realized something: this will always be impossible.


@Leonid

Don't listen to him!

http://home.earthlink.net/~tacosalad/video/dotcrawl.html

There is a vdub comb filter (color crosstalk fixer) with an explaination of how it works. NTSC only.

1st off ... the 'what if it was supposed to look that way' argument does not work. Sound reasonable ... but no. All forms of digital are some form of compression that loose info. They are based upon the fact that it is possible to remove some info and not have it noticed. You can also remove noticable unwanted info and have a minor effect on what you should keep.

Chroma that sneeks into luma (dot crawl) has a definate pattern that can be found and removed. (don't know how well :) ) No doubt some artist would think dot crawl was a nice effect. I would not want to watch it.

The tacosalad filter across frames uses the fact that from frame to frame for NTSC the SIN of chroma moves 180 degrees in and out of phase. :

Intuitively, this wouldn’t seem useful because a software filter receives decoded pixels, not the composite signal. But all the decoding steps entail simple linear convolution except the sine functions, which cancel themselves out. Hence the interference from the sine functions can be cancelled by simple addition, even after the signal is decoded and the samples converted to pixels. The summation removes from the pixels any error caused by interference in the decoded Y channel; essentially it subtracts erroneous signal energy out of Y, which implicity reinstates any energy missing from the I and Q signals. In short, both the brightness and tint are corrected.


If I understand correctly ... same concept as a 2D comb filter.

The tacosalad filter also allows you to blur a bit within one frame ... which believe it or not may not hurt the sharpness if you are just doing it on the chroma rainbows.

.....

So yes you can do a bunch of stuff in digital even after lots of things have mucked with the signal. Don't shot for re-creating the original analog ... just improve the digital.

---------------------
In the spirit of full disclosure ... I rarely do anything with a composite signal ... and I don't really know what the fuss is all about. VHS is not composite. B-Cast is really the only place where this even should be an issue. ;)

[Arachnotron]

Don't listen to him!

http://home.earthlink.net/~tacosala...o/dotcrawl.html

There is a vdub comb filter (color crosstalk fixer) with an explaination of how it works. NTSC only.
Why Trev, thank you :(

ok, my example was bad here. Removing cross talk of luma into chroma can be done in post, since there is a direct relation between the luma and the chroma which you can use to remove it.

However, the question was if you can do *all* combfiltering in post, not fixing the result of improper combfiltering. I still remain sceptical if this is completely possible, though it might be through some iterative process.

I suppose the way to test it would be to capture s-vid while feeding the same composite signal to both Y and C. And you still have to switch off the combfiltering of the chip, since at least for the philips some parts of the combfilter remain active even when capping s-video.

The source would have to be TV or DVD to make sure you really have a good overlap between luma and the chroma subcarrier.

And someone would have to program the filter. Now I still cannot code C well enough. But Trev can. :devil:

[trevlac]

Originally posted by Arachnotron
But Trev can. :devil:

I'm still too busy trying to get my 'cheap' SATA Raid5 working. :(

Maybe you can come over to my house and set it up for me.

[Arachnotron]

Just a quick reminder:
The latest Dscaler (4.1.11) includes a combfilter plugin that can handle both PAL and NTSC and looks to be 3D too. If your card is SAA7134 based you need to change it from the default setting to high quality mode.

Maybe it can be used as a quick check to see how far you can go in post with this (or as a reference to card filters)

[leonid_makarovsky]

Originally posted by Arachnotron
Just a quick reminder:
The latest Dscaler (4.1.11) includes a combfilter plugin that can handle both PAL and NTSC and looks to be 3D too. If your card is SAA7134 based you need to change it from the default setting to high quality mode.

Maybe it can be used as a quick check to see how far you can go in post with this (or as a reference to card filters)

But this filter can only be used in realtime capture, right? Not as postprocessing. I will shoot e-mail to the author of Fly2000 and see whether he has some suggestions...

BTW, so I guess I can bypass 2D comb filter by plugging Composite-to-S-Video adapter to the S-Video input of my capture card. The image there is black and white now. Can I restore the colors?

--Leonid

[leonid_makarovsky]

Originally posted by trevlac
I rarely do anything with a composite signal ... and I don't really know what the fuss is all about. VHS is not composite. B-Cast is really the only place where this even should be an issue. ;)

Hold on a sec. If you don't have a VCR with S-Video out, you have no choice, but digital Y/C separation, now don't you?

With my case, I have an LD player and now I understand why my LD captures look like s&*t comparing to VHS captures.

--Leonid

[Arachnotron]

But this filter can only be used in realtime capture, right? Not as postprocessing. I will shoot e-mail to the author of Fly2000 and see whether he has some suggestions...

BTW, so I guess I can bypass 2D comb filter by plugging Composite-to-S-Video adapter to the S-Video input of my capture card. The image there is black and white now. Can I restore the colors?
Dscaler cannot capture well, and the filter cannot be used like it is as an AVS filter. The Dscaler suggestion was only a theoretical one to see what a postprocessing combfilter is capable of.

A composite to s-vid adapter may have to be self-build. They obviously exist in the form of stand-alone combfilters, but a pure composite splitter is something else. It should not be too difficult, but you would probably need some resistors to get the correct impedance back. The adapters that come with capture cards simply connect the Y pin of the Y/C with composite. So if you use those in reverse, composite is fed into Y and C is left without signal. Hence black and white. They do not feed composite to both Y and C.

You would also need some hack to switch of the 2D comb completely, as by default parts of it remain active even when the chip is in s-video mode.

After reading Trev's comments, I now am beginning to believe you may in the end achieve the same result by capturing composite with the philips and running a motion aware combfilter afterwards in post. Only someone should port the Dscaler filter to avs for that. Or maybe use the vdub filter trev pointed to.

Should be easy to test: simply look at a suitable composite source with dscaler, and see if it improves when you switch on the combfilter plugin.

[fisix]

http://members.aol.com/ajaynejr/vidcomb.htm

i'm still not sure if a 3d comb filter in a tv actually digitizes the signal, filters it, and then re-renders it in analog, OR if it digitizes the signal, and then figures out parameters to use in some analog filter chain that is applied to the analog composite video which is then passed onwards. analog signals can be ran through delay lines to slow them long enough to perform calculations on the side...

in any event, if it is the former, then consumer analog tvs would have been color limited to 8 or 10 bit at some point, which i don't think is the case. you can clearly see color banding in digital rear projection tvs that you can't see in crts (as long as you have a true analog source).

so, the question is: what is the NEC 3d comb chip doing on the compro card? is it supplying Y/C filter parameter info to the CX chip, or is it passing separated (digital or analog?) Y/C to the chip? it's hard to say because most tv's are moving towards a full digital path so it isn't clear what the chip would be used for in a tv.

i'm going to guess and say that the NEC chip actually passes digital Y/C separated video to the CX chip. i don't see anything that would indicate an analog delay line, and i don't think the NEC chip would output analog (separated) video... though i'm not sure the CX chip ever expects to see digital video as an input..

purely a guess. would like to know for sure.

[leonid_makarovsky]

http://www.zoran.com/products/ip_cores/video_decoders/cvd2.html

Is this used any where in TV Tuner cards?

--Leonid

[leonid_makarovsky]

Originally posted by fisix
so, the question is: what is the NEC 3d comb chip doing on the compro card? is it supplying Y/C filter parameter info to the CX chip, or is it passing separated (digital or analog?) Y/C to the

I believe there's no CX chip on Compro, but a Philips one. As a matter of fact Canopus seems to be using NEC chip in their MPEG PRO MVR.

http://www.canopus.us/US/products/MPEGPRO_MVR/pt_MPEGPRO_MVR.asp

--Leonid

[Arachnotron]

@fisix
so, the question is: what is the NEC 3d comb chip doing on the compro card? is it supplying Y/C filter parameter info to the CX chip, or is it passing separated (digital or analog?) Y/C to the chip? it's hard to say because most tv's are moving towards a full digital path so it isn't clear what the chip would be used for in a tv.From the very sparce info available on the nec site this chip can only output two things: a digitized (10 bit?) signal or an analogue Y/C signal.

From the compro site I also get the impression the chip is used as a pre-Y/C separator (no mention of a bypass for instance), and not as a pre-analyzer for the SAA7134. Also, I could not find any indication in the SAA7134 datasheetthat there is any option to steer the combfilter externally. You can switch some features on and off, but from the looks of it that also resets the chip; nothing that would facilitate doing this dynamically on a line-by-line basis.

passing the y/c on in digital would be the best quality-wise. But doing it analogue would be easier to build and program for compro. They would only need to tack it on to an exoisting card without modding the existing drivers and add a separate control unit for the NEC.

My guess is that if the defauls SAA driver control sliders for brightness etc. are still available, they pass on analogue. AFAIKR these controls in the SAA are located before the pixel engine, so if they still work the SAA is still doing digitizing itself. If the NEc does the digitizing and passes the digitized signal on, these controls would be located in the NEC driver only.

Any compro owner care to shed some light on this?

[fisix]

Originally posted by leonid_makarovsky
I believe there's no CX chip on Compro, but a Philips one. sorry, you are right, i was typing away and got confused with all the different cards.

interesting notes on your link:

MPEGPRO pre-filtering features:
> Digital 3D Y/C Separation – Separates the Y/C (S-Video) signal to RGB using chronological information between signals. (applicable for NTSC only)
> Digital 3D Digital Noise Reduction - Eliminates noise in the analog video signal by detecting noise based on the characteristics.
> Digital 3D Frame Synchronizer - Stores incoming signals into a frame buffer to supply a stable, synchronized signal to the MPEG encoder chip.
> Digital Line Time Base Corrector (TBC) - Detects images with strong vertical jitter and employs powerful correction methods to repair the jitter.
> Digital Auto Gain Control (AGC) - Automatically adjusts the input video level.

can the nec chip do all this? or is the second line kinda of a misnomer because some other chip applies the noise reduction (and everything else, probably) to a supplied digital rgb signal? strange.

the one thing about the 3d comb filter, it seems that good ones always have a fallback to a 2d comb, and we know the phillips chip has an internal 2d comb.. but i guess that gets completely bypassed because it probably gets supplied either separated y/c or direct digital rgb (i assume all rgb on these cards is digital).

i'm on the waiting list for one of these cards...

edit> scratch that, mine should arrive sometime monday or tuesday

[leonid_makarovsky]

Originally posted by fisix
can the nec chip do all this? or is the second line kinda of a misnomer because some other chip applies the noise reduction (and everything else, probably) to a supplied digital rgb signal? strange.

I don't know. Now I remember that I was told by a Canopus represantative that ADVC-100 seems to be using NEC digitizer. I don't have this e-mail anymore. For their MPEG products, I honestly don't know. I was actually going to buy one, but then I found out that it converts to MPEG on the fly (h/w) and there's no choice for having LPCM audio - only mpeg2 compression.

I can shoot them e-mail and ask what chip they are using on their boards.

--Leonid

[leonid_makarovsky]

I got a response from Andrey Zolotarev regarding the turning 2D comb filter off in Fly2000. The e-mail is in Russian, here is the translation:

> Is it possible to turn the 2D filter off during the capture in
> FlyVideo 3000 FM and how would I do it?

Why?
2D comb filter can only be turned off during S-Video mode. It's absolutely necessary to have it in Composite mode. According to Philips specs this is what's 2D Comb filter is doing:

PAL - Comb Luma + Comb Chroma
NTSC - Comb Luma + Comb Chroma
SECAM - Comb Chroma

If you turn Comb Luma off during PAL capture, the picture quality will become worse, sharpness will get reduced and colors will be shifted.

I didn't check NTSC, but I think it's going to be the same.

SECAM doesn't give any noticeable changes.

According to specifications both comb filters must be turned off during S-Video capture. In this mode both chroma and luma must be processed separately by SAA713x chip.

> Also do you support this card:
> http://www.comprousa.com/products/vmtvultra.htm

Generally speaking Compro cards *are* supported. Although I don't know about this particular card. There's an additional chip NEC D64083 existence of which is rather confusing. As far as I understand NEC D64083 is fed by the composite signal from RF or Composite input. It then splits the signal into Chroma and Luma and feeds them to Philips which is working in S-Video mode with 2D comb filter turned off. But this is not a true S-Video signal which is supposed to have Chroma and Luma separate in the first place.

I can turn the Comb filter off in Fly2000TV if you really need it.

Generally speaking Compro cards are not that great. I have a card Compro VideoMate Gold+ II which is almost the same as Ultra but without NEC D64083. I didn't like it at all.

1) The RF module from LG is not very good. (very foggy picture)
2) The Radio is based on TEA5767 which is not plugged into SAA7134's input so the sound goes to the output right away.
3) The remote is very inconvenient especially the cross (Leonid: I don't know what he means by "the cross")

1 and 2 disappointed the most.

Sincerely,
Andrey Zolotarev

--Leonid