VHS capture with AG1980

[RS456]

Can anyone see a difference between the two clips? Is the 2nd one an improvement?

One is AG1980 with TBC on -> Hauppage USB live 2

https://streamable.com/r8axcb

The other is AG1980 with TBC on -> Panasonic WJ-MX50A -> Hauppage USB live 2

https://streamable.com/mywtc1

[RS456]

I can barely see a difference. I am asking the opinion of expert eyes.

[latreche34]

The first one is darker, the second one is slightly brighter with a minor quality degradation due to some sort of processing of the video (TBC, DVD recorder ...).

[RS456]

Quote:

Originally Posted by latreche34

The first one is darker, the second one is slightly brighter with a minor quality degradation due to some sort of processing of the video (TBC, DVD recorder ...).

The 2nd one went through a Panasonic Video mixer with the TBC on on the Panasonic VCR. Overall how are the expert clips compared to what you seen regarding VHS digitization.

[dpalomaki]

ALL analog amplifiers and A/D and D/A converters introduce some distortion, noise, and quantization error, however slight. An extract of the Pana's specs is attached. Note especially the 4:1:1 color sampling which will be a concern if you are doing restoration, and the -3 dB frequency response at 4.5 mHz.

With quality gear or low quality source material it may not be visible but in general it is best to minimize the number of devices in the signal chain, adding devices such as proc amps, TBC, etc. only if they provide a visible benefit. In many cases the external TBC is necessary to provide a stable signal.

[RS456]

Quote:

Originally Posted by dpalomaki

ALL analog amplifiers and A/D and D/A converters introduce some distortion, noise, and quantization error, however slight. An extract of the Pana's specs is attached. Note especially the 4:1:1 color sampling which will be a concern if you are doing restoration, and the -3 dB frequency response at 4.5 mHz.

With quality gear or low quality source material it may not be visible but in general it is best to minimize the number of devices in the signal chain, adding devices such as proc amps, TBC, etc. only if they provide a visible benefit. In many cases the external TBC is necessary to provide a stable signal.

What are your suggestions for workflow with AG1980? Capture card, software, and TBC.

[latreche34]

If that mixer uses DV spec to do the video processing then it is the most useless piece of crap, May as well save the space.

[RS456]

Quote:

Originally Posted by latreche34

If that mixer uses DV spec to do the video processing then it is the most useless piece of crap, May as well save the space.

So Panasonic MX50A is useless. I just went with it because it was highly recommended as a TBC substitute by users here and videohelp but I did not realize that the MX50A video processing spec is worse than the previous models. What should be the minimum processing spec I should look for.

[hodgey]

Quote:

Originally Posted by dpalomaki

Note especially the 4:1:1 color sampling which will be a concern if you are doing restoration, and the -3 dB frequency response at 4.5 mHz.

I think the frequency response is fine, at least for standard VHS. VCRs usually low-pass filter the output from standard VHS at around 3 MHz think, so -3 dB ad 4.5 MHz may be okay for standard VHS, it's a fair bit higher than the VHS luma bandwidth. It may be designed to cover the full standard NTSC broadcast signal but I'm not entirely sure. It has a sampling frequency of 13.5 MHz, the same as the SAA7111 chip I think is used in the Datavideo TBCs. (This is the sampling frequency that gives the 720 active pixels per line).

4:1:1 may be a concern though as noted, 4:2:2 (with 8-bit luma) is what you usually get from capture cards. 4:1:1 at the standard sample rate should in theory be enough to cover the chroma bandwidth of the signal you get from a tape, but at least for DV people have reported that it doesn't work as well in practice. It's not going to be compressed like DV though, so you would have to have a look at the chroma output to see if it matters, some of the issue with it for DV is the lossy compression. The Datavideo TBC-1000 and TBC-3000 and AVT-8710 are 4:2:2 (8-bit luma).

[dpalomaki]

Quote:

So Panasonic MX50A is useless.

Not necessarily. It was produced for a specific purpose/use and was in line with the practice of its era. While it provides some TBC effects as a result of doing frame synchronization it was not intended to clean-up poor video - its usefulness there may be limited. However, if it works for you and meets your (or your clients') expectations that is what matters. Just don't use it if the output from the AG-1980 doesn't improve when it is in the signal path.

The AG-1980 TBC is a full field (not full frame) so it may still leave some issues that need to be fixed by a full frame TBC. You have to try the several options and decide which works best for the tape you are capturing.

[RS456]

I have both VHS and 8mm that I need to digitize. The 8mm is being played through a digital 8 camera that is backwards compatible with analog 8mm. I am not 100% sure yet but I think that camera has a internal TBC because my 8mm tapes look a lot richer and smoother then it looked in my old analog 8mm video camera playing those same tapes. So should I consider something else other than the MX50A?

Quote:

Originally Posted by hodgey

I think the frequency response is fine, at least for standard VHS. VCRs usually low-pass filter the output from standard VHS at around 3 MHz think, so -3 dB ad 4.5 MHz may be okay for standard VHS, it's a fair bit higher than the VHS luma bandwidth. It may be designed to cover the full standard NTSC broadcast signal but I'm not entirely sure. It has a sampling frequency of 13.5 MHz, the same as the SAA7111 chip I think is used in the Datavideo TBCs. (This is the sampling frequency that gives the 720 active pixels per line).

4:1:1 may be a concern though as noted, 4:2:2 (with 8-bit luma) is what you usually get from capture cards. 4:1:1 at the standard sample rate should in theory be enough to cover the chroma bandwidth of the signal you get from a tape, but at least for DV people have reported that it doesn't work as well in practice. It's not going to be compressed like DV though, so you would have to have a look at the chroma output to see if it matters, some of the issue with it for DV is the lossy compression. The Datavideo TBC-1000 and TBC-3000 and AVT-8710 are 4:2:2 (8-bit luma).

[RS456]

I have no clients. I am just trying to perfectly digitize my home movies . My tapes are relatively in good condition except for some noise here and there which is something I want to get rid off in the digitized version. Any suggestions for me?

Quote:

Originally Posted by dpalomaki

Not necessarily. It was produced for a specific purpose/use and was in line with the practice of its era. While it provides some TBC effects as a result of doing frame synchronization it was not intended to clean-up poor video - its usefulness there may be limited. However, if it works for you and meets your (or your clients') expectations that is what matters. Just don't use it if the output from the AG-1980 doesn't improve when it is in the signal path.

The AG-1980 TBC is a full field (not full frame) so it may still leave some issues that need to be fixed by a full frame TBC. You have to try the several options and decide which works best for the tape you are capturing.

[dpalomaki]

A quick solution is to use a product like NeatVideo or the denoise filters included in many NLEs. But they are not favored by most of the experts at this site (for starters it is easy to over process and there is an learning curve). Search the forums for suggestions as to what the experts here favor.

FWIW: You are your client. And there is no "perfect," just degrees of less than perfect. When digitized the continuous analog signal ended up coded into a series of steps; it is 25 or 26, not 25.49 for example. The decision is how much time and money do you want to spend before it is close enough to perfect for you.

Some folks detail their car, and others let the rain wash theirs. Ultimately it is a matter of personal preference.

Also keep in mind that what looks pleasing on the screen may bare little relationship to what was in the real world - think Glamor Shots photos as an example.

[RS456]

You are the person who helped me fix my AG1980 when others said it has to be looked at by an expert. I consider your opinions very practical. I looked up NeatVideo, if it really does what is shown then it seems to be a very good plugin. I will look at alternatives also. I am considering trying a cheap TBC that converts S-Video/Composite to HDMI and considering a HDMI capture card that supports upto 1080 60 Hz that clearly supports lossless. Do you think it is worth a try? Also should I stick with VirtualDub as the capture software? The previous samples were made with Hauppage capture software (not lossless). I am actually more use to using MeGui but as far as I know it does not support live capturing.

Quote:

Originally Posted by dpalomaki

A quick solution is to use a product like NeatVideo or the denoise filters included in many NLEs. But they are not favored by most of the experts at this site (for starters it is easy to over process and there is an learning curve). Search the forums for suggestions as to what the experts here favor.

FWIW: You are your client. And there is no "perfect," just degrees of less than perfect. When digitized the continuous analog signal ended up coded into a series of steps; it is 25 or 26, not 25.49 for example. The decision is how much time and money do you want to spend before it is close enough to perfect for you.

Some folks detail their car, and others let the rain wash theirs. Ultimately it is a matter of personal preference.

Also keep in mind that what looks pleasing on the screen may bare little relationship to what was in the real world - think Glamor Shots photos as an example.

[dpalomaki]

Quote:

Originally Posted by RS456

... I am considering trying a cheap TBC that converts S-Video/Composite to HDMI and considering a HDMI capture card that supports upto 1080 60 Hz that clearly supports lossless. Do you think it is worth a try? Also should I stick with VirtualDub as the capture software? ...

If you are comfortable with VirtualDub and it meets your needs I see no reason to change.

No one can comment accurately on the hardware without knowing what it is (make and model). However, in general cheap is usually cheap - you typically get what you pay for. Occasionally you may find a low cost gem but gems are rare. Since the world has moved to HD, most new devices designed with HD in mind treat SD as an afterthought and may give less than optimal results.

An HDMI capture card is nice to have for HDMI sources such as cable boxes, cameras, and some tape-based HDV camcorders, but serves no purpose for legacy SD sources. Upscaling SD to HD makes sense only at the display stage, and most modern TVs and BD players will do a better job of it than cheap devices and/or consumer software.

Keep in mind that if the source material is SD then upscaling to HD for lossless capture, editing, and restoration will add nothing but processing artifacts and bulk. Also, because HD files are substantially larger than SD files they require more processing power/time and storage space.

[RS456]

The device itself is specifically designed to play PAL and NTSC tapes on HDTVs so I was thinking it might be worth a try. There is also an option to keep it in the same ratio.

[latreche34]

Quote:

Originally Posted by hodgey

The Datavideo TBC-1000 and TBC-3000 and AVT-8710 are 4:2:2 (8-bit luma).

I though TBC-1000 uses 10 bit processing, Usually consumer TBC's use 10 bit and the pro ones use 12 bit, Not necessarily a general rule though.

[hodgey]

The specs state 8-bit 4:2:2 and same luma bandwidth as the panasonic mixer. I think it uses the same SAA7111 video chip as the TBC-3000 (though I'm not 100% sure), which has a 8-bit A/D converter I believe. The AVT has a SAA7114. That chip has a 9-bit A/D converter and a bit more options for output formats, but that chip normally outputs 8-bit (luma) 4:2:2 as well unless configured specifically with the 8 extension port output pins. 10-bit processing is a more recent thing I think.

As for the original poster, I would suggest posting a clip of a problem area, the panasonic mixer may well be sufficient. If you have a lossless clip it should be possible to see if there is a noticeable effect on chroma due to 4:1:1 processing as well.

[RS456]

My very first post on this post has the two samples (not lossless). One without the mixer and the other with the mixer in between. I was barely able to tell the difference between the two. I just got the mixer. It has a few issues. Might need recapping (Audio does not work, Static screen is getting a strange greenish tint, and manual settings not working). I haven't gotten as far as trying lossless with the mixer yet. On top of this I am unable to preview the video on VirtualDub when capturing lossless with Diamond VC500 and Hauppage USB2Live. I can view the captured video after I stop the capture process.

Quote:

Originally Posted by hodgey

The specs state 8-bit 4:2:2 and same luma bandwidth as the panasonic mixer. I think it uses the same SAA7111 video chip as the TBC-3000 (though I'm not 100% sure), which has a 8-bit A/D converter I believe. The AVT has a SAA7114. That chip has a 9-bit A/D converter and a bit more options for output formats, but that chip normally outputs 8-bit (luma) 4:2:2 as well unless configured specifically with the 8 extension port output pins. 10-bit processing is a more recent thing I think.

As for the original poster, I would suggest posting a clip of a problem area, the panasonic mixer may well be sufficient. If you have a lossless clip it should be possible to see if there is a noticeable effect on chroma due to 4:1:1 processing as well.

[dpalomaki]

As implied above, it is not unheard of to use, for example, a 10-bit A/D converter chip, and just ignore the two least significant bits essentially using it as an 8-bit A/D. Or more significantly used 8-bit converters for luma and chroma, but only used 6-bits for the chroma. (Reduced memory cost back in the days when ram was not cheap and did not waste limited available processing power on detail that might be below the noise floor). It worked for the typical material that was being recorded and the limited quality displays used for home video back then. It is not up to today's quality standards.

Some devices may be 10-bit for input and internal processing, but only output 8-bit.

The WJ-MX50A sold new from something over $4000 back in the late 1990s. THe MX50 adn MX50A date to the 1990 and they may well suffer from the SMD capacitor problem that plagued the AG-1980s from that era. If the mixer does not provide an improvement in the video I would not use it in the signal path.