Faulty usb device or cure with a TBC

[DB83]

Some time back I acquired a Hauppauge USB2-Live. It was bought for the handful of NTSC tapes that I own (and not available on dvd etc.). I also have a ADVC 300 which have used for many years without issue. I use that for all PAL work.

Never got around to using the usb2-live until yesterday when I was trying out some software in another topic. This is the issue I find with it.

The pictures are captured both PAL and PAL-60 with a 'lined' pattern which is most noticeable on skin-tones but is also present on bright colors. I have no TBC attached to the usb and there is also some horizontal shift which may be related to the 'lines'

However capturing the same tape, using the same compostite leads (s-video is not an option), through the ADVC, which has a rudimentary TBC, the picture is perfect.

So I do wonder id the usb is faulty or simply not up to the required job. Yet others have claimed perfect results. Or will I simply solve the issue with a line-TBC ?

If the latter, can anyone recommend what I should look out for.

Thanks.

[hech54]

This is one of my PAL60 captures(sample - 11MB) with the Hauppauge, WinTV 7.0.30312(CD 2.6C) and a Philips VR-1100(JVC clone).
I don't pretend to know what or if anything is wrong with this capture....looks fine to me.


OOPS.....forgot the link.

https://mega.nz/#!NEYzSTra!lWZDNx2UE3HHUdJOEk8qxNOO_ddeJ-bPfF_ntI9n6Cg

[DB83]

Sample missing ?

[jagabo]

The pictures are captured both PAL and PAL-60 with a 'lined' pattern which is most noticeable on skin-tones but is also present on bright colors.

Post a sample. But it sounds like dot crawl artifacts -- incomplete separation of the chroma subcarrier from the luma. With VHS you can nearly eliminate this by downscaling to ~half width, then upscaling back to 720. It doesn't hurt the sharpness much because VHS only has about half the resolution of a 720 pixel wide cap.

There are also some dedicated dot crawl removal filters for AviSynth. Checkmate, for example. But that one only works with with NTSC.

[DB83]

I would post a sample of a PAL-60 capture. Got it to work once but now the s/w either says 'No Signal' or transmits the PAL from the connected sat receiver.

If I can not fix this then I will resort to a PAL sample.

[DB83]

PAL sample:

Not as obvious as it was in the PAL-60 I viewed earlier (still refuses to work which was the whole point of the purchase)

[DB83]

A couple of the PAL-60 captures which I have now found a way to work.

(BTW I managed that by telling the s/w that the source was NTSC. Obvious B/W result but then it properly switched to PAL-60. Could not go direct from PAL to PAL-60)

[jagabo]

Yes, those are dot crawl artifacts. The more saturated the colors the stronger the dot crawl will be. The following will almost completely eliminate them:

Spline36Resize(width/2,height).Spline36Resize(width,height)

That cap has many other problems that you need to address too.

[DB83]

^^ I'm sure there are. But I was merely looking for a diagnosis of something that I have been spared in previous capture devices.

Certainly my old PCI-based hauppauge card created 'clean' captures but that will not work under Win7.

BTW I am worse that a noob when it comes to avisynth. Can these artifacts be minimised through an NLE or vdub filters.

Did you also check the PAL-60 caps ?. To me they are worse but, like you said, the colors are stonger.

[hech54]

Posted my sample.....silly me.

[jagabo]

But I was merely looking for a diagnosis of something that I have been spared in previous capture devices.

How much dot crawl is left after separating luma and chroma from a composite signal depends on how good the comb filter is. If you're feeding composite to the capture card it's the capture cards comb filter that's in question.

Certainly my old PCI-based hauppauge card created 'clean' captures but that will not work under Win7.

Set up a dual boot system. Or a use a separate computer for capture.

BTW I am worse that a noob when it comes to avisynth. Can these artifacts be minimised through an NLE or vdub filters.

Pretty much all editors have resizing filters. In VirtualDub you could use two instances of the resize filter.


[Attachment 39565 - Click to enlarge]

Did you also check the PAL-60 caps ?. To me they are worse but, like you said, the colors are stonger.

Same thing.

[usually_quiet]

@DB83 The hardware used in the Hauppauge USB2-Live 2 is supposed to include a 2D comb filter to mitigate dot-crawl artifacts, but I've seen some reports like yours indicating that the 2D comb filter doesn't work.

3D comb filters are supposed to be better, but there don't seem to be many SD analog capture devices with a 3D comb filter that are still in production. I did some research for somebody in another thread capturing from Laserdisc and found out that the Hauppauge WinTV HVR 5525 is an internal TV tuner/analog capture device for the PC that has a 3D comb filter for SD analog capture. According to Hauppauge's German product page, it uses software to encode analog video, so it should be possible to do lossless capture. What I don't know is if it can capture PAL60 or not.

[usually_quiet]

Posted my sample.....silly me.

It didn't work because I'm still not seeing a link in any of your posts.

[Edit]Now I see it. All is well.

[jagabo]

This is one of my PAL60 captures(sample - 11MB) with the Hauppauge, WinTV 7.0.30312(CD 2.6C) and a Philips VR-1100(JVC clone).
I don't pretend to know what or if anything is wrong with this capture....looks fine to me.


OOPS.....forgot the link.

https://mega.nz/#!NEYzSTra!lWZDNx2UE3HHUdJOEk8qxNOO_ddeJ-bPfF_ntI9n6Cg

Captured via an s-video cable? That won't have dot crawl artifacts as luma and chroma are sent separately.

[DB83]

@Jagabo

There is clearly improvement in the vdub screen grabs as above.

Now I thought that plain ol' vdub only imported avi. So how can this handle a mpeg2 .ts ? I have not tried it (unless this is vdubmod )

[DB83]

^^ So as expected I get an 'unsupported file type' error message when I attempt to load the .ts in to vdub (same version).

Some plug in ?

[hech54]

This is one of my PAL60 captures(sample - 11MB) with the Hauppauge, WinTV 7.0.30312(CD 2.6C) and a Philips VR-1100(JVC clone).
I don't pretend to know what or if anything is wrong with this capture....looks fine to me.


OOPS.....forgot the link.

https://mega.nz/#!NEYzSTra!lWZDNx2UE3HHUdJOEk8qxNOO_ddeJ-bPfF_ntI9n6Cg

Captured via an s-video cable? That won't have dot crawl artifacts as luma and chroma are sent separately.

That is my normal procedure.

[jagabo]

Now I thought that plain ol' vdub only imported avi. So how can this handle a mpeg2 .ts ?

Get the ffmpeg source plugin for VirtualDub. Put it in Virtualdub's plugins folder.

But VirtualDub isn't great for interlaced YV12 video. It will blur the chroma channels together. The problem will be noticeable in colorful high action scenes.

[DB83]

Now I thought that plain ol' vdub only imported avi. So how can this handle a mpeg2 .ts ?

Get the ffmpeg source plugin for VirtualDub. Put it in Virtualdub's plugins folder.

But VirtualDub isn't great for interlaced YV12 video. It will blur the chroma channels together. The problem will be noticeable in colorful high action scenes.

Yeah. I did read the speel on the vd page and found the appropriate plug-in.

Done a test and it does, as you say, remove some but not all of the artifacts. In fact some left-pointed ones become right-pointed ones

Might try some other editors now I have the basic idea

[jagabo]

You can completely avoid the problem by capturing an s-video signal instead of composite, like hech54 is doing. That generally means you need an expensive S-VHS deck though.

[DB83]

I know. With only a handful of NTSC tapes under my radar that is not really cost-effective.

But I just had a brain-storm. Since the ADVC gives quite good results for pure PAL tapes I just did a capture using that as a pass-thro device. Now I do not need telling that the AR is wrong but that surely is correctable. And despite selecting a NTSC source it records at PAL speed (did another to check for sound sync during singing at all seems fine there. But what of the general picture quality ?

[jagabo]

Passing the PAL60 signal through the ADVC has reduced the dot crawl artifacts and stabilized the color and horizontal time base. Cropping away 96 lines from the bottom of the frame will get you back to the original NTSC frame and aspect ratio. But it has also lost some frames from the 29.97 fps to 25 fps conversion. That will lead to more jerky video. Even advanced AviSynth scipts may not be able to smooth it out.

A better alternative would be to find a true NTSC player for the tapes and capture from that.

The whole point of using a raw YUV capture device is to capture lossless. Capturing as MPEG 2 is losing detail and generating artifacts even before you get to editing/filtering the video. But you need to stabilize the video for your capture device to deal with it. For pure PAL or NTSC video the easiest way is to pass the signal through an old Panasonic DVD recorder (don't record to DVD) which has a line time base corrector and a frame sync. But I don't think they'll deal with PAL60 from your current VHS deck.

[hech54]

You can completely avoid the problem by capturing an s-video signal instead of composite, like hech54 is doing. That generally means you need an expensive S-VHS deck though.

He could also try the SCART to S-Video cable method BUT....that is a big, giant "if".
The machine needs to be fully wired to take advantage of what SCART is capable of......if not it won't help.

[DB83]

^^ With my current deck I doubt that will help. It is not a VCR but a combi. It has s-video out but that only applies for dvd playback so I doubt that the scart will also be wired accordingly. I might just try it some time all the same.

@jagabo. I did the full capture of 'Meatloaf meets Indiana Jones'. There is a lot of fast action in this vid and I could not detect any stuttering. Also sound kept in sync. I know it is not an ideal method but it sure does clean up the picture.

I can use other software which will allow me to capture lossless. I will probably do that just once with this tape as it is now showing signs of wear - forget the precise term but one sees occasional white streaks. I had them once on another tape and did succeed in filtering them out.

[jagabo]

@jagabo. I did the full capture of 'Meatloaf meets Indiana Jones'. There is a lot of fast action in this vid and I could not detect any stuttering.

It's easiest to see in medium speed panning shots. In the uploaded clips it's most obvious in the shot where they guy comes out of a cave holding up a mask. Watch the motion of the ladder in the background. You'll see several little jerks per second in its motion. The motion is much smoother in the PAL60 clip.

These are especially problematic clips because the NTSC tape was made from a PAL source with field blending. And they're from the opening credits where shots are often sped up or slowed down (by frame decimation or duplication, or field blending) to get the lengths they want. The body of the show is less likely to have been manipulated like that.

[DB83]

jag,

What makes you think that the NTSC tape was made from a PAL source ? AFAIK 'Meat' was signed to an American(or Canadian) Record Label and this video, like others before, were shot by American crews with American casts. This particular one was directed by Howard Greenhalgh - the other well-known ones by Michael Bay. As with many promo videos, several versions exist. The one that is on yt is almost identical but has a few minor cuts in the tent scene. I have also seen another version with a different scene late on in the cave.

As for 'credits' there are none. It's very rare to have credits on promos.

But I do not have 20/20 vision. I just looked at the full-length video and do not see 'jerks' you describe.

[jagabo]

As for 'credits' there are none. It's very rare to have credits on promos.

Credits, promos, whatever. The point is those types of clips tend to have shots sped up and slowed down for aesthetic reasons. Or running time reasons.

What makes you think that the NTSC tape was made from a PAL source ?

All the field blending in the PAL60 cap. Though it's possible it's from the speedup/slowdown above.

AFAIK 'Meat' was signed to an American(or Canadian) Record Label and this video, like others before, were shot by American crews with American casts.

It's quite common for NTSC video to made from PAL videos when an NTSC version isn't available for some reason. When this is done with video tape they don't bother to capture the video, inverse telecine back to film frames, then perform a frame rate conversion. They just run the tape through a realtime standards converter and capture the output. Those studio standards converters always use frame blending.

I just looked at the full-length video and do not see 'jerks' you describe.

Lucky you!