What causes ghosts of previous frames on a VHS capture?
 

These images are from an uncompressed raw capture originally made with VirutalDub 1.9.11 so what can be causing the echoes of previous frames from the rotating pyramid to be appearing in them (and what would an artifact like this normally be called)?

It shouldn't be caused by the interlacing (and i've left the last two frames in the sequence so you can see what that looks like) so is it just a muddled codec processing badly or could cabling, settings or something else be doing this, because I'm at a loss :cry:

I'm using Windows 7 Professional 64-bit (which has been on this machine for more than seven years and is as current as windows update can make it, which I appreciate is the quickest way to make software stop working entirely under windows 10 and probably why I needed to use the USB2883's EMBDA64.inf hack to get the ATI working after multiple failures with official Vista drivers, discs and downloads :eek:).

I've tried multiple combinations of player (Panasonic DMR-BR630V, Victor's DHX1, DHX2 and HR-WR5), capture device (IO GV-USB, ATI TV Wonder 600HD USB, Magewell USB Capture AIO and BlackMagic DeckLink PCie card) with and without TBC (BrightEye 75) connected.

The GV-USB managed to (occasionally) just have one ghost frame as did the ATI, while the Magewell and the BE75 combination sometimes made as many as three ghostly back-images per frame as shown in the example.

Any help would be gratefully appreciated :)

3D DNR from player or other device (though don't think any of those capture cards would have that.) could be one cause, which all of those have afaik? I have seen slight luma ghosting on my JVC HR-S8600EK even with TBC turned off and EDIT picture setting. Not too familiar with the japanese vcrs but could be they also don't entirely let you disable 3D DNR, it will at least be likely active with default settings. Panasonic dvd-recorders, whether standalone or combos, at least the european ones, also tend to have 3D DNR enabled by default but can be turned off.

If you want to isolate vcr and see if the capture devices or TBC do it on their own I would test with a more basic vcr that lacks any digital processing.

Does it only happen on this tape, or on other tapes too?

Thanks for the quick reply. :)

I don't have time to do extra tests at the moment but it's very helpful knowing that it could be a simple processing "error" (I'm sure someone at Victor/Panasonic headquarters would claim it's working as intended :mad4:) on the VCR side.

Unfortunately, it is doing that on all the tapes so I'll experiment some more with the settings and see about re-activating my old clunker drive.

I see this ghosting also on my Panasonic combo Sony tv setup, most of the time with recordings that where made at long play recording mode, it's correcting stuff that kicks in, maybe another video output does not have this, component (YUV) doesn't have it with me.

I've still not changed any settings (so busy :mad4:) on my Japanese equipment but I have done a couple of experiments.
I can confirm that even the s-video passthrough on a DCR-TRV30 into WinDV via Firewire also has the problem from the NTSC-J decks but the limited test (as it now appears the machine is dying - the LCD reduces brightness on motor engage and sometimes won't even initiate playback :depressed:) on my composite only NTSC 4:43 capable PAL deck into ATI 600 USB did NOT appear to have the problem.

I may have to search for an old Japanese device in good condition (is there a "yeah, right(!)" emoticon?) but I've also been looking through English manuals for adjacent models and have discovered than Panasonic definitely know it's something that exists and advises turning the 3D and "intedgrated" NR off (as shown in the attached pdf grab), I just hope that's possible in the Japanese version in VHS mode (and before anyone suggests using US decks, I would really prefer to use equipment that definitely knows what the IRE levels should be)

I may have to trick the machine by outputting through a component/hdmi situation, because for this particular set of captures (including first edition Pokemon pre-"Porygon Shock" tapes) frame accuracy is more important than picture quality or colour range.

My guess is the ghosting images happen with tapes that use long play recording, but is not "worked on" via component output, this is the case with my ES35V combo.
Sometimes you need to double check output, because the content can be outputted differently on each output but this not a rule in all cases.
Only, does the DMR-BR630V output the VHS also over the RCA component output ? i guess this Japan only device has also HDMI ? so a HDMI passthrough device would also be an option, because external component to HDMI conversion can have side effects, like to bright or to high luma levels, like cheap converters tend to do.
btw the DMR-BR630V will also have IRE 0 ….where USA NTSC has 7.5 ..so don't use the DMR-BR630V for capturing USA NTSC tapes, but …. i don't know if IRE has effect on the RCA component YUV output ? HDMI could be a different story… maybe just try all the different setups.
Also … is the IRE value stored on VHS ? or only active at playback ?

It could be caused by a bad telecine process or video format conversion (NTSC->PAL/PAL->NTSC or even NTSC->PAL->NTSC). Neither was very noticeable when playing back on a CRT so I think people at the studios just got sloppy sometimes.

These are all from original Japanese SP rental tapes (although it wouldn't surprise me if LP/EP playback would exaggerate the issue). I've yet to try different outputs for internal processing changes.

That would be a very interesting thing to find out as it could increase the range of playback devices without actually damaging the picture intention - especially as I have European machines with NTSC capability (which would presumably be IRE 0 as that's the PAL norm).

Fortunately this is studio NTSC to retail NTSC so that isn't a problem and I'm pretty sure I've confirmed it as a noise reduction process as I made a fresh capture with recently reactivated Panasonic DMR-EZ48VEB in NTSC mode and it doesn't exhibit the issue (although I may need to change some more capacitors as playback does have other problems :depressed:)

yugi 1,2,3.jpg shows the recent EZ48/ATI capture
yugiB 1,2,3 shows the same frames from a DHX2 with the "ghosting" pyramid edges from the previous frame.

For comparison, the ghosting on a scene change (second last image is previous frame) my JVC HR-S8600E_ (appears even if TBC is off and with EDIT picture setting.) With the default settings there is some chroma afterimage as well (I think TBC was on but not 100% sure) The DNR on the Panasonic NV-HS870 is even worse. The HR-S6900E_ only has at most 2D DNR but might be off in EDIT mode and Panasonic NV-HD620 only has analog 1-line NR, vhs-decode from the same HD620 has none, so no ghosting (other than possibly some afterimage that seems to be on the tape itself) is visible). (All other than vhs-decode went via a DMR-ES30, aka ES10+VCR for stabilization, NR on that was off.)

I'm really liking that VHS-Decode picture. :) Maybe I should get some more money out and play with some test probes :D

I'm wondering how useful chip numbers (from service manuals, maybe) could be in targeting machines with the right type of processors because these wildly different results are reminding me of the green vs black AVT TBC situation.
If these are fixed chips (rather than FPGA) maybe it could even be possible to switch out with different versions or cut a pin here and there to forcibly deactivate an undesired function.

What's odd is how the vhs-decode is actually softer than some of the other samples.

The green vs. black situation was partially due to on-chip software, and it was bitlocked. You'll run into a lot of that on AV gear.

I've not done much with ghosting for years now, but oddly I sometimes found certain 90s VHS VCRs to be "better" (less ghosting, but other issues), and always found TMPGEnc Plus filters to work best for some projects. For enjoyment sake, some ghosting could be hidden by Avisynth added noise, even MPEG encode mosquito noise.

? IRE 0 would be only NTSC-J a good european machine does only PAL - IRE 0 and NTSC IRE 7.5
The fault would be made by an consumer converter/capture device.
When i'm using a Blackmagic Design Analog to SDI Mini converter, i can set the correct values for the IRE level,
and the recording's colors are better represented then done by a China converter, which has no option to set IRE, also, keep in mind, for which the IRE value is valid: for the input or the output, while setting "it" somewhere else it could be switched back. (by baked in setting or software) and yes, like Traal says, even the studio "pro's" could already have made that mistake for you.
So that degradation is only limited repairable ?

FYI, most items have some IRE offset, it's never exact. Especially DVD recorders, and anything pure consumer.

Is there any way to "detect" which value is used ? it seems almost impossible to have this correct….

Further, does an NTSC VCR crush the shadows or highlights in the signal from an NTSC-J tape that cannot be recovered with a proc amp before the capture card?

And…. i only see IRE mentioned with composite video… what if one captures from the RCA component output ? would this make a difference ? there is no composite coding this way….