VHS Family Wedding tape restoration

[sanlyn]

[Many people report that decent VCRs (without TBCs) looped through an ES10 or ES15 give as good results. Other people are convinced the ES10 or ES15 will ruin the image. I haven't used mine yet.

I used mine. It replaced one I had years back that developed a bad drive, but I didn't use that one for capture back then (I had TBC VCR's at the time). it certainly didn't ruin anything. But like everything from Panasonic, you have to watch out for a little red pump, be careful with contrast levels, and turn off the DNR. It did quite well keeping lines straight and the other stuff that a line-TBC is noted for. On really rotten tapes it was a bit more effective than my Toshiba machines, although I'd say both are about par with the more "usual" VHS timing problems. Later Panasonics didn't match the ES15 as far as tbc heft, although they're better than none at all. Too bad, because later models are more numerous it seems.

[manono]

A comment was made previously that the SONY codec (default on my PC) used when I captured with WINDV was not a good choice ....
could someone advise what codec I should install ?

Cedocida DV Codec

[DB83]

A comment was made previously that the SONY codec (default on my PC) used when I captured with WINDV was not a good choice ....
could someone advise what codec I should install ?

Cedocida DV Codec

Memo. I really must have that 2nd Shredded Wheat for brekkie 'morrow. Then I might just understand how an installed codec can override a hardware (ADVC) stream which is what we are talking about here.

[Brad]

It is used for decoding said stream.

[DB83]

It is used for decoding said stream.

Ok. But one more silly question before I 'turn in' and dream of 'morrow's brekkie.

Surely, the 'damage', for want of a better word, has already been done by the encoder. The decoder can only report on what it finds.

[Brad]

Certain DV decoders introduce additional errors; in particular poor quality chroma upsampling.

[sanlyn]

Capturing old damaged stained VHS to DV involves chroma issues, is lossy encoding regardless of bitrate, and requires lossy re-encoding if you want playback as anything other than DV. May as well stop wasting time and record it directly to MPEG.

I would repeat that, but......

[DB83]

Certain DV decoders introduce additional errors; in particular poor quality chroma upsampling.

Thnx. (Stating the obvious but a really crap night I had) Methinks the OP should really be checking what DV codec is installed rather than what is being recorded. Just checked my Device Manager and I have the Canopus ones. But since (I know) he has a version of Vegas installed I suspect that decoding really is being done by the Sony codec.

[Tafflad]

When my new ADVC arrives, I'll try a capture then add Cedocida codec see if mediainfo changes.

[DB83]

It won't

You are simply confusing the recorded signal - possibly mis-quoted by mediainfo - and the decoded signal which your installed codec will use.

BTW what does Device Manager (or Win7 equivalent) tell you about installed codecs (of course if you have now installed the suggested codec all bets are off as to whether you were using the Sony codec for decoding)

[Tafflad]

No ... not installed new codec yet ... on my PC I have:
video codec
ICM Microsoft RLE MRLE
ICM Microsoft Video 1
ICM Microsoft YUV UYVY
ICM Intel IYUV codec
ICM Toshiba YUV Codec
ICM Cinepak Codec by Radius
ICM Lagarith Lossless Codec
DMO Mpeg4s Decoder DMO
DMO WMV Screen decoder DMO
DMO WMVideo Decoder DMO
DMO Mpeg43 Decoder DMO
DMO Mpeg4 Decoder DMO

Video Filters
video Microsoft DTV-DVD Video Decoder
audio ffdshow Audio Decoder ffdshow.ax
audio Microsoft DTV-DVD Audio Decoder

[2Bdecided]

A comment was made previously that the SONY codec (default on my PC) used when I captured with WINDV was not a good choice ....

To be 100% clear: the DV codec is not used when capturing DV. Capturing DV is a file transfer process (albeit a real-time one). The DV codec is only used when playing back what you have captured.

Cheers,
David.

[2Bdecided]

Capturing old damaged stained VHS to DV involves chroma issues, is lossy encoding regardless of bitrate, and requires lossy re-encoding if you want playback as anything other than DV. May as well stop wasting time and record it directly to MPEG.

I would repeat that, but......

No need to repeat it Sanlyn, let's take it to another thread where you can show me what I'm missing...
https://forum.videohelp.com/threads/360804-DV-vs-lossless-capture-of-VHS


Cheers,
David.

[sanlyn]

I don't have any DV source or gear. How do you "capture" analog VHS to DV? Obviously it doesn't work the way DV->DV works, because DV->DV isn't captured, it's copied.

[2Bdecided]

I don't have any DV source or gear. How do you "capture" analog VHS to DV? Obviously it doesn't work the way DV->DV works, because DV->DV isn't captured, it's copied.

I still suggest we take this to the other thread, where I've asked for a lossless capture to play with.


https://forum.videohelp.com/threads/360804-DV-vs-lossless-capture-of-VHS


To answer your question (but PLEASE let's take this to the other thread!) at its simplest in the ADVC110 there is an analogue to digital converter and a Hardware DV encoder. The DV format is fed to a firewire output, just like on a DV camcorder. This is then "captured" in the same way as any other DV in the PC - Firewire card and WinDV in my case.


Some DV camcorders with analogue inputs include a TBC and some NR between the ADC and the DV encoder. The ADVC300 reportedly includes something like a line-TBC and some proc-amp controls (before the ADC). The ADVC110 includes neither, so you need a stable signal or an external TBC, and correct levels or an external proc-amp (or near-enough levels that you can fix them in software without causing significant damage).


Do you have a lossless capture and maybe an AVIsynth script that you could contribute to the other thread sanlyn?


Cheers,
David.

[sanlyn]

I have plenty of lossless captures to AVI lying around, but no DV captures to compare them with. You would need a common source that exists as both equal quality VHS and DV source, or at least a VHS source, to capture both ways. That VHS source should be typical: typical noise and other problems common to the medium, etc. Such a source might be a test pattern on VHS. Of course, no one watches test patterns, and test patterns that I've seen don't move like a bride walking across the frame.

[2Bdecided]

sanlyn, we've derailed this thread long enough. The answers are in the other one.



It'll be interesting to see what difference, if any, re-capturing without the TBC in-line makes.


Cheers,
David.

[Tafflad]

...
Did you capture any recent-ish broadcast source (e.g. a VHS tape you recorded from BBC One in the 1990s) using the same kit Tafflad? A sample of that will show whether there's anything fundamentally wrong with the capture set-up.
...
Cheers,
David.

.

Found such a tape with an off-air recording - as requested here is a section from a BBC play
Have now got hold of another ADVC device (could not find a suitable capture card)

VCR >SVideo>ADVC110>firewire>PC (captured as type 2 using WINDV)

Short sample saved as Lagarith YV12
This has had no correction or filtering applied.
The sharp/soft setting on VCR as before left at mid point.
PAL VHS

[2Bdecided]

The luma levels are a little high, but not so high as to be unrecoverable in software. Something like levels(0,1.0,255,4,238,coring=false). As a matter of interest, I get "perfect" luma levels from broadcast recordings using a Panasonic VCR. I don't suppose it matters.
Other than that, it looks fine to me. I don't think a TBC would help that source at all (other tapes may be difference). The "look" of the edges of things is exactly what my Panasonic VCRs look like, and to my eyes it's fine but others may complain.


There's nothing fundamentally wrong with the capture IMO. It's also 1000x better than what many people bring here for help!


For future reference: from a DV source, you can cut losslessly in VirtualDUB by selecting "direct stream copy" on the video menu before saving your edit. No need to decode to Lagarith, which makes it 5x larger for no benefit.



Cheers,
David.

[Tafflad]

OK ..that is good news, shows VCR not doing anything not expected. (which was a possibility)

I'll redo the sample being compared earlier in thread (post #1 sample 2).... so it can be compared without TBC in the path (suggestion it could be softening capture)

I'll with and without DVR in path.

[sanlyn]

The BBC sample does look better than the godawful tapes submitted earlier. However...

There's nothing fundamentally wrong with the capture IMO.

IMO, there is plenty that's wrong with it. Typical results after cleanup of VHS->DV are easily apparent, especially in clay face effects and the impossibility of removing edge ghosts. There is moire and object shifting, evident in the shots where the books in the shelves are shimmering and sparkling. I also noted that sample.avi is progressive. Was it captured as progressive, or deinterlaced later? I'd say it's a pretty good example of the headaches created by VHS->DV capture, especially with budget devices. But you folks suit yourselves. I grew very tired of coping with the challenges of VHS->DV in this clip very early in the game. I'd also mention that the chroma levels exceed RGB 0-255. Facial highlights and white shirts are "hot spots" on playback.

[2Bdecided]

Oh sanlyn, Happy New Year! How I've missed you.

It would be a 25p-in-50i broadcast, as all dramas are in "PAL" countries.


I'm guessing the clip is from this...
http://en.wikipedia.org/wiki/The_Falklands_Play
...so it would be a digital 16x9 720x576 FHA master, letterboxed into the equivalent of a 4x3 704x576 frame for the analogue output of a digital STB if it was taped from the original showing on a digital-only channel. That conversion would introduce a little moire and shimmer as the camera moved.

I assume "motion smearing" is intentional - they don't tend to shoot 25fps content with 1/500th shutter speed.

But you folks suit yourselves.

That's the thing though - I don't think Tafflad is wedded to any approach. He just wants a decent capture of his video using some equipment that he can lay his hands on. I don't think you claiming that every single fault (from the "interesting" tube camera used to shoot his wedding video, to the BBC's decision to shoot drama progressively with a 1/50th shutter) is somehow down to DV capture, is very helpful.

I'm sure, if he can make it work, he'll do equally well with lossless capture. I'm also reasonably sure that he won't do visibly better either.


btw, remember the other thread? Remember vaporeon800 found a sample where he could see a difference between DV and lossless, if he blew both up to 200%, and switched back and forth on a single frame, and looked really really carefully at the raw capture?
https://forum.videohelp.com/threads/360804-DV-vs-lossless-capture-of-VHS?p=2288304&view...=1#post2288304
That's the kind of "issue" you get with DV. Not the ones you're imagining.


Anyway, what card do you capture with sanlyn? If you were in a PAL country Tafflad could have posted his BBC drama tape to you to compare.

Cheers,
David.

[sanlyn]

Thanks for the comments and the background info. If you have to blow up something 200% to see problems, you'll watch anything. UTube thrives on viewers who have what we know down South as CSC syndrome (Can't See Crap). A slight correction: Motion smearing and slow-shutter motion blur aren't the same thing, nor do they look alike.

As I say, suit yourself. Not everyone views videos on a 19" TV. Anyway, this sample leaves me convinced that I would avoid VHS->DV, especially when gear is available today to capture otherwise. Some might call the sample "decent", but I see it as annoying. This particular video might not be all that important, but I'd think that wedding and family videos would be. To each his own, which is what this is all about.

The attached is what I could get before getting tired of it. And for what it's worth:

Import("D:\AVisynth 2.5\plugins\HQdering.avs")
Import("D:\AVisynth 2.5\plugins\ReplaceFramesMC2.avs")

AviSource("E:\tafflad3\G\Sample.avi")

Crop(6,44,-10,-40)
ColorYUV(cont_y=-35,off_y=-9,gamma_y=-15)
HQDering()
MergeChroma(aWarpSharp2(depth=30))
ChromaShift(L=-2)
DeHalo_Alpha()
#--- Efforts to avoid clayface and posterization,
#--- which is minor but very visible.
Dfttest(sigma=12)
DeBlock_QED()
GradFun2DBmod()

# ---- repair broken lines + edges that seem to
# ---- have been imbedded in the BBC broadcast,
# ---- according to comments made in this thread.
# ---- Unfortunately it kills some fine detail, but it
# ---- seems preferable to sputtery DV artifacts.
w = width
h = height
nnedi3_rpow2(opt=2,rfactor=2,cshift="spline36resize").TurnLeft().\
  NNEDI3().TurnRight().NNEDI3().spline36resize(w,h)
AddGrainC(1.5, 1.5)
# ---- repair black stripes in two frames. 
ReplaceFramesMC2(314,2)

#--- YV12->YUY2->RGB is for VirtualDub processing.
#--- (Used NeatVideo turned down to about 40%)
#--- Saved out of VirtualDub as Lagarith YV12 for encoding
ConvertToYUY2(interlaced=false)
ConvertToRGB32(matrix="Rec601",interlaced=false)

#--- restore frame size to 704x576 from original 720x576
AddBorders(0,42,0,42)

[2Bdecided]

The image aspect ratio looks more like the 1.55:1 or 1.6:1 used in European film and early VistaVision, but I'll have to take your word that it was broadcast as letterboxed 16:9 in a PAL SD frame.

I haven't checked, but the "lit up" area of the 4x3 capture will be 14x9. That's how all 16x9 TV programmes (other than films) were played out via analogue from 1998-2012 in the UK - and most (not all) digital STBs converting native 16x9 broadcasts for display on 4x3 TVs would do the same (but not always that well).

If you have to blow up something 200% to see problems, you'll watch anything. UTube thrives on viewers who have what we know down South as CSC syndrome (Can't See Crap). A slight correction: Motion smearing and slow-shutter motion blur aren't the same thing, nor do they look alike. The image aspect ratio looks more like the 1.55:1 or 1.6:1 used in European film and early VistaVision, but I'll have to take your word that it was broadcast as letterboxed 16:9 in a PAL SD frame.

As I say, suit yourself. Not everyone views videos on a 19" TV.

I think you've misunderstood the point, but I'll leave it. While we disagree about what DV does to video, I think we've both learned useful things about restoring videos in PCs from each other, and hope that will continue.


Cheers,
David.

[sanlyn]

I think you've misunderstood the point, but I'll leave it. While we disagree about what DV does to video, I think we've both learned useful things about restoring videos in PCs from each other, and hope that will continue.

I was just about to edit my previous remarks to say the same thing. It's a deal. And I'd add that I've certainly picked up some great restoral tips from you as well. If it hadn't been for your input on that godawful opera that you might recall from a while back, I would never got back into Avisynth to begin with. Thanks again.

[Tafflad]

Hi Guys ...suggested earlier in thread was that my using a TBC 'may' have degraded the capture, by softening too much.
I have gone and purchased another ADVC unit ....

As previously you focussed on Sample 2, I have recaptured that section again .... this time with no TBC and also via a DVDR.


So ... capture path for Sample 5 is VCR <svideo> ADVC <firewire> PC

For Sample 6 is VCR <svideo> DVDR <svideo> ADVC <firewire> PC

and we have original sample 2 (post#1) which was with Full Frame TBC
capture path for Sample 2 was VCR <svideo>TBC <svideo> ADVC <firewire> PC


VCR settings same on each with picture control set to 'mid point', edit 'off'
Capture is with no processor control adjustments, other than setting to PAL.

Would really appreciate comments as to what the difference is ..... and is it significant enough to redo captures compareed to what I have with 'sample 2'

[2Bdecided]

The levels are so different it's hard to compare them directly. They also change throughout the clips relative to each other. This makes visual comparison very tricky.

I used this script to try to make comparison easier. It does this:
1. align the clips frame-by-frame, including correcting for an extra frame near the scene change in the via-DVR version
2. match the levels and saturation as best as I could (matching in the middle of the clip - start and end remain further off)
3. bob them (you can delete this part if you want to see interlaced frames)
4. label them (so you know which is which)
5. interleave all three to make the comparison easier

#load sources for comparison
original=avisource("wedding sample 2.avi")
recap=avisource("Sample 5 straight.avi")
recapdvdr=avisource("Sample 6 via DVDR.avi")




#find same frames in all, try to match levels and colours
original=original.Tweak(sat=0.87,coring=false)
recap=recap.Trim(112,699).Levels(0,1.0,255,1,227,coring=false)
recapdvdr=recapdvdr.Trim(112,474)+recapdvdr.Trim(476,700)
recapdvdr=recapdvdr.Levels(0,1.0,251,0,255,coring=false).Crop(2,0,-0,-0).AddBorders(0,0,2,0)




#bob all
original=original.bob()
recap=recap.bob()
recapdvdr=recapdvdr.bob()




#subtitle (label) all
original=original.Subtitle("original", align=3).ScriptClip("subtitle(string(current_frame),align=1)")
recap=recap.Subtitle("re-capture", align=3).ScriptClip("subtitle(string(current_frame),align=1)")
recapdvdr=recapdvdr.Subtitle("re-capture via DVD-R", align=3).ScriptClip("subtitle(string(current_frame),align=1)")




#interlave all
interleave(original,recap,recapdvdr)

What I'm seeing is...
1. your original capture appears to include extra sharpening and maybe a bit of extra contrast which I haven't removed
2. the direct re-capture is very bright. I've made it match the others but I'd be a little worried that some details might just be irrecoverably clipped at some point.
3. the "via DVD-R" capture has the very right hand side of the image missing, but many timebase errors (wobbles in the picture) corrected. I think it also denoises/softens a little.



I haven't attempted to improve any of these at all, just match them to each other to allow a fair comparison.


I don't like the sharpening in 1, especially in this case because it will make any subsequent anti-aliasing much harder
I'm a bit uncomfortable with the picture cropping and possible softening in 3, but the TBC in the DVD-R is working its magic and the levels are tamed.
I'm a bit uncomfortable that the levels drift slightly between all three clips. Probably some AGCs in there.


Anyway, drop the above script into VirtualDUB, jump to the middle of the clip, then step through frame-by-frame and see what you think.


Cheers,
David.

[Tafflad]

Hi ... appreciate the assistance .... in another thread you have replied to ..
https://forum.videohelp.com/threads/360829-Panasonic-VCR-Settings?p=2293472&viewfull=1#post2293472

There have been comments that having 'edit on' makes a difference .........

I am happy to repeat this sample with 'edit on' .... if required ? just let me know.
BTW ... on the original 'sample 2' it was via Full Frame TBC and an ADVC300 ....... this time I am using an ADVC110 (default other than set to PAL)

Also happy to have slider set to 'sharp' as per your previous comment of I redo sample.

What I am trying to get to is the optimum settings for kit I have ..... and if necessary I'll then recapture whole tape, of all the VHS tapes to capture - the Wedding one is the most important.

[sanlyn]

2B, that's another cool bag of tricks I'll be keeping in my 2BDecided folder -- which is getting to be pretty full. Thanks for sharing.

[Tafflad]

HI Sanlyn ... as it was your original observations that have started me trying recaptures ... do you want me to carry out what I proposed in post #118 or any other mix & match combination.
2Bdecided.. obviously same Q to yourself