Various artifacts after VHS capture?

[hermetic_charm]

Introduction
After over a year of building my capture chain I have started running some test captures. I am at a point where I really need some help 1) Developing vocabulary for issues/artifacts during capture and 2) Understanding the source of the issue (e.g. tape damage, capture chain issue, something to be handled in post).

On point 1: I know there is a wealth of knowledge here already, but I have hit a hurdle in finding the right information without the right vocabulary.

On point 2: I want to learn which issues are unfixable (tape damage), which need to be fixed before capture (hardware or settings related issues in the chain), and which can be handled in post.

My end goal is to get quality "raw" captures of some family VHS tapes which will be archived, and then do some restoration and encoding to make them more accessible to other family members.

Setup
Here is my hardware setup:
JVC HR-S9800U>datavideo TBC-1000>ATI TV Wonder HD 600>Windows 7 SP1 machine (I7-2600k, 16 GB ram, AMD R7 370 GPU, disconnected from internet)

Spyder3 Pro for screen calibration

Software setup:
DisplayCal for screen calibration
graphstudio to adjust proc amp while viewing histogram before capture
VirtualDub 1.9.11 + LS Filter pack for capture
AviSynth for post

VCR Settings:
Video Calibration - OFF
Picture Control - NORM
Superimpose - OFF
Digital R3 - OFF
Video Stabilizer - OFF
Audio Monitor - HiFi
S-VHS Mode - ON

Digital TBC/NR is turned on

I used the guide here to setup VirtualDub: https://www.digitalfaq.com/forum/vid...-settings.html

Capture Issues
I have tested a handful of tapes (commercial and home recordings of TV broadcasts on some unknown consumer grade VCR from the 90s). I have not tested any family videos yet as I want to build my knowledge up first before moving on to those priceless videos. I selected the tape below to start with here as it seems like it has the most issues I have seen so far.

Attached are two clips from the same tape, I only opened in AviSynth to apply a trim to get the file size down to upload. This was a recording of Woodstock '99, it was recorded on a home VCR and if I recall correctly, it was a Pay-per-view event.

Proc amp settings:
Brightness = 124
Contrast = 26
Hue = 64
Saturation = 32
Sharpness = 2

I am sure a pro here will see more than I do but what I noticed was:

1. Strong distortion at the bottom of the screen, more so than what I think is expected in the overscan area. Is this tape damage?

2. The sides of the screen also seem worse than other tapes, with little hair-like lines sticking out

3. It seems like there are green/pink bands throughout, sometimes a much larger band will appear at the top of the screen. Is this a chroma issue? Oddly, this looked much worse on my Windows 7 machine. When I transferred the AVI files to my modern machine it was much less pronounced. I think it could possibly be a difference in the screen quality. I have seen this issue also in some commercial tapes, but only the wide band at the top of the screen and green instead of pink (not sure if this is some kind of copy protection).

Any help on identifying what these issues are called, their source (tape damage, hardware setup, fixable in post), and possible fixes would be greatly appreciated!

[themaster1]

Lots of chroma noise and for the bottom noise you can bet the original tape was recorded with misaligned tape guides (exit guide precisely) . Can be tweaked if there is the will.

I noticed a vertical jump aswell (frame49, 1st video), that's what the "video stabilizer" feature is for on the JVC.
Video decoded in RGB24 on my PC, so something's wrong here.

[hermetic_charm]

Oh man, that's my fault. In capture settings it was set to YUY2. First time saving a video loaded through avisynth script to AVI through VirtualDub and I think I figured out I needed to change some settings while saving in that mode. It seems like Video>Color Depth and Video>Compression are the important points here(?) when saving, not sure if anything else should be modified. But I changed compression to Huffyuv 2.1.1 and color depth to YUY2. This also makes the file size more manageable so I can upload longer clips in the future if neccessary.

Edit: In regard to the video stabilizer, I think that can only be turned on if the TBC/NR is turned off? I guess it's a judgment call then which one would be better for each tape?

Quote:

Can be tweaked if there is the will.

I wouldn't for this tape, but this is good to know if I find this issue with a more important tape.

Quote:

I noticed a vertical jump aswell (frame49...

Thanks for pointing this out and giving the frame #, I didn't even see it before but looking in VirtualDub frame by frame I found it, and learned what this is now and how to look for it!

[aramkolt]

Not claiming to be a pro here either, but here's my 2¢:

The very bottom area of the screen - say where the white bars on either side are black instead - is normal "head switching noise." These are typically masked or cropped later. They are kind of annoying in that it can technically mess with your aspect ratio if you are trying to get the final output to be exactly 4:3 as I believe they count towards the original vertical height. Masking avoids that issue, but leaves a black bar of unused space.

The area just above the head switching noise appears to be poor tracking to me. Usually you can manually adjust the tracking on the VCR by using the up/down or +/- buttons on the front of the VCR as it is playing, or in extreme cases you could adjust the tape guides, but I'd do that with a much less expensive VCR as there's no guarantee that it'll be right after you go back to playing normally aligned tapes.

If I had to guess, it's really hard to tell that that the line TBC is even on which suggests that this could be a second generation tape - meaning a recording of a recording where the second recording baked in any errors caused by the first machine playing it back which you then can't correct.

There also seems to be quite a bit of chroma noise present - that's the red/green rainbow stuff in horizontal lines that is particularly easy to see in the second clip.

There's a fairly significant vertical jitter moment about halfway through the first clip - Normally I'd say this could be due to lack of a full frame TBC, but the TBC1000 definitely is one of those and you have that listed as being in your chain. I question if you've fully recapped the TBC1000 or if you have bypassed the distribution amp? I haven't done much experimenting with mine, but around 80% of the caps on the main TBC board were VERY out of spec and I have heard that the distribution amp (the part that splits the output into 4) can add signal noise as well. LS or others could probably comment better if what you're seeing is typical with bad caps on a TBC1000 since I didn't really do any testing before the recap.

If I was you, I'd start with a known first-generation tape that is from a decent source (or even a commercial release) and make sure that looks good before moving to homemade tapes. I'd also invest in a DMR-ES10 for passthrough (with the line TBC off on the VCR and then put through the ES10 next) to see if that visually improves the image at all. With the line TBC off, you could also see if "video stabilizer" enabled helps too.

I will say after spending that much on the chain, I'd have personally hoped for a better looking result, but it could also just be that the source tape is that bad which is not the fault of the capture chain by any means.


Also, how did you come up with your proc amp settings? Typically those would just be left at defaults unless you have a test color-bar tape that you played to see roughly where to set the black levels and color levels using a vectorscope/waveform monitor. The tape levels likely will vary from tape to tape though, so you'll want to make sure you're not seeing whites "clipping" etc

[hermetic_charm]

Thanks for the feedback! I will see if I can adjust the tracking later today.

It is most likely a second-generation tape (or more). A lot of my family's tapes were recorded on and reused over the years (especially when recording movies and tv shows, hopefully the family videos aren't as bad).

The TBC-1000 has not been recapped or modded in any way. I will look up a guide later and see about testing them. I think it's worth noting though that most other tapes don't have issues like this, and I have also tested tapes where they had frames dropping and then adding the TBC 0 frames dropped.

A DMR-ES10 might be a good idea for the occasions I need to turn off the line TBC, I will start looking around. I've also seen some DVD players used for pass through can defeat Macrovision? I don't think this one can, but that's another tool I might add down the road if I ever decided to do the commercial tapes I have.

For the proc amp settings, everything was left at default except brightness and contrast. For brightness/contrast I played samples of the video with the VirtualDub histogram open, for bright and dark scenes, and adjusted until my levels on the histogram looked good (e.g. between 16-235), I followed a few posts on this here on the forum, but I am having trouble finding the links again.

I will try to get some more captures from other tapes later today and upload them for comparison, and also report back on the TBC capacitors.

-- merged --

I wanted to add an additional detail that might be important: these Woodstock tapes were recorded in ELP mode (~6 hour tapes!).

[themaster1]

Avisynth is your friend. Could take days if you have hours of it
https://imgsli.com/MjQ4ODMw

wood-qtgmc.jpg

You must be logged in to view this content; either login or register for the forum. The attached screen shots, before/after images, photos and graphics are created/posted for the benefit of site members. And you are invited to join our digital media community.

FixRipsP2_Woodstock99.avs

#LoadPlugin("C:\Program Files (x86)\AviSynth 2.6\plugins\KNLMeansCL-v1.1.1\KNLMeansCL.dll")
#LoadPlugin("C:\Program Files (x86)\AviSynth 2.6\plugins\degrainmedian082\degrainmedian.dll")
LoadPlugin("C:\Program Files (x86)\AviSynth 2.6\plugins\svpflow1.dll")
LoadPlugin("C:\Program Files (x86)\AviSynth 2.6\plugins\svpflow2.dll")
# ################ FixRipsP function, modified from lordsmurf median filter ###############
# ######################### for progressive video or SeparateFields() #####################

function FixRipsP2(clip a)
{
a
clense(reduceflicker=false).merge(last,0.5).clense(reduceflicker=false) # Clense is a very brutal temporal cleaner. 
mot=removegrain(8,0).DepanEstimate(range=2)
take2=a.depaninterleave(mot,prev=2,next=2,subpixel=2)
clean1=take2.selectevery(5,2)

sup1 = clean1
 \           .mt_lutxy(clean1,"x 1 + y < x 2 + x 1 - y > x 2 - y ? ?",U=2,V=2)
 \           .msuper(pel=2,sharp=0)
sup2 = a.msuper(pel=2,levels=1,sharp=2)

bv21=sup1.manalyse(isb=true, truemotion=true,global=true,delta=1,blksize=4,overlap=2,search=3,searchparam=5,DCT=2)
fv21=sup1.manalyse(isb=false,truemotion=true,global=true,delta=1,blksize=4,overlap=2,search=3,searchparam=5,DCT=2)
bv22=sup1.manalyse(isb=true, truemotion=true,global=true,delta=2,blksize=4,overlap=2,search=3,searchparam=5,DCT=2)
fv22=sup1.manalyse(isb=false,truemotion=true,global=true,delta=2,blksize=4,overlap=2,search=3,searchparam=5,DCT=2)

interleave(a.mcompensate(sup2,fv22),a.mcompensate(sup2,fv21),a,a.mcompensate(sup2,bv21),a.mcompensate(sup2,bv22))
#SpotLess(RadT=5,ThSAD=1000000,ThSAD2=1000000,pel=2,chroma=false,BlkSz=8,Olap=4,tm=false,glob=false,bBlur=0.0).selectevery(5,2) # proposé par le developpeur du filtre

#SpotLess(RadT=1,ThSAD=300000,ThSAD2=1500000,pel=2,chroma=false,BlkSz=4,Olap=2,tm=false,glob=false,bBlur=0.0).selectevery(5,2) # garde plus de détails (trait chemise)
#SpotLess(RadT=1,ThSAD=1000000,ThSAD2=1500000,pel=2,chroma=false,BlkSz=4,Olap=2,tm=false,glob=false,bBlur=0.0).selectevery(5,2)   # ça enlève moins de neige
#KNLMeansCL(d=3,a=2,s=4,h=3).SpotLess(RadT=2,ThSAD=500000,ThSAD2=499000,pel=2,chroma=false,BlkSz=4,Olap=2,tm=false,glob=false,bBlur=0.0).selectevery(5,2) # pour enlever moins de détails fins: BlkSz=4,Olap=2 et RADT=1 (si plus grand ça enlève plus mais ça créer des artéfacts vérifié X fois)
SpotLess(RadT=2,ThSAD=500000,ThSAD2=499000,pel=2,chroma=true,BlkSz=4,Olap=2,tm=false,glob=false,bBlur=0.0).selectevery(5,2) # pour enlever moins de détails fins: BlkSz=4,Olap=2 et RADT=1 (si plus grand ça enlève plus mais ça créer des artéfacts vérifié X fois)
#.DeGrainMedian(limitY=255,limitUV=5,mode=3,norow=true,interlaced=false)

sup3 = last.msuper(pel=2,sharp=2)
bv31=sup3.manalyse(isb=true, truemotion=true,global=true,delta=1,blksize=4,overlap=2,search=3,searchparam=5,DCT=2)
fv31=sup3.manalyse(isb=false,truemotion=true,global=true,delta=1,blksize=4,overlap=2,search=3,searchparam=5,DCT=2)
#bv32=sup3.manalyse(isb=true, truemotion=true,global=true,delta=2,blksize=4,overlap=2,search=3,searchparam=5,DCT=2)
#fv32=sup3.manalyse(isb=false,truemotion=true,global=true,delta=2,blksize=4,overlap=2,search=3,searchparam=5,DCT=2)
#bv33=sup3.manalyse(isb=true, truemotion=true,global=true,delta=3,blksize=4,overlap=2,search=3,searchparam=5,DCT=2)
#fv33=sup3.manalyse(isb=false,truemotion=true,global=true,delta=3,blksize=4,overlap=2,search=3,searchparam=5,DCT=2)

#last.mdegrain1(sup3,bv31,fv31,bv32,fv32,bv33,fv33,thSAD=499)
last.mdegrain1(sup3,bv31,fv31,thSAD=499)
#last.mdegrain1(sup3,bv31,fv31,thSAD=499)
interleave()
return last
}
######################

################
function MinBlur(clip clp, int r, int "uv")
{
uv   = default(uv,3)
uv2  = (uv==2) ? 1 : uv
rg4  = (uv==3) ? 4 : -1
rg11 = (uv==3) ? 11 : -1
rg20 = (uv==3) ? 20 : -1
medf = (uv==3) ? 1 : -200

RG11D = (r==0) ? mt_makediff(clp,clp.sbr(),U=uv2,V=uv2)
 \    : (r==1) ? mt_makediff(clp,clp.removegrain(11,rg11),U=uv2,V=uv2)
 \    : (r==2) ? mt_makediff(clp,clp.removegrain(11,rg11).removegrain(20,rg20),U=uv2,V=uv2)
 \    :          mt_makediff(clp,clp.removegrain(11,rg11).removegrain(20,rg20).removegrain(20,rg20),U=uv2,V=uv2)
RG4D  = (r<=1) ? mt_makediff(clp,clp.removegrain(4,rg4),U=uv2,V=uv2)
 \    : (r==2) ? mt_makediff(clp,clp.medianblur(2,2*medf,2*medf),U=uv2,V=uv2)
 \    :          mt_makediff(clp,clp.medianblur(3,3*medf,3*medf),U=uv2,V=uv2)
DD    = mt_lutxy(RG11D,RG4D,"x 128 - y 128 - * 0 < 128 x 128 - abs y 128 - abs < x y ? ?",U=uv2,V=uv2)
clp.mt_makediff(DD,U=uv,V=uv)
return(last)
}

Function SpotLess(clip c,int "RadT",int "ThSAD",int "ThSAD2",int "pel",bool "chroma", int "BlkSz",Int "Olap",bool "tm",Bool "glob",Float "bBlur") {
RadT = Default(RadT,1) # Temporal radius. (MCompensate arg)
ThSAD = Default(ThSAD,10000) # SAD threshold at radius 1 (Default Nearly OFF).
ThSAD2 = Default(ThSAD2,ThSAD) # SAD threshold at radius RadT.
Pel = Default(pel,2) # Default 2. 1, 2, or 4. Maybe set 1 for HD+. (1=precision to pixel, 2=precision to half pixel, 4=quarter pixel)
Chroma = Default(chroma,True) # MAnalyse chroma arg. If set to true, use chroma in block matching.
BlkSz = Default(BlkSz,8) # Default 8. MAnalyse BlkSize. Bigger blksz quicker and perhaps better, esp for HD clips. Maybe also better where BIG noise.
OLap = Default(OLap, BlkSz/2) # Default half of BlkSz.
Tm = Default(tm,True) # TrueMotion, Some folk swear MAnalyse(truemotion=false) is better.
Glob = Default(glob,Tm) # Default Tm, Allow set MAnalyse(global) independently of TrueMotion.
Bblur = Default(bblur,0.0) # Default OFF
Assert(1 <= RadT,"SpotLess: 1 <= RadT")
pad = max(BlkSz,8)
sup = (bBlur<=0.0 ? c : c.blur(bblur)).MSuper(hpad=pad,vpad=pad,pel=pel,sharp=2)
sup_rend = (bBlur<=0.0) ? sup : c.MSuper(hpad=pad,vpad=pad,pel=pel,sharp=2,levels=1) # Only 1 Level required where not MAnalyse-ing.
MultiVec = sup.MAnalyse(multi=true, delta=RadT,blksize=BlkSz,overlap=OLap,search=4,chroma=Chroma,truemotion=Tm,global=Glob)
c.MCompensate(sup_rend,MultiVec,tr=RadT,thSad=ThSAD,thSad2=ThSAD2)
MedianBlurTemporal(radiusY=0,radiusU=0,radiusV=0,temporalradius=RadT) # Temporal median blur only [not spatial]
SelectEvery(RadT*2+1,RadT) # Return middle frame
}

woodstock99-2-fixripsP2-progressif.avs

LoadPlugin("C:\Program Files (x86)\AviSynth 2.6\plugins\RgTools-v1.1\x86\RgTools.dll")
LoadPlugin("C:\Program Files (x86)\AviSynth 2.6\plugins\masktools2-v2.2.26\x86\masktools2.dll")
LoadPlugin("C:\Program Files (x86)\AviSynth 2.6\plugins\mvtools-2.7.45-with-depans20210608\x86\mvtools2.dll")
LoadPlugin("C:\Program Files (x86)\AviSynth 2.6\plugins\mvtools-2.7.45-with-depans20210608\x86\DePanEstimate.dll")
LoadPlugin("C:\Program Files (x86)\AviSynth 2.6\plugins\mvtools-2.7.45-with-depans20210608\x86\DePan.dll")
LoadPlugin("C:\Program Files (x86)\AviSynth 2.6\plugins\PlanarTools-0.3.0\PlanarTools.dll")
#LoadPlugin("C:\Program Files (x86)\AviSynth 2.6\plugins\Hdragc-1.8.7\AGC.dll")
#LoadPlugin("C:\Program Files (x86)\AviSynth 2.6\plugins\AutoAdjust-v2.60\x86\AutoAdjust.dll")
LoadPlugin("C:\Program Files (x86)\AviSynth 2.6\plugins\aWarpSharpMT\x86\Release_W7\aWarpsharpMT.dll")
#import("C:\Program Files (x86)\AviSynth 2.6\plugins\LimitedSharpenFaster.avs")
import("C:\Program Files (x86)\AviSynth 2.6\plugins\FixRipsP2_Woodstock99.avs")

SetMemoryMax(4000)
SetMTMode(3,0)
AVISource("C:\Users\gfk\Desktop\woodstock99-2.avi").convertToYV12(interlaced=true)
SetMTMode(2,4)

orig=last
ev=orig.assumetff().separatefields().selecteven()
od=orig.assumetff().separatefields().selectodd()
ev
ue_chroma = UToY(ev).blur(0,1.5).ttempsmooth(maxr=6,lthresh=150, strength=6)
ve_chroma = VToY(ev).blur(0,1.5).ttempsmooth(maxr=6,lthresh=150, strength=6)
YToUV(ue_chroma, ve_chroma)
MergeLuma(ev)
ev_filtered=last
od
uo_chroma = UToY(od).blur(0,1.5).ttempsmooth(maxr=6,lthresh=150, strength=6)
vo_chroma = VToY(od).blur(0,1.5).ttempsmooth(maxr=6,lthresh=150, strength=6)
YToUV(uo_chroma, vo_chroma)
MergeLuma(od)
od_filtered=last
interleave(ev_filtered,od_filtered)
assumefieldbased().assumetff().weave()

#####
QTGMC( Preset="fast", TR2=1, EdiThreads=4)  #)
selecteven()
FixRipsp2() # ON PROGRESSIVE FRAMES ONLY !
##########
##  TO REMOVE PINK/RED STRIPE  ON THE RIGHT SIDE
#ConvertToyv12(interlaced=false)
U = UtoY() # separate U channel
V = VtoY() # separate V channel
V = Crop(V,0,0,-16,-0) #remove discoloured portion and black border
extra=Crop(V,342,0,2,-0).PointResize(12,V.height)
V=StackHorizontal(V,extra).AddBorders(0,0,4,0,$808080) # add the extra, pad to 720 with grey
YtoUV(U, V, last) # mix the separate U and V with the original Y

#return(video_restored)
#####################
#convertToYV12(interlaced=true)

#converttorgb32(matrix="rec601",interlaced=false)

#############
#converttoyv12(matrix="rec601",interlaced=false)

#SetMTMode(2,1)
#input=last
# mode=0: mode contraste (parfait pour PPC)
#Y= AutoAdjust(auto_gain=true,gain_mode=0,avg_safety=1.00,scd_threshold=0,input_tv=false,output_tv=true,dark_limit=17.0,bright_limit=1.50,gamma_limit =1.0,dark_exclude=1.00,temporal_radius=1,high_quality=true,high_bitdepth=false)
#U = UtoY()
#V = VtoY()
#YtoUV(U, V, Y)

#LimitedSharpenFaster()

#stackhorizontal(\
#subtitle(video_org,"video_org",size=20,align=2),\
#subtitle(video_restored,"video_restored",size=20,align=2)\
#)

[lordsmurf]

@themaster1 -- excellent work!

[hermetic_charm]

Quote:

Avisynth is your friend. Could take days if you have hours of it

Thanks, that result looks a lot better than the original.

I did also try yesterday a camcorder color denoise filter, and a chroma noise reduction filter. I admittedly did not play with the levels too much, but the CCD appeared to do almost nothing, the CNR had decent results but not as good as your script.

I plan on using Avisynth quite a bit once I start work on restoring videos. I am a software developer by trade, so the scripting is definitely not an issue, but more so learning about all of the video defects, and which functions are best in what scenarios, and getting a "feel" for thresholds/levels to set. I imagine this is a knowledge pool that takes years to build up. This is a good starting point though. Thank you for the script example I will walk through it and reference the avisnyth documentation to try and start building that knowledge base.

Since I am going to keep "raw" archival files, I know I can rest somewhat easy for now on issues that can be fixed in post, since I could always go back and refilter or experiment with different things. Right now, I want to focus mostly on being able to recognize which defects I can fix in post vs. which ones might need immediate attention and a recapture.

This has been a great example, thanks!

[Selur]

Playing around using some machine learning filters (https://pastebin.com/C6BpujbX), could probably use tweaking, but thought I share it just got give an idea what could be done. (For proper filtering of such a source, scene-by-scene filtering is needed.)