Problem with VHS transcription

[Graviton]

I am trying to remove these horrible, brownish-yellow bands from a transcription of Betacam SP to standard NTSC VHS tape.

The film is in decent color.

The bands sweep slowly from top to bottom of the screen.

Is there any suggestion anyone would have in removing these ugly defects?:

http://i46.tinypic.com/33yoyzq.jpg

Yes, there is color banding in there too, but that solution is already known to me.

I am using Huffyuv 2.1 (lossless) codec contained in an avi for the film capture (from VHS) format, in YUY2.

Your assistance would be *immensely* appreciated, since I have this problem with a lot of their tapes.

PS: A secondary, minor question is what preview editor is best for executing Avisynth scripts, since VirtualDubMod crashes when I load any video file.

[Mounir]

Can you post a video sample so we can see in details the problem

the best editor is avspmod, search on doom9 you'll find

[StainlessS]

Perhaps nothing to do with (EDIT: VD) problem, but, VDMod crashes when bad plugins in plug dir (avisynth) and also
when YV12 is not mod 4 (perhaps other colors spaces, not sure), VD standard dont have the mod 4 prob.

[lisztfr9]

Imho, for each frame it's difficult to detect the correct position of the artifact since it's rather fuzzy. I have a similar problem but my bars aren't moving...

If you apply an overlay mask you should program its position inside a text file and try to find the correct algorithm before

Another idea would be to slice the video horizontally and apply a color correction on parts with the artifact, if it's possible to detect the brown color, i dunno since i'm still struggling with the basics. I would try that, i hope this could work.

[Mounir]

I think the only solution would be to create a mask with the inverted colors which would phase out (= remove) these orange/green bands.
Then you could animate this mask with a roll motion from top to bottom

You'd need a good software to do this after effects, sony vegas etc..

[pandy]

Perhaps some filtering in frequency domain (seems this perfect job for fquiver or defreq)

[StainlessS]

+1 on the sample, ideally vertical strip of supposed uniform color in there somewhere.

There seems to be greenish area above and below brown/yellow bars. (EDIT: Already mentioned by Mournir)
Width of supplied jpg is 873 (???).

EDIT:

Rough hack out from supplied frame:

ImageSource("D:\avs\avi\1.jpg")
Crop(0,0,-1,0)
L=crop(118,0,4,0)
M=crop(590,0,4,0)
R=crop(654,0,4,0)
K=BlankClip(L)

Q=248

Y1=L.Crop(0, 0,0,Q)
Y2=M.Crop(0,Q,0,322-Q)
Y3=R.Crop(0,322,0,0)

Y=StackVertical(Y1,Y2,Y3)
Return Y.BilinearResize(640,480)
StackHorizontal(L,M,R,K,Y)

https://s20.postimg.org/jtgatmr99/image.jpg (https://postimg.org/image/c0pn1nla1/)

Find vertical step between bands


ImageSource("D:\avs\avi\1.jpg")
Crop(0,0,-1,0)
L=crop(118,0,4,0)
M=crop(590,0,4,0)
R=crop(654,0,4,0)
K=BlankClip(L)

Q=248

Y1=L.Crop(0, 0,0,Q)
Y2=M.Crop(0,Q,0,322-Q)
Y3=R.Crop(0,322,0,0)

Y=StackVertical(Y1,Y2,Y3)
#Return Y.BilinearResize(640,480)
#Return StackHorizontal(L,M,R,K,Y)

Y=Y.BilinearResize(640,480)

GScript("""
best = 0.0 IX=0
for(i=201,480-200) {
Tmp =RT_LumaCorrelation(Y,Y,n=0,y=i,h=200,n2=0,y2=0)
RT_Debug("I=",String(i),"Correlation=",String(Tmp))
if(tmp > best) {
best=tmp
ix=i
}
}
RT_Debug("BANDSTEP=",String(ix),"Correlation=",String(Best))
""")
return Y


EDIT: Below output via MS DebugView.

00000002 1.54000688 [5856] RT_Debug: I= 201 Correlation= 0.209317
00000003 1.54353225 [5856] RT_Debug: I= 202 Correlation= 0.206680
00000004 1.54720795 [5856] RT_Debug: I= 203 Correlation= 0.230926
00000005 1.55069411 [5856] RT_Debug: I= 204 Correlation= 0.237002
00000006 1.55416560 [5856] RT_Debug: I= 205 Correlation= 0.248360
00000007 1.55764890 [5856] RT_Debug: I= 206 Correlation= 0.262505
00000008 1.56126904 [5856] RT_Debug: I= 207 Correlation= 0.285732
00000009 1.56477547 [5856] RT_Debug: I= 208 Correlation= 0.285995
00000010 1.56827664 [5856] RT_Debug: I= 209 Correlation= 0.296805
00000011 1.57176948 [5856] RT_Debug: I= 210 Correlation= 0.306889
00000012 1.57633257 [5856] RT_Debug: I= 211 Correlation= 0.315260
00000013 1.57917249 [5856] RT_Debug: I= 212 Correlation= 0.324560
00000014 1.58266473 [5856] RT_Debug: I= 213 Correlation= 0.334283
00000015 1.58615649 [5856] RT_Debug: I= 214 Correlation= 0.349140
00000016 1.58963799 [5856] RT_Debug: I= 215 Correlation= 0.356119
00000017 1.59324574 [5856] RT_Debug: I= 216 Correlation= 0.372216
00000018 1.59681463 [5856] RT_Debug: I= 217 Correlation= 0.369759
00000019 1.60031056 [5856] RT_Debug: I= 218 Correlation= 0.376175
00000020 1.60380268 [5856] RT_Debug: I= 219 Correlation= 0.370589
00000021 1.60749352 [5856] RT_Debug: I= 220 Correlation= 0.394022
00000022 1.61099756 [5856] RT_Debug: I= 221 Correlation= 0.387819
00000023 1.61450195 [5856] RT_Debug: I= 222 Correlation= 0.415276
00000024 1.61800659 [5856] RT_Debug: I= 223 Correlation= 0.420252
00000025 1.62148499 [5856] RT_Debug: I= 224 Correlation= 0.434715
00000026 1.62498093 [5856] RT_Debug: I= 225 Correlation= 0.435403
00000027 1.62849247 [5856] RT_Debug: I= 226 Correlation= 0.448592
00000028 1.63199687 [5856] RT_Debug: I= 227 Correlation= 0.461991
00000029 1.63549852 [5856] RT_Debug: I= 228 Correlation= 0.465885
00000030 1.63928640 [5856] RT_Debug: I= 229 Correlation= 0.475005
00000031 1.64281094 [5856] RT_Debug: I= 230 Correlation= 0.476240
00000032 1.64630890 [5856] RT_Debug: I= 231 Correlation= 0.492250
00000033 1.64981961 [5856] RT_Debug: I= 232 Correlation= 0.482598
00000034 1.65350664 [5856] RT_Debug: I= 233 Correlation= 0.507530
00000035 1.65704083 [5856] RT_Debug: I= 234 Correlation= 0.496039
00000036 1.66055155 [5856] RT_Debug: I= 235 Correlation= 0.521639
00000037 1.66403854 [5856] RT_Debug: I= 236 Correlation= 0.516376
00000038 1.66755068 [5856] RT_Debug: I= 237 Correlation= 0.536051
00000039 1.67198586 [5856] RT_Debug: I= 238 Correlation= 0.533805
00000040 1.67541611 [5856] RT_Debug: I= 239 Correlation= 0.553372
00000041 1.67896843 [5856] RT_Debug: I= 240 Correlation= 0.568634
00000042 1.68256187 [5856] RT_Debug: I= 241 Correlation= 0.581498
00000043 1.68613946 [5856] RT_Debug: I= 242 Correlation= 0.595618
00000044 1.68969584 [5856] RT_Debug: I= 243 Correlation= 0.594343
00000045 1.69320333 [5856] RT_Debug: I= 244 Correlation= 0.633665
00000046 1.69671643 [5856] RT_Debug: I= 245 Correlation= 0.632473
00000047 1.70129740 [5856] RT_Debug: I= 246 Correlation= 0.650002
00000048 1.70407069 [5856] RT_Debug: I= 247 Correlation= 0.650783
00000049 1.70757890 [5856] RT_Debug: I= 248 Correlation= 0.689115
00000050 1.71106470 [5856] RT_Debug: I= 249 Correlation= 0.691440
00000051 1.71455359 [5856] RT_Debug: I= 250 Correlation= 0.699778
00000052 1.71816969 [5856] RT_Debug: I= 251 Correlation= 0.706419
00000053 1.72172332 [5856] RT_Debug: I= 252 Correlation= 0.729977
00000054 1.72524250 [5856] RT_Debug: I= 253 Correlation= 0.725380
00000055 1.72876167 [5856] RT_Debug: I= 254 Correlation= 0.746394
00000056 1.73319268 [5856] RT_Debug: I= 255 Correlation= 0.754065
00000057 1.73674142 [5856] RT_Debug: I= 256 Correlation= 0.770373
00000058 1.74025750 [5856] RT_Debug: I= 257 Correlation= 0.782532
00000059 1.74378252 [5856] RT_Debug: I= 258 Correlation= 0.791189
00000060 1.74746895 [5856] RT_Debug: I= 259 Correlation= 0.783710
00000061 1.75098383 [5856] RT_Debug: I= 260 Correlation= 0.780501
00000062 1.75446475 [5856] RT_Debug: I= 261 Correlation= 0.792011
00000063 1.75795901 [5856] RT_Debug: I= 262 Correlation= 0.785016
00000064 1.76143658 [5856] RT_Debug: I= 263 Correlation= 0.795789
00000065 1.76531148 [5856] RT_Debug: I= 264 Correlation= 0.786895
00000066 1.76882684 [5856] RT_Debug: I= 265 Correlation= 0.805786
00000067 1.77230930 [5856] RT_Debug: I= 266 Correlation= 0.796240
00000068 1.77582359 [5856] RT_Debug: I= 267 Correlation= 0.814535
00000069 1.78099620 [5856] RT_Debug: I= 268 Correlation= 0.804265
00000070 1.78320563 [5856] RT_Debug: I= 269 Correlation= 0.818335
00000071 1.78669989 [5856] RT_Debug: I= 270 Correlation= 0.813854
00000072 1.79020691 [5856] RT_Debug: I= 271 Correlation= 0.825230
00000073 1.79387343 [5856] RT_Debug: I= 272 Correlation= 0.813301
00000074 1.79857433 [5856] RT_Debug: I= 273 Correlation= 0.808953
00000075 1.80229354 [5856] RT_Debug: I= 274 Correlation= 0.806657
00000076 1.80588841 [5856] RT_Debug: I= 275 Correlation= 0.790828
00000077 1.80948567 [5856] RT_Debug: I= 276 Correlation= 0.798685
00000078 1.81295455 [5856] RT_Debug: I= 277 Correlation= 0.760840
00000079 1.81644046 [5856] RT_Debug: I= 278 Correlation= 0.772111
00000080 1.81991923 [5856] RT_Debug: I= 279 Correlation= 0.757441
00000081 1.82340872 [5856] RT_Debug: I= 280 Correlation= 0.768275
00000082 1.82449806 [5856] RT_Debug: BANDSTEP= 271 Correlation= 0.825230


EDIT: Vertically 271 pixels apart.

EDIT: Using RT_YDifference instead of RT_Correlation gives 279 pixels apart, neither is perfect.

[Nightshiver]

Those bands look like they are problems with the actual tape itself. You could try cleaning the physical tape or getting a better VCR.

[lisztfr9]

Did you try Tweak on this ?

http://avisynth.org/mediawiki/Tweak