#1  
04-30-2019, 11:52 AM
ELinder ELinder is offline
Unconfirmed User
 
Join Date: Oct 2018
Posts: 197
Thanked 33 Times in 27 Posts
I've got a tape that has a changing telecine cadence. I'm seeing this for the first time, and I'm not sure what else I can try to get tfm to better detect the out of sequence combed frames and restore them. I'd like to avoid either dropping the frames so there's a big movement jump or ending up with a duplicate frame decimate doesn't remove while still getting to the target 23.976fps. Even if I get an occasional blended frame that'd be preferable to either of the other one of those. I've tried the IVTC options directly in Virtualdb also, but that doesn't perform as well as TIVTC.

This clip is from the VCR(with TBC setting on) --> TBC-4000 --> VC500 USB. I've also tried with the VCR(TBC setting off) --> EC15 --> TBC-4000 --> VC500, but the result is siginificantly worse.

This is giving me the best result so far.

ConvertToYV12()
tfm(order=1, mode=5, pp=6, field=1, slow=2)
TDecimate()
Vinverse2()
Santiag()


Attached Files
File Type: avi EML-VC500-TBC-on-clip.avi (33.87 MB, 3 downloads)
Reply With Quote
Someday, 12:01 PM
admin's Avatar
Ads / Sponsors
 
Join Date: ∞
Posts: 42
Thanks: ∞
Thanked 42 Times in 42 Posts
  #2  
04-30-2019, 09:10 PM
sanlyn sanlyn is offline
Premium Member
 
Join Date: Aug 2009
Location: N. Carolina and NY, USA
Posts: 3,648
Thanked 1,307 Times in 982 Posts
Good try, but incorrect use of mode 5. Read what mode 5 requires, from the TIVTC html that comes with the TIVTC package:
Quote:
Mode 5 = This is similar to mode 3 (mkv vfr, and requires hybrid=2 and cycleR=1), but as part of a two pass process. It requires a complete input file (obtained via the output parameter on a previous pass), and a complete tfmIn file (see the batch option for a way around these requirements).
Why are you using vInverse and santiag?

I don't see any combing. But i do see the following in the results of your script, which you can verify nfor yourself if mou open the result in VirtualDub and scroll one frame at a time:
Dropped frame between frames 55 and 56.
61 and 62 are duplicate frames.
Dropped frame between frames 74 and 75.
Dropped frame between frames 103 and 104.
104 and 105 are duplicate frames.

It's a very uneven cadence from start to finish. It appears that the original has some dropped and inserted fields, likely in the problem areas noted above. Looks like a PAL to NTSC conversion.

BTW. TIVTC works in YUY2. If you're going to convert to YV12, you must specify whether interlaced is true or false.

I get very similar results with this:

Code:
AssumeTFF()
TFM().TDecimate()
return last
Reply With Quote
  #3  
04-30-2019, 09:53 PM
ELinder ELinder is offline
Unconfirmed User
 
Join Date: Oct 2018
Posts: 197
Thanked 33 Times in 27 Posts
I don't see that description for mode 5 on either https://www.afterdawn.com/guides/arc...th_page_11.cfm or http://avisynth.nl/index.php/TIVTC/TFM. Where is that html page?

Vinverse is in there for later in the capture where there's some quick motion of not quite vertical lines, not just for this clip.

It's definitely not a PAL conversion, it's a first generation tape from the studio. I've also tried capturing on a second vcr with the same results. Vdub reports no dropped frames on any of my capture gear combinations. Just after this section the telecine stabilizes into a normal cadence, but about 10 minutes later in the tape there's again a few seconds of this problem, then it stabilizes again. Two different episodes on the same tape exhibit no problems at all.

Is it possible the problem happened when they did the original transfer from film to video?

Erich
Reply With Quote
  #4  
05-01-2019, 12:30 AM
sanlyn sanlyn is offline
Premium Member
 
Join Date: Aug 2009
Location: N. Carolina and NY, USA
Posts: 3,648
Thanked 1,307 Times in 982 Posts
It just looks like faulty mastering work. I have some pretty ugly examples of that kind of stuff myself, and the makers had the nerve to sell it retail. I spent months cleaning it up.

[QUOTE=ELinder;61150]I don't see that description for mode 5 on either https://www.afterdawn.com/guides/arc...th_page_11.cfm or http://avisynth.nl/index.php/TIVTC/TFM. Where is that html page?
If you downloadwed the version 1.0.5 zip or 7z TIVTC package and decompressed it, there are a bunch of files in that package, not just the filters. All the docs are in there. It's tivtc_tdecimate.htm. There is also tivtc_tfm.htm.

If you have the newer version 1.0.10 that's posted on the wiki, the docs inside the .7z download are "TDecimate - READ ME.txt" and "TFM - READ ME.txt".

That video also has lots of grungy noise, ugly color, smeared reds, screechy overmodulated audio, and chroma ghosting all over the place. Source looks like it came off a JVC that needs work. The pulldown pattern is absurdly broken, it seems to approach a cycle of 10 repeats every 42 frames, but that's only an approximation. I managed to play around with the sample and made an Excel spreadsheet of the frame numbers and matching repeat frames after running TFM. What a waste of time. I got mostly coherent frames out of it, but the time phasing is goofy or something because it plays with jerks/missing frames in the formerly bad areas anyway. I had to do similar grunt work with an Aussie import a few years back. But it wasn't as bad as this one.

For what it's worth (which isn't much): Conversion to RGB32 was for Virtualdub filters to reduce red over-saturation (red "blooming") and a quickie denoise with NeatVideo at 65%. The NeatVideo noise profile was taken of the background walls in the sample. In the SelectEvery() statements below you can see how certain patterns repeat, but it's still random. Results attached as EML_bad_telecine_23.976.mp4:

Code:
AviSource("D:\forum\faq\ELinder\EML-VC500-TBC-on-clip.avi")
AssumeTFF()
TFM(pp=0)
FixChromaBleeding()
srce=last

V1=srce.Trim(0,4).SelectEvery(5,0,2,3,4).AssumeFPS("ntsc_Film")
V2=srce.Trim(5,29).SelectEvery(5,1,2,3,4).AssumeFPS("ntsc_Film")
V3=srce.Trim(30,36).SelectEvery(7,1,2,3,4,5,6).AssumeFPS("ntsc_Film")
V4=srce.Trim(38,41).SelectEvery(4,0,1,2).AssumeFPS("ntsc_Film")
V5=srce.Trim(42,61).SelectEvery(5,1,2,3,4).AssumeFPS("ntsc_Film")
V6=srce.Trim(62,68).SelectEvery(7,1,2,3,4,5,6).AssumeFPS("ntsc_Film")
V7=srce.Trim(70,76).SelectEvery(7,1,2,3,5,6).AssumeFPS("ntsc_Film")
V8=srce.Trim(77,86).SelectEvery(5,1,2,3,4).AssumeFPS("ntsc_Film")
V9=srce.Trim(87,93).SelectEvery(7,1,2,3,4,6).AssumeFPS("ntsc_Film")
V10=srce.Trim(94,123).SelectEvery(5,1,2,3,4).AssumeFPS("ntsc_Film")
V11=srce.Trim(124,130).SelectEvery(7,1,2,3,4,6).AssumeFPS("ntsc_Film")
VAA=srce.Trim(131,140).SelectEvery(5,1,2,3,4).AssumeFPS("ntsc_Film")
VAB=srce.Trim(141,142).SelectEvery(2,1).AssumeFPS("ntsc_Film")

Out1=V1 + V2 + V3 + V4 + V5 + V6 + V7 + V8 + V9 + V10 + V11 + VAA + VAB
Out1
AssumeFPS("ntsc_Film",sync_audio=true)
ConvertToRGB32(interlaced=false,matrix="Rec601")
Crop(12,2,-6,-6).AddBorders(10,4,8,4)
return last
Most of the time, bad source like this isn't worth this much trouble. But it's good practice.


Attached Files
File Type: mp4 EML_bad_telecine_23.976.mp4 (2.95 MB, 2 downloads)
Reply With Quote
  #5  
05-01-2019, 08:51 AM
ELinder ELinder is offline
Unconfirmed User
 
Join Date: Oct 2018
Posts: 197
Thanked 33 Times in 27 Posts
First, thanks for taking such a close look at this. Thankfully it's the only time I've run across this kind of problem so far...knock on wood.

I must not be understanding the relationship between tfm and decimate, as I thought I was using the tfm mode=5 only in tfm to increase where and how it looks for combed frames and matches. Is that not different than calling for the tdecimate mode=5?

I'm also surprised by your comment about the JVC deck looking like it needs work, but you've obviously seen exponentially more of this kind of thing than I have. Both the JVC SR-V101US and HR-S7800U, both of which I bought from LS here about 6 months ago have very similar image quality. Both decks, with both the VC500 and 510-USB capture devices, need a healthy procamp brightness increase and slight contrast decrease to get a proper capture histogram without low or high clipping. Also, a light scrubbing thru NeatVideo is part of my later workflow after, but I have the OFX version, so I'm doing it in Resolve.

Erich
Reply With Quote
  #6  
05-01-2019, 01:41 PM
sanlyn sanlyn is offline
Premium Member
 
Join Date: Aug 2009
Location: N. Carolina and NY, USA
Posts: 3,648
Thanked 1,307 Times in 982 Posts
I'm a former JVC high-end user myself and went through 4 of them (5, actually, #4 and #5 were consecutive copies, both defective out of the box, brand new). Both JVC's you mention are noted for good images. That's not what the video sample looks like. It's reminiscent of faulty images we've seen from JVC's that need maintenance. If indeed that's how your sample plays nowadays, it looks like sloppy mastering work before it came into your possession.
Reply With Quote
Reply




Similar Threads
Thread Thread Starter Forum Replies Last Post
Dealing with tricky Video8 tapes hodgey Capture, Record, Transfer 8 05-10-2023 06:18 AM
Confused about inverse telecine stevevid Restore, Filter, Improve Quality 5 12-20-2018 11:35 PM
How to identify telecine pulldown? ragu0012 Restore, Filter, Improve Quality 2 04-10-2018 02:21 PM
Best scripts for de-telecine ragu0012 Restore, Filter, Improve Quality 8 03-17-2018 08:36 PM
Dealing with Audio on VHS captures - EQ and mixers? VideoTechMan Capture, Record, Transfer 2 05-10-2012 01:15 PM

Thread Tools



 
All times are GMT -5. The time now is 04:23 AM