Panasonic camcorder (1988) transfer oddities?

[disco444]

Hello, I am looking for recommendations on optimizing everything before transferring the family VHS home movies, most of which were made with the Panasonic PV-420D camcorder from 1988, and in particular, regarding a banding effect on the JVC HR-DVS3U.

(1) Banding issue: With the Auto picture setting on my JVC HR-DVS3U I am seeing a banding effect that I do not see on the Sharp or Soft setting. It is present on all tapes recorded with that camcorder, prominent on some scenes more than others. Attached are some samples with the Auto and Sharp settings of the JVC as well as a transfer direct from the original camcorder. The effect is also present with the same tapes on my other VCRs, so it is not a defect of the JVC in my estimation, but more of a difference in Panasonic's 1980s VHS standards.

In summary, the effect persists on the Auto setting, but is not really noticeable on the Sharp setting nor when the tape is played back on the original camcorder (though various artifacts are visible this way - I can definitely see the advantage of the JVC VCR with TBC on).

I can also confirm that this banding issue is not limited to this specific camcorder but is likely present on various models from different manufacturers. Other footage recorded with two different 1980s camcorders, one an early 1980s model with a separate VCR recording component and camera and another mid-1980s camcorder, produced the same banding result. Besides that, there is no problem transferring tapes not made with these 1980s camcorders (they look great and have no banding), so it is most certainly something to do with the possibly out of spec designs of some 1980s Panasonic stuff as discussed elsewhere on the forums. Everything is SP. The tapes are not copies but are tapes that were in the camcorder while the movies were recorded. I have tried with various settings on and off and they did not address the issue (TBC ON / stabilizer off, stabilizer on and TBC necessarily off, both off, the various audio monitor settings hifi, hifi L, hifi R, norm, mix, etc.).

Is "Sharp" the best option or can something else be adjusted with the equipment I have?

(2) Recommendations: I will have to do some more reading on the forums but if anyone has any advice and recommendations specific to processing of old 1980s camcorder videos I would appreciate anyone sharing them here. I was planning on at least doing some color balance adjustments and cropping and have used Avisynth to process videos in the past.

Setup:
JVC HR-DVS3U > Monster gold plated S-Video cable > ATI AIW X1900 PCIe > Dedicated XP SP3 PC with Core 2 Quad CPU
Also tested other VCRs and original camcorder through composite output and a high quality, heavy, gold plated composite cable connected to the same AIW breakout box.
0 dropped and 0 inserted frames outside of the start of capture and places where there are recording gaps-for example when a movie was videotaped, then we played it back, cued the tape to just after the end of that clip, and recorded again after that).
no frame tbc, no ES-15 or similar

VCR settings:
Calibration ON (this model is noted as similar to the SR-VS30 which benefits from this setting - I tried it on and off and can confirm the result is worse without it), picture control (Auto and Edit have the issue, Sharp does not, Soft does not but the image is poor for what I am transferring), Digital R3 off, stabilizer off (tried with it on and TBC off), TBC on

VirtualDub 1.19.11 settings:
Video cropping to determine histogram levels, then it is all set to 0 before recording
Video preview, preview acceleration-progressive both fields, histogram on, video source-s video, capture pin YUY2, output 720x480
graphstudionext open to change the video proc amp settings (156 brightness, 90 contrast, hue/sat both unchanged at 128) to adjust the histogram so there are no red values during test playback. As was mentioned a number of times on other forum threads, these old camcorders have high red values so the contrast has to be brought down and the image looks bad with even slightly lower brightness settings due to black crushing)
An odd bug - changing the audio to the ATI analog capture instead of the capture default causes the capture to crash soon after. I leave the audio selection to the default "0 capture device" and all is well.
set Compression lagarith, configure-YUY2, multithreading (cpu usage is about 20-30% while capturing)
Capture timing-drop / insert frames both checked, sync video to audio (because I am syncing with an external audio source, sync audio to video does not work as well in my testing to keep the a/v sync)
press F6 to capture, esc to stop

Side notes:
To add to the discussion of old Panasonic mono linear audio not playing well on 1990s+ decks, I can attest to this with the movies recorded with this camcorder. The audio plays with very good quality when the tape is played back on our original camcorder, but plays poorly on any of our other VCRs (this 2004 JVC, a ~2000 Mitsubishi, and a ~2000 Sony). This is why I am capturing audio both through the JVC and separately while the tape is played by the camcorder.

I can easily see the quality difference between my current setup and my previous setup involving a Canon HV-30 with analog-DV passthrough. The old image has comparably little detail and appears washed out compared to my initial transfer of a couple tapes on this new setup. I can definitely recommend the best practices described on this forum for those looking for high quality transfers.

I also used composite as a test for the original camcorder and for my other VCRs in comparison with the JVC because they have no S-Video port. My testing showed that the JVC on Auto / Norm / Edit and my other 1990s and later VCRs produced the same banding issue with footage from this specific camcorder while JVC on Sharp and playback on the original camcorder did not, though I am wary of using sharp from what I have read.

I tried Edit and it produced the same issues as Auto and Norm (Norm only appeared when I turned off video calibration, so I tried Norm at that time).

My post mentioned using the camcorder as a test to show the lack of banding (but the picture artifacts make this choice not worthwhile) and also to transfer the old linear mono audio, which did not transfer properly on any 1990s and later VCRs that I own. Barring a purchase of another VCR made in the 1980s I am not sure of a better solution to transfer the audio at the same or better quality as the original camcorder, but I am not worried about the audio for now - primarily the banding issue and also I am looking for advice and recommendations specific to processing of old 1980s camcorder videos.

Update: Yesterday I recorded some new footage with the original camcorder and can confirm the same conclusions as noted earlier. I am hopeful that this discussion will help others who are transferring consumer 1980s camcorder footage. The issue is not limited to the capture card either. This issue is noticeable playing back the tape to other capture devices or to a TV connected directly.

[Sharc]

By "banding" you mean the horizontal lines?

On the auto.mp4 you could try something like

LWLIbavVideoSource("156 90 Auto.mp4",cache=false)
QTGMC(InputType=3)
vinverse()

The Sharp has strong sharpening artifacts.
The PV-420D has rainbows, color noise and flicker.

How did you deinterlace the video?

[disco444]

The issue is the darker horizontal lines that are visible most noticeably on solid areas such as the nearly-solid tile background. These are different from the sharp lines when I play an interlaced file with no deinterlacing.

I ran the script and have attached the results as well as the original TFF interlaced file and a screenshot of the deinterlaced file. While this does not solve the issue I think that the QTGMC recommendation will be a good option to deinterlace the footage after capture.

[disco444]

Any ideas on the banding issue?

[disco444]

The issue of different-color lines is identical with home movies from three different VHS camcorders in videos recorded in the 80s when the camcorders were relatively new (as well as in my testing of a new recording last month), so I am thinking it has something to do with different behavior of some 80s VCRs in camcorders and/or standalone models. I doubt three camcorders manufactured in different years were factory misaligned in the same exact manner.

I'd describe this issue as lines across the screen, most visible on the lighter colors in the video (the tiles, for example). In my 5/19 post I attached an enlarged still from the deinterlaced video captured on the JVC in Auto, with the VCR settings mentioned in my OP.

I do have the original camcorder but as shown in the above posts, the results are noticeably poorer, with visual artifacts, compared to my transfer on the high end JVC VCR.

I have seen videos with tapes out of alignment and folks gradually adjusting the screws on the VCR head for a cleaner picture but those same big symptoms are not present here. As noted earlier, the banding is significantly reduced or eliminated with the Sharp setting on JVC, but it adds unpleasant artifacts to the captured image.

This may be a red herring but I do count about 120 lines from top to bottom, which is half of the 240 interlaced lines of resolution for VHS. I don't think it's an issue with my setup since all other tapes besides ones made with these three camcorders transfer without these artifacts, including those made with a 1996 VHS-C camcorder.

There is much good discussion about specific frame/external TBC models that are recommended for best results, but my understanding is that they do not play a visible role in the picture but more with the timing and syncing between frames, so they may not address this specific issue. The issue appears to be that mostly lighter parts of the image are not consistent and show lines when played back on 1990s/2000s decks such as those I have tested.

[oln]

The banding/horizontal lines effect is common on older camcorder recordings. Most likely the difference between VCRs a difference in how much the noise reduction in the VCRs "blend" it out.

[disco444]

Thanks for that advice. Since there may not be a way to prevent these lines at the capture stage, I am wondering if anyone could try to remove them at the processing stage with Avisynth, without noticeably affecting the rest of the image as a plugin that does noise reduction may do?

I recommend testing with the clip that has the most noticeable artifacts, linked below:

156 90 JVC Auto interlaced.avi

https://forum.videohelp.com/attachments/71145-1684520806/156%2090%20JVC%20Auto%20interlaced.avi

I have tried ChromaDestripe and Mergechroma with various settings including the settings below.

ChromaDestripe(top=480, blur=6, thresh=6)

MergeChroma(last, SeparateFields().Blur(0,1.0).Sharpen(0,0.7).Weave( ))

I did some research on Hanover bars but that appears to be for PAL; plugins that appear to remove Hanover bars like Chromadestripe do not seem to work with the lines in my NTSC sample clips.

[poisondeathray]

Why is the lagarith file "156 90 JVC Auto interlaced.avi" in RGB ? Double check your procedure

[jagabo]

The horizontal stripes are easy to get rid of:

This isn't the best method but it's easy:

LWLibavVideoSource("156 90 JVC Auto interlaced.avi") 
AssumeTFF()
ConvertToYV12(interlaced=true)
SeparateFields()
vInverse() # gets rid of the hz lines
Weave()
QTGMC() # for viewing

Before bothering with other methods you should make sure you get a proper capture. Do not capture as RGB or convert to RGB. Capture as YUV 4:2:2 (YUY2, UYVY, etc.) and use a lossless codec in YUV mode; do not force/let the program/codec convert to RGB.

[disco444]

Thanks jagabo - your suggestion works well!

I am curious, what are other methods that may be even better? This one seems to be great.

Should I be using ConvertToYV16(interlaced=true) instead of YV12 if I am capturing in YUV422 and want to keep things the same for archival storage of the original capture and the processed version?

Apologies for the incorrect color space in the sample files. I capture in YUY2 with Lagarith with the process described in my OP. I use Magix Vegas to do final edits but should have trimmed the clips directly in VirtualDub2 for upload to avoid a possible issue in Vegas. I can confirm the original virtualdub-captured nonedited files have the same issue.

[jagabo]

I am curious, what are other methods that may be even better?

vInverse works by blurring the horizontal lines (when it detects them) away. That leads to a slightly less sharp picture. With this type of problem you may be able to lighten or darken scan lines instead. Modifying the earlier script:

LWLibavVideoSource("156 90 JVC Auto interlaced.avi") 
AssumeTFF()
ConvertToYV12(interlaced=true)

SeparateRows(4)

row0 = SelectEvery(4, 0) # a clip of rows 0, 4, 8, 12...
row1 = SelectEvery(4, 1) # a clip of rows 1, 5, 9, 13...
row2 = SelectEvery(4, 2).ColorYUV(gain_y=5) # a clip of rows 2, 6, 10, 14...
row3 = SelectEvery(4, 3).ColorYUV(gain_y=3) # a clip of rows 3, 7, 11, 15...

Interleave(row0, row1, row2, row3) # rows back to sequential frames
WeaveRows(4) # weave them back to the original frames

QTGMC() # for viewing

Row 2 of every 4 (row numbering starts at 0) gets a gain_y of 5, row 3 of every 4 gets a gain_y of 3. After weaving them back together into a single frame the horizontal lines are nearly gone. I think the chroma needs similar adjustments but didn't address it here.

Should I be using ConvertToYV16(interlaced=true) instead of YV12 if I am capturing in YUV422 and want to keep things the same for archival storage of the original capture and the processed version?

Yes. Or you can leave it as YUY2 if the filters work in YUY2. This script works in YUY2.

[disco444]

Thanks again jagabo - your second script works even better! After looking more closely, I now see what you mean by the blurring from the first script.

Many thanks to all for the help with troubleshooting. I appreciate it!

I think this camcorder lines issue is happily resolved now. Hopefully other folks with the same issue will be able to refer to your avisynth script to fix things.

I am curious, what are other methods that may be even better?

vInverse works by blurring the horizontal lines (when it detects them) away. That leads to a slightly less sharp picture. With this type of problem you may be able to lighten or darken scan lines instead. Modifying the earlier script:

LWLibavVideoSource("156 90 JVC Auto interlaced.avi") 
AssumeTFF()
ConvertToYV12(interlaced=true)

SeparateRows(4)

row0 = SelectEvery(4, 0) # a clip of rows 0, 4, 8, 12...
row1 = SelectEvery(4, 1) # a clip of rows 1, 5, 9, 13...
row2 = SelectEvery(4, 2).ColorYUV(gain_y=5) # a clip of rows 2, 6, 10, 14...
row3 = SelectEvery(4, 3).ColorYUV(gain_y=3) # a clip of rows 3, 7, 11, 15...

Interleave(row0, row1, row2, row3) # rows back to sequential frames
WeaveRows(4) # weave them back to the original frames

QTGMC() # for viewing

Row 2 of every 4 (row numbering starts at 0) gets a gain_y of 5, row 3 of every 4 gets a gain_y of 3. After weaving them back together into a single frame the horizontal lines are nearly gone. I think the chroma needs similar adjustments but didn't address it here.

Should I be using ConvertToYV16(interlaced=true) instead of YV12 if I am capturing in YUV422 and want to keep things the same for archival storage of the original capture and the processed version?

Yes. Or you can leave it as YUY2 if the filters work in YUY2. This script works in YUY2.

[jagabo]

The exact values in the script are specific to your video (VCR?) but the idea can be adapted to other videos with a similar problem. I've seen other videos like yours here before.

Have you tried a good head cleaning? That may eliminate the problem at the source.

[disco444]

I can confirm your script works with other videos captured with the same camcorder. I will check over the coming weeks to see if the script works with VHS tapes from other camcorders or requires modification.

Yes I have cleaned the JVC VCR and have tried on some other VCRs as well, including a Mitsubishi and a Sony. All show the exact same lines with the same VHS tape tried in each VCR. Also last month after cleaning the original camcorder's tape path as well, I recorded some video on the original camcorder to a new part of VHS tape and the same issues were present.

As noted earlier, the issue of different-color lines is identical with home movies from three different VHS camcorders in videos recorded in the 80s when the camcorders were relatively new (as well as in my testing of a new recording last month), so I am thinking it has something to do with different behavior of some 80s VCRs in camcorders and/or standalone models. I doubt three camcorders manufactured in different years were factory misaligned/worn in the same exact manner or were not cleaned in a way that produced the same artifacts.

Since the issue shows up with various VCRs and original tapes from various original camcorders I do not think the issue is with the alignment or cleaning of a particular VCR or camcorder but I will see how things go with the transfers.