Dark "wipes" across the video

[wcndave2]

Does anyone know what causes these?

I was thinking dirty tape vs damaged tape and wondering if anything could be done.


[Attachment 93859 - Click to enlarge]

[lollo]

Looks like "dark comets".

Better to post a sample of the video rather than a fixed image to find a solution.

[wcndave2]

Looks like "dark comets".

Better to post a sample of the video rather than a fixed image to find a solution.

Thanks, I have attached a file here - I think...

[lollo]

problematic tape: dark comets everywhere (but somehow moving from frame to frame, so a fix could be achieved); defective frames 33, 34.

For first try median on multiple captures or on a single capture despot, removeDirtMC, spotless, FixRipsP2 or similar.

For second try FixRipsP2 (very destructive) or interpolate 33 and 34 from frame 32 and 35.

[Sharc]

... and if this is an unprocessed capture it is way oversharpened (and looks poorly deinterlaced) which emphasizes the comets.

[dellsam34]

This is more likely baked in the original recording, A satellite program recording but seen it in off-air and DVB-T TV programing. I also believe a S-VHS recording played by Quasi S-VHS or a VHS machine can produce such effect.

[lollo]

A satellite program recording but seen it in off-air and DVB-T TV programing.

More likely in old analog satellite broadcast, not in DVB-S/T streams.

The old times with hours spent to perfectly align an analog dish for perfect reception (almost impossbile with bad wheater)

[dellsam34]

I've seen it in digital satellite programing as well but that could be the fault of the source and not the transmission. But yes, the old good analog satellite days, but Teletext comes through just fine, It was a lot of fun.

[Sharc]

Try this. Not perfect but ....

LWLibavVideoSource("p-seg.avi")
converttoYV16()
noisy = last
nonoise = SpotLess(last,ThSAD=5000,ThSAD2=1000,pel=4,chroma=true,BlkSz=24)
diff = 1                        # Difference between denoised and noisy clips
m = mt_lutxy(noisy,nonoise,"y x - "+string(diff)+" - 255 *")
mt_merge(noisy,nonoise,m)

Left: source
right: processed

[wcndave2]

Lots of replies, very helpful community, if somewhat over my head at this time...

This is a purchased VHS from 1987.

I have no idea where it's been stored - it was given to me.

I am ripping using JVC SVHS S7722EK, to Elgato Video Capture.

Using AmarecTV, and what you're seeing is directly what came out of there.

various features on the player like auto-sharpen/soft/etc are all off.
B.E.S.T. mode is on

I have all the "adjustments" in amarectv set to the default neutral position.

The video is 25fps non-interlaced.

[dellsam34]

I'm curious if this is caused by the VCR or the capture device, Because it is unusual to see this in Pre-recorded tapes. A hookup to a TV may rule out one of the two guesses if it's possible.

Some may know me that I like to troubleshoot the root cause of the problem before jumping to software solutions, this is the reason for such questions.

[wcndave2]

I'm curious if this is caused by the VCR or the capture device, Because it is unusual to see this in Pre-recorded tapes. A hookup to a TV may rule out one of the two guesses if it's possible.

Some may know me that I like to troubleshoot the root cause of the problem before jumping to software solutions, this is the reason for such questions.

Sounds sensible.

I have captured 20 tapes in the last 2 weeks, this was one where it was obvious. A lot of the same tape is fine, and other tapes fine.

Take a look at the attached p-seg3, there's the same issue and then when it cuts to the next "scene", it's completely gone.
Then when it goes back to the next scene similar to the first one, it's back again. (p-seg2)

[Sharc]

Black wiskers (comets) have been observed when SVHS tapes are played on standard VHS machines. It seems that this hasn't been the case here though. Also, it has been reported that the comets disappeared when the tape was played with a different VCR. All this comes just from my memory, so no evidence or proof. Comets may originate from a VCR tracking adjustment as well.
Secondly, the Elgato does not have the best reputation on this froum, but I doubt whether it is responsible for the comets. I wonder however whether the OP's source is interlaced and the Elgato did the deinterlacing on the fly (?). If so, it would be sub par. Capture should be lossless interlaced 4:2:2.
Deinterlacing can spread the comets over multiple scan lines and multiple frames. So one normally would remove comets first and deinterlace after.
Maybe the commercial tape was PsF however (Progressive segmented frames), I don't know. Only the OP could tell.
Just guessing, maybe others will chime in ....

P.S. and these ~static artwork title snippets don't tell much. Use scenes with motion for analysis.

[wcndave2]

Black wiskers (comets) have been observed when SVHS tapes are played on standard VHS machines. It seems that this hasn't been the case here though. Also, it has been reported that the comets disappeared when the tape was played with a different VCR. All this just from my memory, so no evidence or proof.
Secondly, the Elgato does not have the best reputation on this froum, but I doubt whether it is responsible for the comets. I wonder however whether the OP's source is interlaced and the Elgato did the deinterlacing on the fly. If so, it would be sub par. Capture should be lossless interlaced 4:2:2. Maybe the commercial tape was PsF however (Progressive segmented frames), I don't know. Only the OP could tell.
Just guessing, maybe others will chime in ....

I got the Elgato for other reasons than VHS capture, and I definitely don't use their bundled software. Compared to those cheap VCAP capture devices they seem very good. I then saw the next level was PCI cards that cost many hundreds... Happy to be advised on a better option that's a reasonable budget.

I have captured plenty that were interlaced, and it's very clear that they are. I also have quite a lot that are a mix of interlaced/progressive.
These were filmed rather than TV cameras, so I guess there's quite a few music videos that are 25fps progressive.

There is no deinterlacing happening in the elgato - otherwise I wouldn't have 70% of my captures in a state that clearly need deinterlacing.

Apart from the fact that it progresses move/still/move/still when one tries to deinterlace a progressive capture, how would one know if it was PsF?

[VWestlife]

Which Elgato are you talking about? Their HDMI capture cards are reportedly pretty good. But their analog video capture device is literally just a rebadged EasyCap-type device, using the same Conexant chip as found in many of them, and sold at a huge markup.
https://www.linuxtv.org/wiki/index.php/Conexant_CX2310x

[wcndave2]

P.S. and these ~static artwork title snippets don't tell much. Use scenes with motion for analysis.

Well, there is motion within them, and this is when the comets happens on that capture.

Here's a screenshot of later on in the video, left is the original, and right is deinterlaced.
This is just a test mode I use to check if the video is interlaced or not.


[Attachment 93867 - Click to enlarge]

Also se p-seg5 attached, which has a lot of movement, clearly interlaced. However never any comets...

[wcndave2]

Which Elgato are you talking about? Their HDMI capture cards are reportedly pretty good. But their analog video capture device is literally just a rebadged EasyCap-type device, using the same Conexant chip as found in many of them, and sold at a huge markup.
https://www.linuxtv.org/wiki/index.php/Conexant_CX2310x

This one: https://amzn.eu/d/0feI5oIf

[lollo]

Which Elgato are you talking about? Their HDMI capture cards are reportedly pretty good. But their analog video capture device is literally just a rebadged EasyCap-type device, using the same Conexant chip as found in many of them, and sold at a huge markup.
https://www.linuxtv.org/wiki/index.php/Conexant_CX2310x

You must be joking. There are no easycap devices using Conexant video ICs.

The CX23102 video IC, which provides excellent quality for video capture, is used in Diamond VC500, Hauppauge USB-Live 2 and the Elgato Video Capture shown by OP in his link.

[Sharc]

P.S. and these ~static artwork title snippets don't tell much. Use scenes with motion for analysis.

Well, there is motion within them, and this is when the comets happens on that capture.

Here's a screenshot of later on in the video, left is the original, and right is deinterlaced.
This is just a test mode I use to check if the video is interlaced or not.


[Attachment 93867 - Click to enlarge]

Also se p-seg5 attached, which has a lot of movement, clearly interlaced. However never any comets...

Yes. basically interlaced with occasional hickups of dropped and repeated fields (p-seg5) of which I do have some doubts as to these are intended motion artwork or just capture flaws.
Anyway, if you are happy with your capture setup just fix the comets using Avisynth or Vapoursynth scripts as proposed. There is no simple QTGMC setting for this.

Added: Similar comet artifacts can also be produced by TBC errors. Maybe substitute the TBC of your VCR with a DVD recorder in passthrough.

[lollo]

I got the Elgato for other reasons than VHS capture, and I definitely don't use their bundled software.

Good choice.

Compared to those cheap VCAP capture devices they seem very good.

Yes.

I then saw the next level was PCI cards that cost many hundreds...

The quality of the card matters more than the "connection". PCI or USB makes no difference in that sense.

Apart from the fact that it progresses move/still/move/still when one tries to deinterlace a progressive capture, how would one know if it was PsF?

Separates the fields and move across them. in a PsF video architecture there is no movement between top and bottom field

I never used an Elgato, but Sharc's remarks are ok. Move the sharpness in the settings close to minimum. The settings on the VCR you listed are ok.

[dellsam34]

Are these black comets appear mostly in bright zones and during a transition between zones of different levels? If so this indicates a bandwidth clipping, severe bandwidth limitation can cause distinct dark or black streaks, but they manifest in a highly specific way known as streaking, trailing, or smearing rather than random horizontal lines across the frame right? a bandwidth issue creates horizontal streaks that specifically drag or "bleed" to the right of bright objects, kind of the ones you see when playing back a S-VHS tape in a VHS machine.

Here is a copy paste from the net about bandwidth clipping:

How Bandwidth Limitations Create Black Streaks:

Analog video translates image brightness directly into voltage levels over time along a horizontal line. High frequencies represent sharp details, crisp edges, and sudden changes in contrast (like transitioning from a white object to a black background).

- Slow Voltage Transitions: If an analog circuit or cable has severely restricted or clipped bandwidth, it acts as a low-pass filter. The signal cannot change voltage fast enough to capture sudden drops in brightness.

- The "Right-Hand" Smear: When the scanning beam finishes drawing a very bright object (high voltage) and suddenly encounters a dark background (low voltage), a bandwidth-clipped system cannot drop the voltage instantly. The voltage takes time to decay, causing a dark, smeared "trail" or streak to stretch out to the right of the object across the screen.

Common Sources of Bandwidth Clipping:

- Poor Coaxial Cable Quality: Utilizing exceptionally long cable runs, unshielded wires, or severely damaged cables will drastically roll off high frequencies, reducing available signal bandwidth.

- Improper Impedance Matching: Analog video relies on a strict 75-ohm impedance across cables, connectors, and equipment inputs. Mishandling this requirement—such as utilizing 50-ohm RF cables or failing to terminate a loop-through monitor correctly—creates signal reflections and severe frequency attenuation.

- Failed Amplifier/Filter Components: Aging capacitors or faulty inline amplifiers within video switchers, VCRs, or CCTV systems can restrict the signal path, artificially clipping the high-frequency spectrum required for a stable picture.

[Sharc]

^^^^ Unlikely that bandwidth constraints ("bandwidth clipping") are the root cause of these comets, IMO.
Slow voltage transitions due to analog bandwidth limitations make sharp transitions soft or gradual, so instead of sharp white->black (or vice versa) transitions the signal changes gradually through greytones, like white->graytones...->black.
If the transition from passband to stopband is very sharp (narrow) one will get ringing artifacts which manifest as "halos" in the video rather than comets.

Improper Impedance Matching: Yes, on long cables this causes signal reflections and - in case of multiple reflections - ripple in the frequency response, depending on the severity of the mismatch. Reflected signals (reflected signals bouncing forth an back) cause echoes or ghosts in video, depending on signal travel time in the cable - which depends on the cable length. 50/75 Ohm cable joints attenuate the signal by 0.177 dB only (96% transferred power, 4% reflected power) so this attenuation is normally not an issue.
For typical short cables on the workbench (few feet) echoes and "mismatch" issues are hardly noticeable as for short lines the load’s impedance dominates the circuit and the cable's characteristic impedance is of little consequence to the circuit’s behavior. Nethertheless decent cabling is always highly recommended. Lot of crap cables exist.

Bottom line: IMO these comets are tape or VCR issues, and I wouldn't rule out TBC glitches of the VCR.

"Bandwidth clipping" in the digital domain (e.g. peak bitrate limits or capping of the data transfer rate) produces its own effects/issues, but I don't think these will ever manifest as comets either.