Capture/reconstruct video from RF signal

[swiego]

I’ve found a few references to work done capturing a raw RF signal of a magnetic tape then using software to reconstruct sync, scan lines and frames, chroma and luma info etc. but what’s out there seems proto / conceptual for the most part.

A company claims to have a technology set with Betacam as their donor platform, but their site is unclear as to maturity. I haven’t contacted them, myself. https://www.cube-tec.com/en/solutions/video/quadriga-video/quadriga-video-rf/analog

I also found a short clip on YouTube that’s tantalizing but no details were provided. https://youtu.be/t0eLxQY6-RI

I’m starting this thread mostly with the intention of collating information about the current state of the art on this topic. The above two leads are the only ones I’ve been able to find myself.... I’m hoping there’s more out there! Also hoping for some discussion about the long term efficacy, limitations of this set of techniques for signal acquisition and image reconstruction.

[Cornucopia]

I don't understand what you intend to recover that cannot be recovered by simply playing back the tape(s) on appropriate & maintained decks and capturing via the most straightforward, highest quality format given the deck's output options.
RF capturing requires much higher bandwidth (aka higher samplerate) than standard baseband video. Supporting this capability is certainly not economical.

Or are you just trying to see if it can be done from a scientific, theoretical standpoint?

Scott

[johnmeyer]

This reminds me of the efforts to play phonograph records by trying to reconstruct the audio from reading a waveform created by a laser beam reflected off the grooves. It can be done, but even the modern efforts, using the CPU/GPU powers of modern computing, fail to fully reconstruct the audio, and the resulting audio is inferior to what can be achieved with a good cartridge and stylus.

You didn't make it clear if you are trying to use a signal that comes from a standard helical scan tape mechanism, but I assume that you are, since I don't think you can get the signal off the tape any other way (i.e., there isn't a "laser beam pickup" alternative).

Perhaps you could provide a simple block diagram of a typical VHS circuit (they are easily available on the Internet) and show where in that signal flow you are trying to read the signal.

[pandy]

I don't understand what you intend to recover that cannot be recovered by simply playing back the tape(s) on appropriate & maintained decks and capturing via the most straightforward, highest quality format given the deck's output options.
RF capturing requires much higher bandwidth (aka higher samplerate) than standard baseband video. Supporting this capability is certainly not economical.

Or are you just trying to see if it can be done from a scientific, theoretical standpoint?

Scott

RF approach require lower bandwidth (RF video is usually bellow 4MHz thus doubling this give us 8MHz an ADC resolution 8 bit is sufficient as signal is FM modulated) - standard baseband require approximately 27MB/s vs 8 - 10MB/s on RF. Also demodulating FM and all additional steps performed in software allow for highest precision. IMHO even best VCR can't provide this kind of quality.

This reminds me of the efforts to play phonograph records by trying to reconstruct the audio from reading a waveform created by a laser beam reflected off the grooves. It can be done, but even the modern efforts, using the CPU/GPU powers of modern computing, fail to fully reconstruct the audio, and the resulting audio is inferior to what can be achieved with a good cartridge and stylus.

You didn't make it clear if you are trying to use a signal that comes from a standard helical scan tape mechanism, but I assume that you are, since I don't think you can get the signal off the tape any other way (i.e., there isn't a "laser beam pickup" alternative).

Perhaps you could provide a simple block diagram of a typical VHS circuit (they are easily available on the Internet) and show where in that signal flow you are trying to read the signal.

Issue for analogue laser turntable was dirt and surface not tracking precision superior to any mechanical approach.
Side to this novel approach with 2d high resolution photography allow to read whole plate at once.
both methods are superior to mechanical stylus as they are not damaging surface of plate.
For clean groove mechanical approach is less accurate.
There is many possibilities to read tape without head - but usually it is not necessary and head signal is easily available in VCR.
RF FM signal from head amplifier is routed to further processing - seem this is best place to start DSP.
Crucial can be combining FM head signal with servo/head sync pulses thus being able to compensate unavoidable mechanic issues.

Thinking on this from very long time ADC USB3 based acquisition board, perhaps RF (or rather 5 - 8MHz bandwidth 8 bit acquisition i.e. 16MHz sampling rate plus additional digital channel for head servo plus 2 audio 20kHz 16 bit channels for classic linear audio) all signal simultaneously sampled thus perfect synchronization provided. Core signal processing based around GNU Radio

[lingyi]

Most of the this goes over my head, but the company in the OP's first link seems to be focused on selling their proprietary hardware (internal and external) with the only internal modification being to the video output stage with an internal board. Interestingly, at least that I could fine, there's no videos or even screencaps that compare their product's output to a conventional one.

[JVRaines]

RF approach require lower bandwidth (RF video is usually bellow 4MHz thus doubling this give us 8MHz an ADC resolution 8 bit is sufficient as signal is FM modulated) - standard baseband require approximately 27MB/s vs 8 - 10MB/s on RF.

Exactly right. An analog-to-digital converter working on a standard NTSC signal is already running at over 13 MHz; sampling the FM output from videotape, which lies chiefly in the 1–5 MHz range, is no problem at all. Many professional players already have RF outputs for dubbing. The work lies in demodulating the signal to recover brightness/color, blanking, and sync. Assuming the player tracks reasonably well, it would even be possible to perform timebase correction in software.

[pandy]

RF approach require lower bandwidth (RF video is usually bellow 4MHz thus doubling this give us 8MHz an ADC resolution 8 bit is sufficient as signal is FM modulated) - standard baseband require approximately 27MB/s vs 8 - 10MB/s on RF.

Exactly right. An analog-to-digital converter working on a standard NTSC signal is already running at over 13 MHz; sampling the FM output from videotape, which lies chiefly in the 1–5 MHz range, is no problem at all. Many professional players already have RF outputs for dubbing. The work lies in demodulating the signal to recover brightness/color, blanking, and sync. Assuming the player tracks reasonably well, it would even be possible to perform timebase correction in software.

Even with poor tracking overall result with sufficient oversampling should be better when compared to classical analog steps. TBC can be performed as servo head signal together with video sync pulses may provide more accurate timing. Color decoding is less problematic as it is quite well documented.

[lingyi]

From my layman's minimal understanding, what's being discussed is practical theory of what could be done. Could anyone address what they think is being done in the link given in the OP's first post https://www.cube-tec.com/en/solutions/video/quadriga-video/quadriga-video-rf/analog, which consists of replacement of an internal board and output to a proprietary digitizer:

"Simple but efficient

In order to modify simply open your machine and remove the DEC-42 board, if you are modifying a recorder. This is not needed for playback. In pure player devices the slot next to the DM-56 board will be already empty.

The DM-56 board has to be taken out and modified.

You only have to solder the supplied cables onto the Y/C demodulator board in order to directly tap the off-tape RF signals.

The next step is to open the back of the machine and solder the second set of supplied cables to two existing BNC connectors. For example, two of the composite outputs may be used to serve as outputs for the direct RF read out after modification.

Place the modified Y/C demodulator in its allocated space.

Than connect the QUADRIGA•Video RF Amplifier board and place it in the free slot next to the DM-56 board.

Now your device is ready to be connected to the external QUADRIGA•Video RF Digitizer. You are now able to digitize with 14 bit for a maximum of signal quality."

As has been stated, "Many professional players already have RF outputs for dubbing", so how is is different or superior?

[JVRaines]

Well for one thing, I don't think the Betacam BVW series has a stock RF output. Their custom board must provide some signal conditioning and then gets jury-rigged to a couple of BNC outputs.

[swiego]

Yes, for helical scan my thought was signal acquisition right off the drum head leads. I imagine that in addition to the software challenge of reconstructing image information (which my instinct tells me is solvable) the other challenge is having a shielded path to the aquisition system. Does anyone know what voltage/current levels are found coming right off the head?

[betwixt]

A few mV into a high impedance, that's why the head amplifier is screened and fitted close to the head itself. It's to minimize noise pick-up and raise the video 'RF' level before it gets lost in the background. Using DSP techniques to process the head signal isn't new but substituting digital for analog circuitry doesn't fix the underlying problems of timing and frequency response because the same head mechanism, which accounts for almost all the signal degradation, still has to be used. There is some merit in performing timebase correction on the signal before it has gone through the normal VCR electronics instead of on the processed output but to be honest I doubt most people would notice the difference.

Brian.

[pandy]

Yes, for helical scan my thought was signal acquisition right off the drum head leads. I imagine that in addition to the software challenge of reconstructing image information (which my instinct tells me is solvable) the other challenge is having a shielded path to the aquisition system. Does anyone know what voltage/current levels are found coming right off the head?

Depends on VCR age - modern VCR's are usually equipped with "active" head - pre-amplifier is built in head drum and on VCR mainboard FM signal is around few hundreds mV, for example service manual for "passive" head Philips VR6443 shows oscillograms where at point TP203 playback signal from preamplifier BA7252S is around 150 - 200mV and formed to square wave head switching signal is 5V.
BA7252S datasheet describe head signal level at around 0.15 to 0.6mV i.e. 150 to 600uV - not so bad (usually MC pickup or tape audio magnetic head signal is bellow single uV around few hundred nV), gain is around 55dB i.e. few hundred (up to 1000) times - no problem for modern OPAMP's. But there is no need to build amplifier if VCR's amp can be used nicely so ADC must be able to deal with signals around 200mV and bandwidth maximum 10MHz - easy.
Checking datasheet seem in modern VCR's signal was sampled anyway - they began to use CCD as analog memory thus moving signal processing toward digital domain as another level of evolution. And looking on some problems with SECAM and uncommon video standards quite justified (not economically).

A few mV into a high impedance, that's why the head amplifier is screened and fitted close to the head itself. It's to minimize noise pick-up and raise the video 'RF' level before it gets lost in the background. Using DSP techniques to process the head signal isn't new but substituting digital for analog circuitry doesn't fix the underlying problems of timing and frequency response because the same head mechanism, which accounts for almost all the signal degradation, still has to be used. There is some merit in performing timebase correction on the signal before it has gone through the normal VCR electronics instead of on the processed output but to be honest I doubt most people would notice the difference.

Brian.

Partially agree, moving DSP to low level allow to capture even deteriorated tapes which are lost for normal approach but still, if head can rotate and read "something" then you can capture this and perform various strategies in digital domain (like dropout compensation). I'm not saying this is easy and i'm not saying this is economically justified to be consumer standard procedure but when you can digitize RF FM playback signal then later you can work with this signal as many times as you wish.