Sounds like you should have the basics, except for experience with analog circuits and analog design. You can find simple designs for analog buffers with level adjustment in the schematics for VCRs. The following comments are offered more for the benefit of other readers.

Composite video signal does not have to be bad. It is worth noting that the video terminal output of many professional analog camcorders was a composite signal. High end signal separators in monitors and other gear used comb filters to derive Y and C effectively. But those filters added cost. Often in home/consumer/modest cost equipment the separation was accomplished simply (and cheaply) by cutoff filtering at around 3 mHz with below 3 mHz for Y and above for C. These limits are inline with VHS recording capability and account in part for lack of S-VIDEO outputs on VHS gear. However, they are well below broadcast signal capability. (a bit over 4mHz for the NTSC Y signal component) and S-VHS, DVDs, STBs, and DV/MiniDV thus the addition of S-VIDEO outputs. Feeding a composite output from quality sources can look poor compared to S-VIDEO or component.

While digital is relatively tolerant of frequency response and linearity variations in the signal path, analog is not, A high frequency roll-off causes a loss in sharpness/resolution and non-linear amps cause gray scale distortions. That is where the multiburst, cable sweep and ramp signals come into play in testing/alignment.

Matching impedance is necessary to ensure video levels are correct (whites and blacks) are at the proper levels at the display. Within limits this can be compensated for with contrast and brightness controls on the viewing set but that becomes application specific and can reduce cross component compatibility of the recorded signal.

Very interesting thanks!

Back to the modding of an older VCR with no S-Video output for a second.

So basically any VCR is going to have the Y and C signals at some point on the circuit board. The issue is going to lie in what voltage levels they're at as they may be tuned for something like the RF modulator specification, rather than composite (like what I believe is happening in the N1700 example above) and also that tapping in to them may cause issues with the existing outputs on the board (as we may load them too much causing the signal to droop).

Aren't both these issues solved with a high impedance op-amp with adjustable gain? Like if you have a rail to rail op-amp with a high impedance input, couldn't you just run it at 1V, then play a tape that maxes out Y/C and adjust your op-amp until you get your desired peak to peak values? It would mean you could add s-video to any VCR and keep the VCR itself running as if it wasn't modified.

Quick edit: I mean an Op-Amp circuit for both signals.

I assume there is more to this, otherwise someone would be selling this already.

These games are, of course, nice, but I wouldn't want to digitize anything from any Son of Frankenstein unless I'm absolutely sure that everything is as it should be. It's not that simple - find a connection point, get a signal out, and then enjoy! This is why my NV-W1 still has only composite out.

VHS VCR's store the signal on tape as a form of Y/C. However, the C signal, which (for NTSC) is suppressed carrier amplitude and phase modulated centered on 3.579545 mHz, for VHS tape it is down converted to center on about 650 kHz, HiFi audio is recorded as FM at 1.3 mHz and 1.7 mHz and Y is recorded as FM above all this. S-VHS uses the same C and HiFi audio recording method but shifts the Y band to higher frequencies to give more resolution. For output the C is up-converted by the VCR back to 3.57 mHz

Thus you should be able to pull find a Y and a C signal just before an adding circuit feeding the composite output. The levels and ability to drive an external circuit will depend on the details you your source point.

An OpAmp is a workable approach. You will need flat frequency response out to about 6 mHz and the ability to drive a 75 ohm load at 1 volt peak to peak. And since they operate closed-loop and will be at relatively low gain be be sure of stability.