50 years of Test Card F

That's the quid pro quo for not having 'proper' PAL decoding

You get half the theoretical 'perfect PAL-S' resolution (which is the same as 'real' NTSC would deliver at a 4.43MHz subcarrier freq and 625 line rate), but the PAL-D decoder itself averages consecutive lines (that's how the phase errors are cancelled), so you get a theoretical reduction of vertical resolution with a PAL-D decoder too, though not as much as the half of the Sony 'Not-quite-PAL' technique (which presumably also introduced some aliasing in the vertical chroma domain?)

Isn't it the same with half the vertical colour resolution with digital when we're using 4:2:0 colour subsampling, or am I getting that wrong?

Yes - 4:2:0 has half vertical resolution chroma, but it achieves it by vertical sub-sample filtering, not discarding alternate lines! (Or it should be)

The Sony decoder simply discarded alternate lines of chroma, and repeated the previous, and would have given you nice jagged alias edges I suspect - if the CRTs had been sharp enough for you to see them...

I think the answer lies in your last sentence. From memory, in the days of CRTs driven by colour difference signals, you could turn down the brightness of the telly (it might have been contrast) and on the screen you were effectively left with the displayed chroma signal. It really showed up the lack of pretty much any definition in the colour signals which is no surprise really. Quite an 'other worldly, ghostly effect'.

To throw a wobbler in! We had a fancy zone plate generator to push variable 'concentric circles' through some of the analogue kit and look at the results. Not sure it really told us anything really, we'd switch filters in and out, push it through a DVE etc. I think the engineering department had a bit of cash left over one year so we bought one, more of a lab tool I suspect than a day to day production engineering need!

You can do that with digital video too (I've done it in VLC) and the effects are pretty much the same. In fact on quite a few things I notice the chroma only signal sometimes isn't properly interlaced (so it looks filmised or whatever), but added to the 50i luma image you can't tell it apart from sometimes if there's some very saturated colours and fast movement together (I've noticed this a bit on some news reports over the last few years). It does show, how bizzarely little colour information there actually is in a picture.

It would have shown how horrible a lot of DVE processing was back in the day... Lots of aliasing...

I've played with one - you don't want to push it into domestic displays with non-1920x1080 panels (or those that have not have overscan disabled), and it shows the limitations of certain compression codecs quite effectively as well (particularly those that subsample)

Yep - there is definitely a 'chroma bug' knocking around in some processing. I've seen video where the chroma has been treated as 25p, but the luminance as 50i - so saturated red and blue content looks juddery, but less saturated stuff looks nice and fluid...

Some ffmpeg processing used to do with 4:2:2 1080/50i sources deinterlaced to 1080/50p.

The stuff I digitised from my VHS/Video 8 collections ended up like that too, I presume it must have been how the capture card/software works. As I said though you don't even notice it 99% of the time. And I'm not going back and re-digitising it all either, it took me months.

How did you digitise it? Was the capture card also doing a hardware MPEG2 or H264 compression process, or was it captured uncompressed with a software compression ? I assume it was kept 576i?

I'm not sure exactly how the card (well, USB dongle) captures video, but I captured through VirtualDub into DV format using, I think, the ffdshow encoder. It seemed to be the best solution I could get to work with it with all the software I had- the software that came with it only seemed to have a maximum capture bitrate of MPEG-2 at something like 8mbps (a limitation of the software, not the hardware) and other software I had didn't seem to capture the video properly.

The DV files that came out of it were pretty much artifact free, so I presume the card captured either uncompressed, or at a very high bitate.

And yes, the footage was 720x576 50i all the way.

The stuff I transferred from my MiniDV camcorder doesn't have the chroma 25p issue, but then I didn't use the capture card for that, I imported it directly from the tape via Adobe Premiere so there wasn't any re-encoding or anything.

The capture card comes up as "USB 2681 device" in the capture software, if that gives you any clues about it. I bought it of Amazon back in 2012.