720x480 is the standard for Standard Definition NTSC signal sampling.
That's 720 samples per horizontal line.
That's 480 samples per vertical line.
When Analog signals for video are measured they are defined by the number of vertical lines side by side which can be distinguished from one another.
That is if you took a bunch of vertical "bamboo sticks" or "straws" and stood them up on end and lined them up shoulder to shoulder next to one another.. then stood way back from them. How many could you stand up side by side, shoulder to shoulder, next to one another before they appeared to "blur" together and you could no longer distinguish them from one another?
If you have only Five and spread them evenly, again shoulder to shoulder, so there were spaces evenly spaced between them.. you would have a better chance of seeing you have "five" from far away.
But as you stack more and more side by side.. decreasing that even space between them.. they crowd together.. until visually they seem to "blur".
These are called "Vertical Lines of Resolution" or how dense can you stack a forest of Trees (or straws) until its meaningless.
A VHS tape typically can produce a signal with 320 lines of vertical resolution, a Broadcast signal can produce something closer to 500 lines of vertical resolution.
With a Digital sampling of 720 points along a horizontal rod laid across all of those straws, means your over sampling by a factor of 2 samples per each straw.. which can have issues of aliasing.. or interference patterns.. but generally smoothing and anti-aliasing can compensate for this interference.
So basically in a perfect scenario 320 samples per digital horizontal lines is enough for a VHS tape.
DV takes a grid of 4:1:1 or 4 luma points along a horizontal line per 1 color sample on that same horizontal, and 1 color sample per vertical field line
DV is standard video only, so it samples on the field not the frame so its sampling over a tvdl 320x240 with a sample grid of 720x240.
If you convert this into progressive (assuming fields into frames) before digital sampling this becomes
720x480 over 320x480
Scaling 4:1:1 and reducing for redundancy means 2:1/2:1
That means for a 720 dot line, two samples per Luma, and 1 sample per two chroma, the third is 1:1 so it is loss less.
The actual loss depends upon the signal resolution being near perfect at 320 which includes (overscan) normally not seen because of CRT bezels and thus normally avoided in televised or recorded SD content.
And if not using SP speed but LP or EP speed it can be even worse.
It should also be realized the NTSC signal saved on VHS tape is recorded as a Color Under signal and reproduced from that Compression scheme.. meaning it is already loosing 1/2 the color horizontal dimension.. so 2:1:1 .. it is possible the claims of 320 Chroma tvdl may actually be claimed, but its more likely that is for Luma only.. and the Chroma is actually 160 in a best case scenario.
It may "seem" like a reduction in color sampling from analog to digital, but in reality it is over sampling only the Luma and sampling the Chroma in a 1:1 ratio for available signal from a VHS source.
Sampling a digital conversion from a higher resolution source such as a Betamax Composite or Broadcast signal may reach for up to 500 lines of tvdl on paper.. in a studio.. but in real world scenarios over losses over a transmission line or due to broadcast and reception on less than perfect equipment will bring that down substantially. S-VHS, S-Video and EP speeds may claim to capture more of the signal on tape, but cumulative losses are likely to claim some of that resolution.. some of which must be sacrificed to Timebase Correction and Frame Sync corrective actions from older tapes.
With Color Under NTSC transmission;
If 4 is the Luma for an analog signal and is 320 and we are over sampling by x2 for 720 we have the potential for aliasing and the potential for using the extra information for slight sharpening.
If 1 is the Chroma for an analog signal and it is actually 160 from the Analog signal and its over sampling by x2 - "nothing is being lost" - the capture resolution is actually 1 to 1.
If 1 is the Chroma for an analog signal per "field" and not frame because it is DV and only works on an interlaced signal - "nothing is being lost".
4:2:0 makes better sense when converting a real 4:4:4 situation and favors the vertical dimension, but more over it compresses pscyhological teasing out intraframe compression opportunities, but suffers from Macro blocking artifacts.
Mosquito noise is an artifact of condensing or composing a Progressive frame from two Interlaced fields with a separation in the temporal dimension and get worse with high motion, or low motion with sharp edges, which is where it is most often observed.
MPEG-2 or h.262 introduced Macro blocking and inter-frame and well as progressive compression opportunities with different complexity profiles.
MPEG4 part 10 AVC or h.264 introduced golden ratio spiral or more complex search patterns of psycho aural visual inter-frame as well as progressive compression opportunities, again with different complexity profiles.
DV was appropriate for its time, but of very low compression advantage, while retaining the best picture available from signals of its time. It would never be appropriate for HDTV or HD signals of today.
But it gets maligned quite often as an inferior capture format, when in reality, considering the signal source, its more than adequate. It was also widely adopted by many of the operating systems of the day and remains a simple format to decode and present.
MPEG-2 was more appropriate for studio film conversions to digital and storage on DVD playback media, it was a little over hyped for its time, but over delivered for systems not ready yet to deal with the format, it was anticipated to deal with the better quality signals available from Component or highspeed Cable or Fiber networks, leaving the SD and S-Video / S-VHS era behind.
Microsoft came up with their own version in VC-1 to circumvent licensing issues, and partially due to those licensing issues Apple chose to sponsor and develop the h.264 MOV/MP4 standards. This bifurcation led to a lot of market confusion. Video format wars of the VHS/Betamax/Laserdisk or Blu-ray/HDVD formats did not help.
The format wars insured that the OS vendors would eventually opt out and natively include neither.. but marginally continue to support DV and the more open h.264/MP4 standards which were not license encumbered.
Microsoft stayed with VC-1, and optionally bundled licensed MPEG-2 codecs and decoders for a fee. Microsoft formerly abandoned the nonlinear editor business after offering Windows Media Encoder 9 and Windows Expression Encoder 3.
Apple went with h.264 MOV/MP4 for free but also offered Professional addon codecs with nonlinear editor suites like Final Cut X.
Microsoft in later versions of windows began including h.264/mp4 as a natively supported codec with little fan fair.. as it offered no competitive advantage.. and withdrew the limited MPEG-2 codec support by withdrawing Windows Media Center from the market. Ceding most nonlinear editor business to Adobe/Premiere and other companies.. such a Grass Valley/EDIUS or Blackmagic/Media Express, AJA/Studio, (former SONY) Vegas .. among others.
