Trying to Understand Capture Card Choices

[trhouse]

Here is Macrovisions original patent,

http://patft.uspto.gov/netacgi/nph-Parser?Sect1=PTO1&Sect2=HITOFF&d=PALL&p=1&u=/netaht...&RS=PN/4631603

Granted Dec. 1986 so it expired after 17 years Dec. 2003 presumably.

[edit]

USPTO now has in force a change in the term of patents. The later of 17 years from issuance or 20 years from first filing. The Macrovision patent was filed April 17, 1985 so it expired April 17, 2005.

[edDV]

Ok. Here we go

@ edDV

The Absolute Capture Setup ...

If what you say is true, in your last two paragraphs, then that explains
a lot more (at least to me) You see, I'm under the impression that all
TV broadcasts are (at least these days) MPEG, which in my understanding
is 4:2:0 in the end. And this same 4:2:0 is YUV color space.

I was talking about the way video is acquired, edited and presented to the transmission system in a typical TV station. To summarize:

- most local camcorders these days are DVCPro or DVCAM (both 4:1:1 DV)
- some acquisition and editing is done in Digital Betacam (4:2:2, 2x compressed)
- analog tape material is transcoded to SDI serial digital 4:2:2
- tape material for programs and ads usually arrive in Digital Betacam.
- network satellites output analog NTSC or SDI (serial 4:2:2 at SDTV or HDTV resolutions). SDTV material is transmitted directly or recorded to Digital Betacam or to a 4:2:2 server. HDTV material is usually not recorded, it is just broadcast directly from the satellite.

So, SDTV material that goes to air is a mixture of raw 4:2:2 and 4:1:1 digital.

transmission to home looks something like this

transmission ----> analog transmitter -> analog NTSC *
server -----------> digital transmitter ->ATSC DTV

Ignoring DTV for now, the home receives NTSC over the air or via analog cable to the analog tuner (TV, cable box, PC tuner, VCR or DVD Recorder) where it is decoded.

* Regardless of upstream wideband digital sources, NTSC is NTSC so that means it contains a maximum of 4.2 MHz luminance and modulated chroma C that has at most 600-1000KHz of U and V (aka I and Q) bandwidth. Unless Y and C are split with a sophisticated comb or notch filter, only around 3MHz (240lines) of luminance resolution Y is decoded by a typical capture device. Typical chroma bandwidth is 600KHz per component.

3D comb filters are just beginning to be available in tuner/capture cards (e.g. ATI Elite). These can produce up to 4.2MHz luminance and up to 1000KHz U and V bandwidths (336 horizontal lines Y and 80 lines U and V)

Cable boxes usually produce only slightly better decodes than capture cards but mileage varies. Newer cable boxes have better filters and will usually provide a better NTSC decode over the S-Video connection than a typical PC tuner can do on its own. The S-video picture is almost always better for the MPeg2 "digital" cable channels from the cable box.

Bottom Line:
If you are capturing from a NTSC source, the most you have to work with is 3-4.2MHz luminance and 600-1000KHz chroma. After decoding, the result is just baseband analog YUV. There is no remaining footprint of digtal 4:2:2, 4:1:1 or 4:2:0 all of which were much wider in bandwidth. What you have here is the equivalent of 2 : 0.3 : 0.3 in relative bandwidth. For a PC NTSC capture, this is your starting point.

Cable digital channels (MPeg2) and DBS also fall into this bandwidth range at the S-Video (Y/C) connector. In these cases, the actual NTSC decode was done at the cable/sat uplink end with presumably a high quality decoder or the source was digital. MPeg2 is sent to the cable box in YUV components. YUV is then encoded to Y/C where U and V are modulated around a 3.58MHz subcarrier to make C in the same way as done for NTSC. The difference is Y is kept separate which is good. Again it doesn't matter whether the MPeg2 was 4:1:1 or 4:2:0 or some other scheme. Once chroma components are modulated into C you are back to baseband analog 600-1000KHz chroma after decode or at best 3 : 0.7 : 0.7.

Bottom bottom line :
Decoded NTSC is typically 2 : 0.3 : 0.3 and that is the best you can expect in capture bandwidth. Oversampling won't help.