HD and special effects (1 Viewer)

[ChrisWiggles]

GAAAAAH. Talk about bad information, this is NOT the case at *ALL*. This is PROFOUNDLY incorrect, I can't even begin to fathom where this stuff comes from. HD is *NOT* film, or intended to be a replacement for film content.

As has been alluded to already, many commercial 35mm prints are poor for various reasons, they are many generations removed from the original, they are handled poorly, run through poor machines, dirty, scratched etc etc. An average 35mm print in your average movie theater probably isn't going to be as good as a prisine 1080p transfer, however a really good 35mm presentation will still probably be significantly better. Things like 70mm, Imax things like that are going to be yet another giant leap in MTF capabilities that HD resolutions still won't touch.

Digital Cinema is definitely moving beyond 2K, and most scans nowadays for HD transfers and archives and things like that are done at 4K.

This list of "Facts" is pretty much totally erroneous, and I don't know where these ideas come from, but HD is not at all a replacement for film.

You also have to realize that there is more than sheer resolution in play. HD content is nonlinear 8-bit 4:2:0, and heavily MPEG compressed. D-cinema will be of significantly higher bit-depths, full 4:4:4 uncompressed, and perhaps even linearly coded.

Film that's scanned at 4K, 12-bit 4:4:4 is something like 10 Terabytes for a tw-hour film. This is orders of magnitude greater in both resolution, sampling, and bit-depth capability than something like 20gb for a two hour film in HD.

Please do not confuse D-Cinema capabilities with HD. Two, VERY different beasts.

 

[ChristopherDAC]

Heh. My data are from the NHK book, from the original papers by Fujio et alii in the SMPTE Journal, the IEEE Transactions, NHK Laboratories Note, &c. ATSC broadcast HD is all that. High Definition itself is not. If the ATSC broadcast standard falls short of the full capability of the High Definition concept, in the same way that ordinary theatrical projection falls short of the possibilities of 35mm film, it is probably to a less extent.
Read the original sources for an idea of what HiVision was supposed to be, and the reasons behind the selection of 1125 lines, 30 MHz baseband width, and all the other specifications. It is explicitly stated that the intent is to surpass the image quality [embracing not only resolution, but also motion rendering, colour gamut, and other factors] of 35mm film as it is ordinarily presented, and to provide a production medium of sufficient quality to replace 35mm film.
Here in the US we never really made any effort to indentify the desirable characteristics of the next-generation television format. The Japanese did, and that's why the first Japanese HDTV sets went on the market in 1979.

 

[ChrisWiggles]

Well, you need to make it clear that you are talking about non-existent content that is unavailable to consumers, and not what everyone else considers "high definition content."

I think it's reasonable to consider the following HD content:

OTA Broadcast HD (720 and 1080)
Cable and Satellite HD broadcast
D-Theater
WMV-HD content
VC-1 and H.264 HD-DVD or BRD

The examples you seem to want to include under the umbrella of "HD" include D-5 tape and higher, and D-5 is still a long long way from D-Cinema stuff, and D-5 content is hardly available to consumers.

If that's what you want to say, you should be VERY VERY crystal clear what you are defining as "High Definition." Personally, I think that definition is ridiculous, and hideously misleading to just enter into a conversation where everyone rightly assumes HD content at 720 and 1080 resolutions, compressed, 8-bit, etc, and make broad statements that "HD is designed to supplant film." Clearly, the HD being discussed here is not at ALL designed to supplant film. It's almost akin to including UHDTV in this thread without saying that that's what you're talking about. Grossly misleading to say the least, since is the HT software forum, not the "nonexistent, unavailable to consumers future formats" forum.

 

[ChristopherDAC]

I'm not talking about some mystical "content unavailable to consumers". I'm talking about the studio standard. It should be obvious that, when motion-picture production is the matter at hand, the studio standard is what is under discussion, not the various approximations thereunto which have been used for broadcast purposes.
So what if ATSC uses 8-bit nonlinear sampling, while the studio standard calls for 10-bit linear? So what if ATSC doesn't even try to follow the constant-luminance principle, unlike even the Japanese analog MUSE format [which encountered technical problems with true constant-luminance, but implemented what they called "quasi-constant luminance"? I think everyone understands that motion pictures are not going to be produced within the limitations of ATSC broadcast formats, any more than network TV is produced within the limitations of VHS.

Let me define the normative meaning of High Definition, from NHK's original proposals, as adopted by the SMPTE [before certain later emendations].

Frame rate: 30 Hz
Scanning: 2:1 interlace
Scan lines: 1125
Active scan lines: 1035
Aspect ratio: 16:9 [initially 5:3]
Bandwidth: 30 MHz each R G and B
Randwidth: 30 MHz Y, 15 MHz each R-Y B-Y
Signal-noise ratio: 55 dB minimum, weighted

Fundamental sampling rate: 74.25 MHz
Samples per line time: 2200
Active samples per line time: 1920
Sampling word length: 10 bits [gamma-corrected linear]

I can give you the colorimetry equations if you like, they're here in my desk drawer. Suffice to say the colour reproduction is both wider in range and more accurate within its range than any one-strip colour film process.

 

[ChrisWiggles]

I'm not sure what you're getting at here either, but most video does not follow this principle.

Can you provide a specific reference to this format? Because as I mentioned before, it's kind of silly to be discussing irrelevant formats, and based on what you posted here I still think it's not appropriate to consider it a replacement for film, because "1035 scan lines" is less than 1080, which is current consumer HD, and is a far cry from full film capability(idealized).

Further, things like this: "Sampling word length: 10 bits [gamma-corrected linear]" are non-sensical. It's either linear or it's not. Gamma-corrected linear is an oxymoron.

The essence of what you posted was this:
Conclusion? What you see in HD is what the director saw.

Which is flat-out ridiculous. You seemed to say that the HD transfer that we see at home currently (ignoring the NHK stuff for a minute, because this contradicts your vapor-ware "HD") is then transferred to film for distribution, at current consumer HD resolutions and attributes. This is flat wrong. Not even D-5 tapes are close to what the print is scanned at, and I don't see anyone with D-5 content in their homes, let alone something beyond that. One day, hopefully, but certainly not now.

 

[ChrisWiggles]

I just realized something, you aren't referring to SMPTE 240M, are you? :b

 

[ChristopherDAC]

Well, it's a relative.

Some of that was from memory, I admit, and I'm thinking you may be right about the 11 bits. When I say "gamma-corrected linear" I'm just trying to indicate that the samples are linear, and taken from the gamma-corrected, that is to say linearised, analog waveform -- it is possible to sample the uncorrected waveform linearly, after all. If you have a copy of SMPTE 240M, I'd like to see it -- I haven't been able to get it so far. Of course, I'd prefer to have the original as-adopted version, rather than the current amended version, but that's life.

Anyway, I've got a 1" broadcast VTR in my living room, so the distinction between professional and consumer isn't very clear to me. :b

 

[ChrisWiggles]

I don't have 240M that I can distribute, but it's not really that interesting, it's an obsolete HD system adopted way back when. It's been thoroughly supplanted by Rec. 709 (thankfully).

 

[ChristopherDAC]

Ach, I'm a student of the history of technology, in my own small way. If you think SMPTE 240M is bad, you should have a look at ITU-R BO.787, which standardises the dreadful European 1250-50 HD-MAC transmission system. I'll just have to keep hunting, if you can't provide it; perhaps I can get it from the SMPTE Journal?
Of course, I'm of the opinion that even a poor standard is better than no standard, and the ridiculous variety of allowable timings and scan structures under ATSC strikes me as being very much like having no standard at all. I know there are arguments for this state of affairs, but I don't much agree with them.

 

[ChrisWiggles]

ATSC isn't that ridiculous, really. Yeah it's kind of messy, and I wish it were of higher quality, but oh well.

I expect HD-DVD and BRD to be significantly better than Broadcast HD, and hopefully better than D-theater. There is conflicting info about that, but we'll see if the engineers can be crafty enougyh to get the quality up in the face of penny crunchers at the studios.