Michael Krupnick
Forum Replies Created
-
Yes, Gary…I ferreted around and got some info on the math. As you knew off the top of your head, but I needed to be reminded of, HDCAM is1440H EXTRAPOLATED to 1920. It still seems a bit fishy to me, math-wise, that 1280H is not more distorted, but, then when you add the half-sampled chroma, you get 1920. That raises another question (not really important, I s`pose) as to what really counts as a pixel. In NTSC, pixels are luma-only, with the color added on top. Digital sampling is a different story, apparently. According to esteemed colleagues, TRUE 1920X1080 as a displayed picture from the digital chain doesn’t happen anymore – the last time was when HD was analog. To me it’s kinda sad that we’re lowering the quality bar despite the technical capability to raise it. But then, I like Showscan film way over Panavision, which I know is futile because it is prohibitively expensive. And there are alternatives, like Dalsa, that really DO seek to make digital exceptionally fine.
We transcoded the HDCAM footage to DVCProHD to get it to play reliably on the XServ SAN we’re using at my client BMP. You probably already know XServ isn’t yet able to do uncompressed HD multiple streams for large installations. They have many many seats using the fiber net, in mixed resolutions. That’s their interest in that experiment, since they are looking forward to HD delivered to their network clients.
My interest is that I’m looking to expand into more HD on a boutique level. HDV is OK in theory for consumer use, but I worry that the MPEG transport stream won’t hold up to critical professional scrutiny and workflow. I’m looking at the Panny HVX200 as a pretty good way to bridge the gap between HDV and full-on pro-grade HD, both in terms of quality and lower storage overhead. The key to that, in my head, is that DVCProHD codec. It seems to be a reasonable compromise. My colleagues worry about the P2 card storage limits, but I think it’s probably a safe bet that P2 technology will get fatter and cheaper pretty quickly. There are already viable work-arounds like Firestore drives. And the promise is that the 200 will have the widest range of frame/aspect/rate capture options of anything near the price. That’s probably important for tapping the indy film market. I wish I could get a Viper, a D6, a beautiful hunk of glass in front of it all, and a huge pile of Ultra 320 behind it, but…
My eyes tell me to go for the best, but my wallet…well, I’m sure you know. Alas.Cheers.
m.MixMasters/PixelLabs
https://mkrupnick.go.cc -
Hey Gary, chill, will ya? When you say “broadcast”, that IS about transmission, in this country. You work in 720p. So do I. I like it, too. But that wasn’t the subject of my query. I was asking about 1080i. A simple question it was, too: 1280 vs 1920 at 1080 INTERLACE. I’m not advocating one over the other or asking about marketing, just math. But you seem to be wanting to go everywhere else but there. I’m not making an argument – YOU are. I’m not talking down any format; I was just trying to get an answer or opinion on a simple specific question, which you obviously are unwilling to focus on. The idea was to stimulate a discussion on the discrepancy in pixel count on a line in what should be identical rasters, get it? But since this subject is political rather than technical for you, never mind. I’m sorry this upsets you so much. Thread CLOSED.
-
You said:
All HDTV broadcast video signals are 1.5 GBs not just 1080.Not correct, Gary. Only the baseband signals carried through wires are that high. NOBODY broadcasts a signal like that. While at CBS network operating their first HD post bay, I learned alot about about bandwidth limits and compression. When you speak of BROADCAST, you are talking about MPEG2 streams of only 19MB. That’s a lot more compression than anything going to tape (except HDV) , as it must be to squeeze onto transmitters.
If you examine the codec tables for HD in FCP, you will see two different widths for 1080: 1920 and 1280. Forget 720, forget HDV; and think about your fourth-grade geometry class. THAT’S what I am talking about. DVCProHD (100) 1080 qualifies and bringing in other formats only confuses the issue. I guess very few people have actually looked into the 1080 variants and just make general assumptions as to what HD is limited to. I have been discussing only one raster – you keep detouring to another. I think you’re missing my point.
Some interesting reading on that subject is available at https://www.atd.net/HDTV_faq.html and even includes some mention of your beloved 720 line raster. The article is from DoD, NASA, DARPA, and NIST, who, unlike manufacturers, have no interest in selling a format but doing an understanding evaluation of the true nature of these beasts. It’s indeed very interesting how communications pros like us can get so confused talking about this stuff, ain’t it? In a time when the technology is popularizing ever more complex tools, the facts still seem to conspire to limited comprehension.
Cheers,
m. -
Jeez, Gary; I’m sorry I called you Andy. That’s waht I get for talking on a cell while driving and posting -hee hee-
-
Hi Andy. I stand corrected on HDCAM sampling, but actually, DVCProHD IS spec’d at 4:2:2 sampling. While I understand what you’re trying to say by example, I think it is somewhat misleading to equate anamorphic optical processing with data compression. People might think there is some geometric distortion, and we both know that’s not really what we’re talking about. And HDV wasn’t even on the table as a subject…bringing that up is only adding to the ambiguity of the situation. There is already alot of confusion in HD, much of which has happened because of improper analogies or sometimes even misinformation about the basics. Almost all popular tape formats are compressed to accomodate the huge amount of data in HD. Baseband 1080i/59.94 HD is 1.5 GB/second, enough to make the signal cable warm on a live switcher trunk. I just wish all the manufacturers would agree on a common benchmark terminology to define their specs instead of juicing up the figures for their marketing. Engineering and production benefit when everyone is on the same page. I’ve had some experience with trying to deal with that aspect of HD “salesmanship” vs objectivity, and the headaches it caused were migraines, for sure. The crux of my problem in this thread came from a relatively simple math dilemma: two codeccs both claiming to 1080 lines high and 16X9 aspect had different horizontal dimensions (1920 and 1280). That is not possible even in elementary-school math. The reason that occurred is because Sony and Panasonic differ on what they call a pixel. I was hoping after all these years of television there was agreement on that basic fact by now. That’s the important point for me. Again, thanks, Andy.
-
Gary (and JBB):
Thanks. I fiddled around with the settings and finally got something that worked. But my head aches. My assumption was that 4:2:2 samples were counted the same as pixels by Sony and Panasonic…but apparently they’re not identical. Nomenclature is the confusion for me, apparently. And the fact that these numbers do not (readily) appear in publications of SMPTE or ATSC doesn’t add clarity, either. I naiively just figured that all pixels were equal. When does 1920=1280? When marketing and engineers conspire -hee hee-.
Cheers,
m -
Have you tried dropping a de-interlace on the slo-mo’d footage?