The firewire misunderstanding

[intomisery]

Let me unload my knowledge. As far as I know, YUY2 capturing full frame video needs 20MBytes/s, while DV requires 25MBytes/s. An avg. 7200 rpm HD cannot keep up with this.
Yet, there are Firewire adapter cards that fit nicely into any PC and claim that you can capture lossless video via the direct digital transfer. How is this possible if this requires 25MB/s?

[Dr Mordrid]

You're confusing megaBYTES with megaBITS.

Generally speaking megabits/8 = megabytes.

YUY2 uses 20 megabytes/s while DV is using 25 megabits/s.

DV's 25 mbits/sec comes out to just about 3.125 mbytes/s plus another tad more for the audio and a total bitrate of about 3.6 mbytes/sec.

I usually make the distinction by using MB/s for megabytes and mb/s or mbits/s when talking megabits. While these aren't official nomenclatures it usually gets the point across.

DV providing lossless editing is totally an urban legend. DV's losslessness extends ONLY to transferring the data to and from the camcorder. When DV is edited and effects, filters or titles added it IS lossy.

While DV is not as lossy as other DCT (discrete cosine transform) based compressions like MPEG or MJPeg, but still it is lossy. As such itt still suffers generational degradations that depend on the specific DV codec used. Yes, and there are several out there. Microsoft, Sony, Canopus, Matrox RT-2000 etc. etc.

The Microsoft MSDV is the worst showing horrid color shifting and vertical striations after just a couple of generations of recompression. The Canopus and RT-2000 DV codecs are among the best showing little or none of these effects out to 10 or more generations.

Another problem is DV's lower colorspace relative to other codecs like MPEG and MJPeg. DV uses 4:1:1 (NTSC)/4:2:0 (PAL) while the others use 4:2:2, a much better setup for doing overlays, bluescreening etc.

Dr. Mordrid



[This message has been edited by Dr Mordrid (edited 08 August 2000).]

[intomisery]

Does this mean that I will get much better overall quality if I use the YUY2 with the Marvel G400 and analog inputs than if I bought a $150 DV firewire card (or perhaps Raptor) and use its firewire inputs (from the same mini-DV camera source)?

[Eric VanHeest]

The video on the digital tape in the camera is already compressed, so you aren't going to improve its quality by converting it back to analog and capturing in YUY. It's possible that your quality might be better if you capture directly from the camera's analog output while recording, since presumable the DV compression isn't in effect there (although my knowledge of DV cameras is paltry and it may not be that way).

Once it's in DV compressed format, though, you might as well not lose any more quality by using analog transfer after that; the firewire card will let you get the lossy video off of the tape without losing anything in the camera->PC transfer.

[intomisery]

ok, Say I'm using a nice 3 grand mini-DV camera. Why would I buy something as expensive as these other firewire editing solutions for professionals vs. the $150 Pyro card or Raptor? Besides the the cool real-time stuff it can do, the quality which is the DV compression will be identical since they are both losslessly transfering the exact same data from the camera, then outputting it back to the camera... correct?

[cjyo~]

Doc,

On this MB/s vs. Mbps question can you answer me question here? :
http://forums.murc.ws/ubb/Forum2/HTML/004614.html

[Walrus]

Actually, if you're going to do bluescreening, overlays, fades and the rest for DV materials for the absolute best quality you need to capture in DV, and then convert to YUV or RGB to do all of the processing, and then render to your output format. This of course requires ludicrous amounts of hard drive space, but it minimizes generational loss.

[Phil Pugh]

intomisery:

The lossless transfer only applies to the moving of the data from the camcorder to the PC and back again. If you process the video on your PC with anything other than simple cuts then portions of your video will be re-rendered. This is where the lossy bit comes in. As mention before (Dr. Mordid) some CODECs are not very good whilst others are more than adequate for normal needs. That's why (amongst other things) a RAPTOR costs a bit more than a PYRO. However - how many times do you capture video, edit it, write it out, re-capture it , re-edit it etc.... Not very often I would guess so you may not notice any artifacts even on the less-good CODECS.

[intomisery]

Ok, so to losslessly put data on my computer from a DV cam, then transfer it back would result in the exact same quality video no matter what card I use. Thank you!

[Brian Ellis]

cjyo

I've explained this in the past. The IT guys totally ignore the ISO standards on units and their abbreviations.

- a bit is an abbreviation for a Binary unIT and should never be re-abbreviated

- a byte is a neologism. b is already much used, so by is a legitimate abbreviation for a byte, although, being short, it could be spelt out in full (plural unnecessary)

- B is a baud (usually 6 bit bytes/min, used in telegraphy data rates). It must NEVER, NEVER, NEVER be used as an abbreviation for bit or byte (capital abbreviations are quasi-uniquely reserved for units named after proper nouns, but the units are not capitalised)

-M, as a prefix, signifies a multiplier of 10^6 or one million

-m, as a prefix, signifies a mulitiplier of 10^-3 or one thousandth

-per or p should never be used in a unit. The correct thing is a slash (/) as a divisor, although, in printed scientific docs, it is preferable to use the negative exponent eg Mbit/s becomes Mbit.s^-1. However, this is no use in e-mailing, where the slash is preferred.

Doc's abbreviations are therefore, IMHO, just perpetuating the confusion. 1 mbit/s is an extremely slow data rate (~3 bits per hour!) and 1 mb/s is not a data rate at all, whereas 1 MB/s represents a telex starting off very fast and accelerating to the speed of light in a few seconds (a baud is already a data rate!).

Now, how do we get the IT industry to understand that units are, by definition, something which is rigid and can't be mucked around with like it does? I don't know, but as someone who is an award-winning freelance technical journalist in his spare (?) time, I do my nut when I see the liberties taken with something that has been internationally agreed for many years. Oh!, and yes, even the USA is a signatory to the ISO conventions on units.


------------------
Brian (the terrible)

PS I bet this kills the thread stone dead

[This message has been edited by Brian Ellis (edited 30 August 2000).]

[peterclifton]

Brian

Thanks for the message, it made me smile as well as reminding me that I'm just as guilty as the next man about using the wrong conventions, I'm not even consistant between postings.

I remember getting just as steamed about the complete bastardization of the RS232 standard.

Peter

[Sillyname]

I'm not familiar (read: comfortable) with "B" being the abbreviation for Baud. In my experience baud is not abbreviated. MB is megabyte. Mb is megabit. As for the headscratching involving "p" for "per", I don't think of that as being an issue. Even schematics in electronics stray from standardized component symbols. And the Chinese read from right to left! What's with that! Commies!

------------------
Deep is not the root word of depression.

[Pertti]

I'm with Sillyname,

The capital B for "byte" is used quite widely throughout the PC world.

Talking about standards, bigger violations are being made by the "Big" names here.
At least I was taught that 1000000=1000k=1M

[Brian Ellis]

I'm sorry, please let me quote ISO Standard 31-0:1992(E), paragraph 3.2.1 International symbols for units:

"When international symbols for units exist, they, and no other, shall be used. They shall be printed in roman (upright) type (irrespective of the type used in the rest of the text), shall remain unaltered in the plural, shall be written without a final full stop (period) except for normal punctuation, e.g. at the end of a sentence.

"Any attachment to a unit symbol as a means of giving information about the special nature of the quantity or context of measurement under consideration is incorrect.

[an example is given here, impossible to reproduce precisely in HTML]

"The unit symbols shall in general be printed in lower case letters except that the first letter is printed in upper case when the name of the unit is derived from a proper name.

"EXAMPLES
m metre
s second
A ampere
Wb weber"

B for byte is therefore wrong as there never was a Mr. Byte having given his name to the unit. b would be acceptable, provided there was no confusion with any other unit, but by is better as being unambiguous. I maintain B is used for Baud which is an internationally accepted data rate unit for telegraphic purposes, named after a Monsieur Baudet. The fact that its use has diminished over the last 20 years is no excuse to usurp B for byte which, apart from being incorrect, may lead to confusion because there is a close relationship between a data rate and the size of data.

I agree the IT industry has been sloppily using wrongly designated units, to the confusion of all for some time, but it is only by educating users into the correct designations that we will end all this confusion, once and for all. The fact that this thread is discussing it is derived exactly from such a confusion. The ISO system is totally unambiguous and ISO 31 entitled 'Quantities and Units' is in as many as 13 parts, averaging about 25 pages each. ISO 1000 1.2.2. also repeats the above text but adds that a space must be placed between the numerical value and the unit symbol (Brits, please note!) and the name (not the unit symbol) shall be written in lower case even if derived from a proper name. BTW, I see from this document that b is a reserved symbol derived unit from the SI units for a barn which is 10^-28 m^2 or a small fraction of a square angstrom! It cannot therefore be used either for bit or byte!

------------------
Brian (the terrible)

[Pertti]

Brian,

How does dB fit into this, as the "d" is 10^-1 ;D