Posted by PTravel on 01/12/37 11:53

"Martin Heffels" <youwishyouwouldknow@nottellinya.com> wrote in message
news:ufl6b29br5sqi0lk2tm3a6hv1q4uqmihhp@4ax.com...
> On Tue, 11 Jul 2006 07:38:03 GMT, "PTravel" <ptravel@travelersvideo.com>
> wrote:
>
>>> You can answer that question based on your own experience of course.
>>> What
>>> these numbers tell you is that there is a lot happenening that you don't
>>> even notice, repaired unnoticeable to you, but noticeable to the player.
>>
>>And what the numbers don't tell me are how many of them are uncorrectable.
>>Is sticking to the subject really that difficult?
>
> Did you miss out that I said "You can answer that question based on your
> own experience"?

I have answered it based on my own experience: not one uncorrectable error
in 200 hours.

>
>>> Think about it: 100 drop-outs per minute. They won't all be
>>> correcteable,
>>
>>How do you know? That's a rhetorical question, by the way. The answer
>>is,
>>you don't and the Sony sheet that you posted doesn't provide the answer.
>
> That's right. But I keep looking for more info.

I appreciate that, and I look forward to seeing it when you find it.

>
>>
>>> so repair will take place by copying/interpolating data. Seeing the
>>> amoun,
>>> I really start to believe that every copy you make, you loose data and
>>> thus
>>> you loose a generation.
>>
>>You believe a lot of things. I'm not interested in your beliefs. I'm
>>interested facts. So far, you've presented none.
>
> I presented you with the fact that there are already a lot of errors when
> you playback your tape. I bet you didn't even realised that, and that is
> all because you didn't see it.

I know that miniDV uses ECC. I said that in my first post. The number of
corrected errors is irrelevant.

> On the DSR1800-deck I have seen the meter
> frequently drop into the red bar, but the picture still looked ok. That
> means that the error-correction is doing a bloody good job in correcting
> the errors somehow.

I never suggested otherwise.

>
>>No need. Once more: D-25 video has better quality than DVD-compliant
>>mpeg.
>>Period.
>
> Not at the same bitrate.

That's like saying apples are equal to oranges, except for the difference in
peel, fruit, texture and taste. DVD-compliant mpeg doesn't have the same
bit rate as D-25.

> That's what I said. Proved in yet another
> SMPTE-article. And I think I also posted a link to a codec-quality
> comparison which tells the same.

Again, your comparing apples and horseflies. It doesn't matter which codec
is better. What matters is whether the video quality of D-25 is better than
DVD-compliant mpeg. It is, and it always will be.


>
>>> Many more articles written about this with the arrival of HDV which
>>> tells
>>> you I am right.
>>
>>Irrelevant to the discussion. We're not talking about HDV, but the
>>comparison between D-25 and mpeg2.
>
> And, what compression does HDV use again? :-)

HDV isn't D-25. DVD-compliant mpeg2 is a subset of the mpeg spec. Again,
HDV is irrelevant to the question of whether D-25 produces better video
quality than DVD-compliant mpeg2.

> HDV has a higher pixel-count
> than DV25, yet it is compressed to MPEG2 with the same bitrate as DV25,
> and
> still looks better.

Except that HDV mpeg2 has more than 2-1/2 times the bit rate of
DVD-compliant mpeg2. That's why it's irrelevant to the comparison of D-25
and DVD-compliant mpeg2.

> Tells me that MPEG2 is a much more efficient way of
> compressing video than what is used with DV25.

I'm sure it is. So what? mpeg2, in its DVD-compliant flavor, produces
poorer quality video than D-25.

>
>>Typical non-lawyer misunderstanding of what that means, which is this:
>>the
>>state bears the burden of proving, beyond a reasonable doubt, the
>>existence
>>of facts that compel a finding of guilt. You and Richard keep insisting
>>that that D-25 results in generational loss because of the existence of
>>uncorrectable errors.
>
> I will make it even worse for you: even with corrected errors you get
> generation loss. Because the errors can be corrected so _you_ don't see
> them, but the bits are different from what they originally where, so you
> have lost a generation.

Now you've just re-defined corrected errors. Sorry, I'll stick with the
definitions we've been using, i.e. there are two kinds of error correction,
one that results in bit-accurate data, and one that interpolates and doesn't
result in bit-accurate data. Or are you now claiming that all ECC results
in interpolated, non-bit-accurate data?

>
>>I've asked you to quantify the number of
>>uncorrectable errors, since a miniscule amount is meaningless for the
>>purpose of this discussion, which is: will D-25 result in more
>>generational
>>loss than SuperBeta?
>
> It had to do with component vs firewire copy to begin with.

That's right. And component is . . . analog. And firewire is . . .
digital. So the question posed was: does digital result in greater
generational loss than analog? And the answer, of course, is it does not,
because there is no generational loss for digital. Did you have something
in mind other than D-25 (or, for that matter, D-50? D-1?)

>Then generation
> loss popped up, in which DV25 and Beta-SP where mentioned, and now the
> severeness of the error-correction.
>
>>You haven't met your burden of proof because (1) you've provided no
>>evidence
>>but, more importantly, (2) you don't know.
>
> I do know, but I am looking for more facts. That's what you would do in
> your profession as well.

Exactly. And what I would never do in my profession is present my
unsupported belief as fact. See, what's interesting here, and as I've said
several times, I'm perfectly willing to be persuaded that you and Richard
are right. All it will take is a fact or two.

>
>>Entirely predicated upon speculation, with no foundation in fact. All
>>right, here's some more speculation. I'll take Richard's 8 bits per 10
>>minute error guestimate, resulting in a reliability for D-25 of
>>99.99999975%
>>
>>So, 100 errors per minute * 30 minutes = 3000 errors, .00000025% of which
>>are uncorrectable and result in data loss. That's a total of 0.00075
>>errors
>>per 30 minutes, or 0.0015 errors per hour of video. Put another way,
>>you'd
>>have to shoot 666 hours of video before you hit one uncorrectable error.
>
> 666 hours? Eeek :-)

That's the math. Maybe there's a hidden message there.

> But let me bring this forward again and again: errors
> can be corrected without you noticing it, by borrowing or interpolating,
> but they content of the bits has changed.

I know. You've said it many times. And each time, my reply is the same:
how often do corrected-by-borrowing-or-interpolating
errors-so-that-I-wouldn't-notice-but-the-data-changed-anyway errors occur?
Because if its the number in Richard's guestimate, then even the busiest,
most hard-working pro in this newsgroup isn't going to hit one more than
once or twice a year.

>
>>Using Richard's assumption, as well as yours.
>>
>>Now, me -- I don't like to assume. I like hard data.
>>
>>Let me know if you get some.
>
> Only if you promise to read them :-)

Absolutely. I'd like to know the answer to this.

>
> [...]
>>I missed the link, though, as it turns out, it wasn't relevant to anything
>>in this discussion.
>
> No, it wasn't. It's probably not what you wanted to hear.

Show many anywhere in the sites at your links that quantifies the number of
corrected errors that result in changed data.

>
>>> So whatever I find, I will keep for myself, and maybe
>>> share it with mr Crowley, and Craig the Tapeguy, and let you live in
>>> your
>>> perfect world, so you don't have to worry too much. It might distract
>>> you
>>> from shooting your nice city-footage ;-)
>>
>>Yes, feel free to be the Guardian of Secrets. That's not why you post on
>>Usenet anyway, is it?
>
> I am a Knight of The Templars and guard even bigger secrets. But don't
> tell
> anyone.

Too late.

>
>>
>>
>>>
>>>>That means that 99.99999975% of the time, the data is accurate, i.e. a
>>>>bit-for-bit copy.
>>>
>>> That is not a bit-for-bit copy, because that would be 100%!
>>
>>I'll tell you one thing that it's not, though: "generational loss."
>
> It _is_ generational loss, because you have something which is different
> from the original.

One 8-bit error for every 666 hours of video. That means 665 miniDV tapes
would be pristine. How would you quantify that "generational loss"?

> Just like an analogue video has more noise and
> chroma-errors once it has been copied, which means it is different from
> the
> original caused by copying.
>
> cheers
>
> -martin-
> --
> "If he can he'll smile 'cos he's a Royal Crocodile."

[Back to original message]