DV-List: Audio dynamic range,RTMac ship dates,VERTICAL RESOLUTION,Wouldn't AGC do the following

.footer { }

Knowledge

Hardware

Software

DV-Movies

HowTo

Misc

A DV(L)-FAQ [e]

DVL-Digest 535 - Postings:
Index

Audio dynamic range
RTMac ship dates
VERTICAL RESOLUTION
Wouldn't AGC do the following

Audio dynamic range - Adam Wilt

> Most consumer camcorders are limited to the 12bit 32kHz option of the DV
> format. This would give a maximum dynamic range of 72 dB with no headroom.
Ah, but the 12-bit mode uses a perceptually-coded nonlinear compression; its
dynamic range is still pretty much 96 dB.
> To put all this into context, a few years ago the very best broadcast
> video tape machines had a dynamic range of perhaps 50dB with a following
> wind!
Quite so! Be it 72 dB or 96 dB, we don't realize how good we've got it!
Cheers,
AJW

RTMac ship dates - "Adam J. Wilt"

> If Matrox ships the RTMac in July it will be the first video product they
> have built in the last two years to ship when promised...
Who promised July? The earliest timeframe I heard at NAB was September or
October, and more likely late Q4.
July may have been promised for the Pinnacle TargaCine card, but not the
RTMac.
Cheers,
Adam Wilt

VERTICAL RESOLUTION - "Adam J. Wilt"

> 4. Every reason to assume XL1 has the same = 340 TVL -- unless pixel
> off-set is working in non-Frame Mode. Someone needs to test!
It is the same. Vertical pixel offset only occurs in Frame Movie Mode and in
digital zoom.
> 5. Switching VX1000 to 1/30th S: I have NO idea what Adam means by "The
> VX1000 has full resolution vertically (340 TVL) in the raw DV data but
> sets the field-doubling bit so that analog output is single-field,
> doubled. Rendering the pix to lose the bit is NG because at 1/30 on the
> VX1000 the fields still show interlacing."
Looking at the pix on analog outputs, the resolution drops by half and the
picture is clearly a single field, line-doubled, when switching to 1/30.
However the same pix taken into an NLE and viewed fullscreen show no such
halving; both (interlaced) fields are still present.
What this tells me is that the field mode flag in the DV datastream is set to
replicate one field; the hardware codec in the camera sees this, but the
software codecs I use don't, and show both fields regardless.
> Is the resolution still 340 TVL?
On the tape? Yes. But if the field-doubling bit is set, you'll see 170 lines
on the analog outputs, and a coarse, jaggy rendering at that.
> 6. Switching VX2000 to 1/30th S: I'm not sure what Adam means by "The
> VX2000, alas, appears to interpolate one field from the other prior to
> recording; its vertical resolution at 1/30 is half of that at 1/60." Is
> the resolution only 170 TVL?
Well, half of 340 is...? ;-)
The pix from the 2000 at 1/30 are identical whether viewed from the camera's
analog output or on the screen of an NLE. One field appears to have been
interpolated (fairly smoothly) from the other. There are no obvious jaggies
as a result, but the vertical resolution is, well, about 170 TVL.
> 7. XL1 in Frame Mode at 1/60th S (second field interpolated using
> pixel-offset from first) = 360 TVL. I assume CCD row-pair summation is
> still used ???
CCD row-pair summation is used, but vertical pixel offset is also employed
(the green CCD is shifted one scanline (physical CCD row) with respect to the
red and blue CCDs and the Y signal is a weighted average of resultant
scanlines from the G and R/B CCDs. As a result vertical resolution is roughly
halfway between 340 and 170; I don't have the exact # handy because I'm
travelling and my test tape isn't.
> 8. XL1 in 1/30th S ???
No data.
> 12. Measure of XL1?
Looks comparable to GL1 with GL1's sharpness one notch from minimum, but I
don't have a hard figure. As I recall the GL1 was around 490-510 TVL/ph when
aliasing took over, but again, my test tape is 2000 miles away.

Wouldn't AGC do the following - "Adam J. Wilt"

> My assumption that AGC is a DSP function is a guess. Sony seems to take
> related functions and place them in LSI. I expect there is a single
> small chip with:
>
> Voltage controlled op-amp
> anti-alias filter
> A/D...
There's an awful lot of assuming going on in this whole thread, and as a
result plenty of folks are getting their exercise by jumping to conclusions.
Until we know what's going on, a detailed discussion of the supposed systems
and their side effects is about as useful as a discussion of how many
IBP-frame MPEG-2 compressed angels can dance on the head of a PIN photodiode.
> Remember that to increase gain in NLE post by a factor of 2 only
> requires a "shift 1 bit left." No increase in noise. A 4X requires
> "shift 2 bits left." No increase in noise.
Alas, quantization noise (the audible equivalent of banding, and the
accompanying loss of fine detail) increases in the process. Subtle details
and clarity will be lost.
Dolby has a very nice demo of one-bit audio recording and reproduction. It's
listenable, if you don't mind the dither noise!
And remember, the limiting factor with most of our gear is not the 96 dB or
so of dynamic range in digital domain, it's the self-noise of the analog
electronics, RF pickup in cables, the noise floor of the mic, and the quality
of the A/D and D/A circuits.
IMNSH and very cranky O, it still pays to set levels based on the audio
content of the shot. AGC is all fine and well, but in my experience a low
recorded level is more problematic in post than a level that peaks in the
"visible" range around -12dBfs.

(diese posts stammen von der DV-L Mailingliste - THX to Adam Wilt and Perry Mitchell :-)

[up]

last update : 21.Februar 2024 - 18:02 - slashCAM is a project by channelunit GmbH- mail : slashcam@--antispam:7465--slashcam.de - deutsche Version