On Wed, 5 Feb 2003 03:22:30 +1100, "Hughy" <antispam@spamkiller.net> wrote:

 

>Sony claim in their operating manual (for TRV-900's "gain shift" of -3dB):

>

>"When shooting in bright conditions ....... adjusts the gain shift to -3dB

>as much as possible .....to shoot pictures with less noise in bright

>conditions".

>

>On manual,  you step all the way to wide open with -3dB gain operative.  As

>you wind the wheel down through wide open, the gain then steps as below:

>

>f1.6 (or 1.7 ... can't remember)

>-3dB (lens wide open)

>0dB (lens wide open)

>+ 3dB

>(and so on up to +18)

>

>Assuming Sony did their noise figure tests and calcs correctly, then

>shooting with gain shift **should** help if you produce DVD's and don't get

>your exposure spot on.  What degree of noise reduction is achieved is pure

>conjecture, but I'd guess maybe 2 to 3 dB increase in S/N ratio.  Maybe

>someone has some figures?

>

>The MPEG2 encoder needs as little noise as possible, because many (most??)

>encoders algorithms will faithfully replicate the noise in the resultant

>output, if noise filters aren't cut in.

>

>eg. concerts under changing light intensities.  If you underexpose somewhat,

>and need to add some gain in post, it helps if you shot for a slightly lower

>noise level than normal to begin with.  With a better input S/N ratio, you

>get an output S/N ratio that is also better (but obviously must still be

>degraded over the I/P S/N ratio).  This should only help practically with

>small to medium underexposure.  If rushes are way down, nothing I can afford

>would rescue them.

>

>That's the theory of it, anyway.  I intend to try the TRV-900 on -3dB "gain

>shift" setting next concert (don't hold your breath, they're next

>November/December).  We always shoot 4 or 5 cameras at the end of year

>concerts, on the standard assumption that we'll lose at least one (or maybe

>even two) cams.

>

>I can vouch for MPEG2 bit rates rising dramatically with noisy rushes, I

>recently observed a 2 MBit/sec encoding difference between two bits of

>footage shot with XL1 and TRV-900 when the temporarily diminished light

>exceeded the TRV-900's low level "low noise" operating point.  And the noise

>wasn't all that bad, either.

>

>regards,

>Hughy.

 

I have used the "-3db" setting on both the TRV900 and

VX1000. It offsets the aperture setting sequence, moving

the lens to a slightly wider stop until wide open (and

then moves the gain up...). Until the lens is wide open,

the effects are ones of using the lens slightly wider

(with, generally, the slightly worse performance of these

particular lenses in the wider few stops range), and also

theoretically slightly reduced noise (until the gain starts

to rise, at exactly the same illumination point it would

have with "-3db" not set) - but in practice, the noise

reduction in bright light is not useful, and there is

no advantage in low light (after the lens is wide-open),

though there can be a slight noise advantage just before

the lens goes wide open (though it is not very great...).

I have stopped bothering with it...