On Mon, 22 Apr 2002 15:08:14 -0400, "Mark & Mary Ann Weiss"
wrote:

>> >I think MSDV is seriously flawed. It clamps the highlights and ruins all
>my
>> >exported and rendered footage for DVD authoring. If I want to author a
>DVD,
>> >I have to accept blown highlights as the norm. That is unacceptable.
>>
>> MSDV shouldn't be entering into the picture for MPEG-2 encoding, unless
>> you're re-rendering DV format files in the intermediate stages. If that's
>> the case, (if you even need to save intermediates at all, i.e. if you're
>> using one program for all of your edits and then using TMPGEnc or
>something
>> to make the MPEG-2 files) you can avoid doing so by saving intermediate
>> streams as HuffYUV or even uncompressed AVI files.
>>
>> Not that this makes MSDV any better, but if you're going to DVD in the
>end,
>> it doesn't make sense to re-render intermediate stuff with a different
>> compression codec.

>I'm getting really confused about where the problem lies. I just performed a
>test, an objective test where I could measure RGB levels.
>
>Here's what I did:
>
>Create a Photoshop file with RGB255 background.
>Put vertical strips of RGB235
>Make the lower half background RGB235 and put strips of RGB217 in that
>background.
>
>Load into Premiere and make it 10 seconds long.
>
>Export the clip to the Sony VX2000 through DV-fire wire.
>Capture the resulting video back into Premiere from the VX2000.
>
>Compare the two by exporting single frames to TIFF and reading RGB values in
>PhotoShop.
>Result: RGB255 stayed the same.
> RGB235 became 234 (hardly changed)
> RGB217 became 216 (again, insignificant).
>
>But... when I SHOOT video of the physical world outside, THAT video gets
>clipped!
>
>Hypothesis: The VX2000 (and the TRV900 as well) are producing RGB values
>greater than 255 somehow. (I know they are 14-bit per color, but more info
>on how these out of bounds intensities come to exist is needed.)
>
>So I've been able to create scales of white levels in the PC, export them to
>DV, reimport them and find their fidelity absolutely intact. But real
>footage optically aquired by the camera seems to exceed these values and
>gets crushed.
>
>So now the question remains, what is going on here?
>
>Anyone care to take a stab? Perhaps the CODEC is not the real problem.

Interesting tests...
Since I also use Premiere and the VX2000, and often shoot
the sun, sun reflections off water and other items, white
flowers, skies, interiors with windows showing skies and
snow, and other very "white" items, and often use filters,
transitions, motion, etc. on parts of the footage, causing
a mixing of rendered and unrendered footage within the same
clips without experiencing "clamping" of the lighter tones
(with Canopus Raptor and its excellent codec), I conclude
that it is either a codec problem, a settings problem,
or...? But it is unlikely a VX2000 or Premiere issue...
BTW, the highlight-rendering of the Sony 3-chip cameras is
better than most others - the Canons, for instance, push
the lighter tones relatively higher in value, "blowing out"
detail in the highlights (this should result in even more
obvious "clamping", I would think...). I'm not surprised
that you are almost alone in reporting this difficulty,
though, since it appears most people don't notice most
picture defects, so they may not notice this one...;-)
(More on video image characteristics and defects is at:
www.David-Ruether-Photography.com/vid_pict_characts.htm).
I wish you good luck in finding a solution...!