At 0127 PM 1/1/02 +1100, you wrote
David,
A happy new year to you! Below is intended to assist.
I have some doubt about your reply to the poster regarding proscan -
mind
you, I am not able to independently verify the "cut and paste" info
below.
My understanding is that proscan does indeed provide an increase in
resolution in all cases, not just for stills.
As an electronics and communications technician myself, I believe Alan
Roberts info is logical and makes considerable sense. He has
contributed
some very insightful stuff to the newsgroup in the (now distant) past.
His
credentials are impeccable and I hold him in high esteem. Until I
traded
posts with him, I had been unable to find out exactly what happened
inside
the camera when proscan is engaged. Since that time, I have seen a few
other posts that went even deeper and added to his info. Damn, I can't
find
them now so assume I didn't save them, unfortunately.
See the exchange below, with Alan replying to my post paragraph by
paragraph, with my paragraphs preceded by the colon. It is definitely
worth
the read if you have the time ..... The punch line is Alans last
paragraph,
where Proscans mode of operation and reason for the resolution increase,
suddenly becomes clear.
_______________________________________________
Hugh Marchant
So - in a DV camera we have a CCD scanning an incoming image - and
(normally) outputing a single field at a time. ie. half of a full
raster,
which would be 262 lines for each 525 NTSC field (I am more at home
with
PAL).
The entire ccd is usually copied to RAM first, at field rate, rather
than
reading it at video rate. In the RAM, the interlaced fields are made by
averaging adjacent pairs of lines, so the first field sends out line
1+2,
3+4, 5+6 etc and the second one sends out lines 2+3, 4+5, 6+7. This way,
all the light is used for each field, the output lines are correctly
phased in the display frame, and each field is from a fresh instant in
time.
Each field
is encoded then written to tape and is also fed to the
1394 socket.
Don't forget the chroma subsampling, and the data compression. What is
fed to 1394 is a data stream describing the frequency content of the
coding blocks, it isn't video.
The capture
card presumably assembles full bytes and feeds each field to data bus
and
on to the hard drive.
Tha capture card simply routes this bit-stream to the hard drive. Or, at
most separates the audio into .wav and wraps the video to make it look
like .avi or .mov.
Now, in progressive mode, 525 raster lines are scanned in the same time
it
would normally take for 262 lines. Double the data has to be stored
for a
single image scan.
Not really, the ccd is read into the RAM as before, but only once per
frame instaed of once per field. The rest of the processing is the same,
so you still get an interlaced pair of fields from each frame, but this
time they're both from the same instant in time.
You can check all this yourself by doing a trial recording. Set your
camera to normal mode, and record something moving (a metronome is very
good for this). Step through the tape a frame at a time and you'll see
that the two fields of the frame are displaced spatially resulting from
their temporal displacement (i.e. you'll see combs on the edges of
objects).
Now set the camera to frame or still mode and shoot again. Now you'll
see
that each frame is like a photograph, blur from motion integration, but
no combing on edges. You've made a progressively scanned recording.
If you've set the shutter to 1/50 (or 1/59.94), you'll obviously get
some blur from the integratioon time, after all, you're taking a photo
using a camera, and that's what they do.
And another thing - presumably bitmaps that Premiere makes from
undeinterlaced video are constructed from all the lines from field 1
and
field 2. If so, then I dont understand claims of "twice the vertical
resolution" for Prog Scan - where Premiere (4.2) is used, anyway. ie.
you
still get exactly 525 lines worth of data into your bitmaps (or TIFFs
etc)
whether it is Prog Scan or Normal Video.
You should be able to see that frames shot in normal mode have lower
resolution simply because of the adjacent line averaging in the camera.
Proscan doesn't do that, so is sharper vertically.
--
******* Alan Roberts ******* BBC Research & Development Department
*******
* My views, not necessarily Auntie's, but they might be, you never know.
*
************************************************************************
**
"Hugh Marchant"
----- Original Message -----
From "Neuman - Ruether"
Newsgroups rec.video.production
Sent Saturday, December 29, 2001 312 AM
Subject Re shooting progressive scan video. Why?
> On Fri, 28 Dec 2001 121443 GMT, "kenobi"
> wrote
>
> >Considering shooting my next short using progressive scan mode of an
XL1.
> >What are the real benefits of shooting this way? The Cannon site says
> >"increased vertical resolution, and less aliasing", this may be true
but
it
> >also incures a slight stutter look. If my final product is to be
viewed
> >mainly on TV and I output it to video, will the progressive mode have
been
> >of any use as its being viewed on an interlaced tv?
>
> The "increased vertical resolution" applies only to stills
> taken as frame-grabs during moments of motion - the
> resolution is actually somewhat reduced for motion-video
> for both static moments and during motion (as you have
> noticed). PS-mode, in my opinion, does little but degrade
> TV-viewed motion-video - though some hold the opinion that,
> since it better simulates a failing of film, it better
> simulates the look of film, a "virtue" I find difficult to
> appreciate...;-) This is video - shoot it in a way that
> optimizes its advantages (one of which is smooth and sharp
> rendition of motion-subjects); shoot non-PS-mode for best
> *video* results...
>
> David Ruether
> d_ruether@hotmail.com
> http//www.David-Ruether-Photography.com
> Hey, check out www.visitithaca.com too...!
Hi--
Thanks for the above...
The original poster was asking about the advantages/disadvantages
of shooting "frame-mode" with the XL-1...
"Considering shooting my next short using progressive scan mode of an
XL1.
What are the real benefits of shooting this way? The Cannon site says
"increased vertical resolution, and less aliasing", this may be true but
it
also incurs a slight stutter look. If my final product is to be viewed
mainly on TV and I output it to video, will the progressive mode have
been
of any use as its being viewed on an interlaced tv?"
Logically, unless something is either added to, or removed from, the
PS or interlaced system, the spatial resolution would be unchanged
between these systems (though the temporal resolution could be
degraded by mis-matching). I pointed out that in fact the XL-1
PS-mode spatial resolution is degraded relative to interlaced (and
Loren Amelong passed on the excellent article by Adam Wilt showing
why this is true - at
http//www.dv.com/magazine/2000/1100/wilt1100.html).
In this article, AW also supports my other premise (as I recall...;-),
that PS-mode shooting also relatively degrades temporal resolution,
since it provides info on motion only at 1/30th second intervals instead
of at 1/60th second intervals (and the original poster also noted the
"stutter look"...). Since I have never heard of "You should be able to
see that frames shot in normal mode have lower resolution simply because
of the adjacent line averaging in the camera. Proscan doesn't do that,
so
is sharper vertically.", unless I have missed something (quite possible!
;-),
I find it hard to believe that normal frame-rate PS-mode shooting for
TV-viewing of NTSC video images offers any advantages - but it does
offer
obvious disadvantages...
BTW, you may want to post something on scan line averaging in this
thread,
and see if someone more knowledgeable than I am picks up on it. There
may
also be relevant differences in the chroma/luma of PAL vs. NTSC
involved...