Path: utzoo!mnetor!uunet!husc6!hao!gatech!bbn!denbeste
From: denbeste@bbn.COM (Steven Den Beste)
Newsgroups: comp.sys.amiga
Subject: Digi-view praise and a bug
Message-ID: <5998@cc5.bbn.COM>
Date: 14 Jan 88 14:10:15 GMT
Organization: Bolt Beranek and Newman, Inc., Cambridge, MA
Lines: 75

I bought myself a Digi-view for Christmas, and also a TV camera. I got the
camera they recommend in the manual (Panasonic WV-1410, I think - I'm typing
this from work and don't have it here), and I'm running it on a 1000 with an
Insider (total 1.5M) with two floppy drives and a Cltd 24M HD.

Since Digi-view implicitly comes with a "dongle" (the conversion module) they
don't copy protect anything, so I copied the software onto my HD and it works
BEAUTIFULLY! I am astounded!

I've been using it continuously ever since 12/25 (and I'll have you know it was
a significant struggle to keep from opening it before then).

I've had no gurus, and only one strangeness has shown up - which is pretty
remarkable for a software package as complex as this one. I can recommend this
product without qualifications for quality and features and useability. (Some
of the display modes are a little slow, but since I understand some of what
they are doing, I'm surprised they are as fast as they are - if I had written
them they would have been ten times slower! Nonetheless, I wonder: Would
"sharpness" and "saturation" run faster with a 68881?)

That one strangeness: In order to avoid any danger of etching the video camera,
I've taken to converting 7-10 pictures at a time and storing them in ".RGB"
format (which takes between 250K and 300K per picture - DAMN HD's are nice),
then powering down and lens-capping the camera and going back and
post-processing them to convert into HAM format.

When I do this, the picture that results has a distinct green tinge.

I originally thought I was being zapped by the auto-exposure of the camera
combined with too wide an iris setting (which causes the camera to throttle
down the red more than the green, so the net effect is a relative boost in
green. Indeed I was doing this, but I have figured out how to avoid that and it
isn't now my problem.

Let me describe the experiment I did last night to prove to myself that there
was a problem:

I captured a picture (after making sure the light levels weren't too high) and
with all "color" settings at neutral saved a "4096" HAM image. I then saved the
"RGB" and executed "new". I reloaded the "RGB" and saved another "4096" HAM
image, then exited (actually used LEFT-A-N to get back to my main CLI) ran
"show" (PD, so far as I know) and looked at both.

The second image was unquestionably dimmer, less contrasty and greener than the
first. Just to prove that this isn't subjective, the second HAM file was
also about 2.5% smaller than the first, though it could just as well have been
larger. The point is that given the same data, I would have expected the same
output or very similar (given the random effects of dithering, set to 1).

I did this with 320*400 mode. The info-frame says "Digi-View V2.0 Rev 0".


I have a conjecture as to what has happened: There is a function they call
after a camera capture which normalizes the color. You can see its results in
the second B/W sweep after the capture of each color, usually in the image
getting more contrasty and less layered.

The "RGB" file format should be, and is advertised in the manual as being, the
actual raw capture data. I conjecture that when an RGB file is read this same
function should be, but isn't being, called. As a result, you are now operating
on the unnormalized data the second time - and you don't have a manual control
which allows you to do this normalizing by hand (I think it is multiplicative,
and the color sliders appear to be additive). The results have always been less
satisfactory, since in order to get something approaching a normal color I have
to run the contrast, brightness, saturation and color sliders all over the
place - and by doing this effective data is lost and the picture gets noticable
layers.


If there are other Digi-View owners out there, I would appreciate it if they
would run the same experiment and see if they get the same result. Is it me, or
is it Memorex?
-- 
Steven C. Den Beste,   Bolt Beranek & Newman, Cambridge MA
denbeste@bbn.com(ARPA/CSNET/UUCP)    harvard!bbn.com!denbeste(UUCP)