Conference hydra::amiga_v1

    re: .0
    
    400 lines?  Are you sure?  This would require 800 pixels across, which
    seems like a lot, and an SVHS or better monitor.  Was this 400 lines
    number from the specification, from the salesperson, from an
    independent test lab, or from measurements you performed yourself?
        John Sauter

    That figure is from the manual.

    re: .2
    
    Did it say how the resolution was measured?  It would be interesting to
    repeat the measurement procedure to see if you could get the same
    results.  A good TV shows about 240 lines across.
        John Sauter

    A good monitor shows 400 lines.
    
    My HandyCam has 400 line resolution too.

Re: .2,.3

There are different ways of measuring resolution:

The first and the more traditional way is to see how many fine lines of
alternating black and white you can show before the lines resolve into
a smear.  The resolution is the number of white (or is it black) lines
that you show.

The second method, which is more popular with computer video, is to
count the number of scan lines output.

By the first of these measures, a high quality TV will have a resolution
of about 250 lines.

By the second measurement method, the same TV will have a resolution
of about 500 lines.

Note that there is a builtin factor of two difference because someone
changed the definition.

An NTSC signal is spec'ed at 525 lines (made from two interlaced fields
of  262.5 lines).  No all of these lines are viewable on a TV set
since most sets are calibrated to show only the middle lines of the signal.
The reason for this is that with time the image on a TV contracts.  TV
manufactures assume that people feel better losing some picture area
on the new set so that in a few years they don't have a dead area
bordering the picture.

I suspect that the camera in question outputs the equivalent of an
Amiga's interlaced signal: 400 lines of data within an interlaced
screen of 525 lines.

I suspect that note .2 is using the first definition of resolution,
and that note .3 is using the second.

RE:.3

>   Did it say how the resolution was measured?  It would be interesting to
>   repeat the measurement procedure to see if you could get the same
>   results.  A good TV shows about 240 lines across.

John,

The manual refers to lines as in scan lines, not horizontal
resolution. The CCD sensor has about 350K pixels. The images as
recorded (analog) on the floppy has 350 scan lines of resolution. How
this is translated into 400 lines I do not know.

The realistic resolution of the xapshot is like 320wX400h. The
sampling frequency is just too low for a 640 pixel picture to look
*really* sharp. It does still look pretty good.

With a Digiview setup, use the Pro-res color seperator. It is quite
amazing and well worth the extra money (over the sunrise). It has a
mode designed for using with still video cameras and allows RGB
preview (switchable with the AMiga video) of the image. I was not
able to tawke advantage of this since it comes with the cable for
a 1080 monitor and did not care to make a new one for a loner piece
of equipment.

I am currently using a mimetics framebuffer and getting close to the
same quality as Digiview. Sometimes better. Sometimes worse. It is
faster than Digiview and has more overscan. It is also inside the
Amiga 2500 and does not need a external accessory like the pro-res.

	-SES

    re: .5
    
    I reject the notion that there are two ways of measuring resolution,
    with the same name, which differ by a factor of two in their results.
    (I don't say that Randy is wrong, I just reject it.)  Having two such
    systems is bound to lead to confusion, so I refuse to accept it.
    
    To my mind, you measure resolution in "lines" by doing just what Randy
    described: see how many distinct lines you can view at one time. The
    vertical measurement will be half the number of scan lines, unless the
    device is badly flawed.  The horizontal measurement will be, at best,
    half the number of pixels, if the device is digital.  If the device is
    analog you will get a good idea of its bandwidth from this measure.
    
    I have seen CCD cameras advertised in magazines with around 350K pixels,
    but the ads never give the geometry.  Let's suppose that the picture is
    525 scan lines high, since the CCD is probably a commodity product
    intended for use with regular video cameras, which have to output all
    of the 525 lines specified by NTSC.  The number of horizontal pixels
    then comes to less than 700, which isn't enough to produce 400 lines
    of resolution, by the only definition of "lines of resolution" that
    I accept.
    
    Maybe I've got the CCD geometry wrong, or maybe they're playing games
    with the specification, such as doubling the number of lines to provide
    the "big numbers" that marketing people are so fond of.
    
    Does the Xapshot have a provision for dumping its digital data
    directly, perhaps through an RS-232 interface?  That might be less
    convenient than grabbing the frame, but it would be less expensive and
    probably yield higher quality images.
    
    re: .6
    
    400 scan lines?  350 scan lines of resolution?  I'm afraid I don't
    understand.  You said that the sampling frequency is too low for
    640 by 400, but that's only 256K pixels.  Perhaps there's something
    other than the number of pixels that limits the image quality.
    
    re: .4
    
    How do you measure "lines of resolution" (since the ambiguity of that
    term has been suggested).
        John Sauter
    

    The problem with digitizing a Xapshot picture or a VCR freeze-frame is
    the slight jitter that shows up as a raggedly edge of red, green, and
    blue pixels along vertical lines (such as a subject's face). A
    framegrabber can probably avoid this since it grabs an image in around
    1/60 of a second. But for Digiview in color-camera mode where it takes
    around 90 seconds, the problems stand out. 4096+ mode eliminates some,
    but not all.
    
    Tell me more about the Mimetics ... speed, price, software, etc.

re: .7
    
>    I reject the notion that there are two ways of measuring resolution,
>    with the same name, which differ by a factor of two in their results.
>    (I don't say that Randy is wrong, I just reject it.)  Having two such
>    systems is bound to lead to confusion, so I refuse to accept it.

I admire your taste and respect your conviction.

    re .8
    
    	I also have the mimetics framebuffer/grabber. It is a single
    amiga board and uses an standard zorro slot. The card has two
    connectors, one in and one out. Both are NTSC composite video.
    The card uses 24 bit per pixel in and out.
    
    	In grabber mode the card passes the video through. You click
    on a gadget on the amiga screen and your looking a still frame.
    It must capture it in real time. At this time the data is displayed
    as 24 bit/pixel overscan ( about 730 x 460,I forget the exact numbers)
    and the image is in onboard memory. You then can save the image as
    24 bit ( 3 different formats ) or IFF ( standard resolutions and colors
    including ham ). The video output blanks during the save.
    
    	I display mode the board displays 24 bit ( 16 million colors ) or
    translates any IFF to 24 bit and displays it. Many HAM pictures look
    better. The output blanks while the data is loaded.
    
    	Speed ?
    	The capture is real time. On my system, ( 33 mhz 68030 with 32 bit
    mem ) I can move a 24 bit/pixel image into or out of the board in 
    about 3-4 seconds to RAM disk. The IFF convertion takes longer.
    
    	Cost ?
    	about $750-$800 with display, capture, and RAM. They sell the board
    without RAM and without capture for $550. The capture can be added
    later. The RAM is 2MB of 256K dips ( I think 100 ns ).
    
    	Software ?
    	I comes with a control program. It does everything. The version
    I have doesn't support AREXX but Mimetics keeps saying their working
    on an update. It also somes with a Sculpt-3D/4D driver.
    
    
    			Steve Peters
    

RE:.10

>    It must capture it in real time. At this time the data is displayed
>    as 24 bit/pixel overscan ( about 730 x 460,I forget the exact numbers)

The saved image geometry is 746x484. This includes 2-4 pixels of garbage
along the left edge of most images I have Digitized with mine. I am quite
dissapointed with the software and its speed. Mimetics seems to have decent
hardware, but are weak in software. The current revision of the mimetics
software has problems with its file requester when used on an A3000. It 
seems they break some rules around 24bit addressing vs 32 bit. sigh.

RE:.7

>    I reject the notion that there are two ways of measuring resolution,
>    with the same name, which differ by a factor of two in their results.
>    (I don't say that Randy is wrong, I just reject it.)  Having two such
>    systems is bound to lead to confusion, so I refuse to accept it.

John,

Everyone should have the freedom to close their mind as they so desire! 

Please note, though, that "lines" are an inherent part to NTSC and raster
scan video. When one talks about lines in NTSC context the absolute limit
vertical pixel resolution depends on the number of horizontal scan lines 
displayed. These lines are generally visible in a video display and can be
counted. Are you refering to these lines? Are you refering to line pairs
as used in photography? Are you counting the dark spaces of these line
pairs or the light lines? Since you have a very definite idea of what
"should" be please explain it!

I agree that multiple ways of measuring resolution can be confusing, but 
they can also be useful. Do you feel the same way about measuring audio
power output? If so you must have a difficult time dealing with that.

>    Does the Xapshot have a provision for dumping its digital data
>    directly, perhaps through an RS-232 interface? 

No. It outputs NTSC composite only. The image is stored on the floppy in
an analog format. Well, maybe. There is a several thousand dollar disk
player that has a RS232 output. I have only seen it in brochures. 

	-SES

    re: .11
    
    I was referring to the term "lines of resolution".  That term should
    have only one definition in an imaging system, not two!  "Lines" is
    something else, I agree.
    
    I don't object to having multiple ways of measuring resolution, or
    audio power output, as long as they have different names.  As far as
    I know, the various audio measures have different names.
    
    A several thousand dollar disk player with an RS-232 output could be
    doing the digitizing itself.  Thus, the image may well be stored in
    an analog format, despite the existence of this machine.
    
    If you take a picture of a picket fence, how many pickets can you get
    in the frame before the sticks fade into the background and you can't
    count them any more?  This procedure is, in my opinion, a good way to
    measure "lines of resolution".
        John Sauter