Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!cs.utexas.edu!swrinde!zaphod.mps.ohio-state.edu!pacific.mps.ohio-state.edu!tut.cis.ohio-state.edu!ucbvax!bloom-beacon!bu.edu!bu-cs!mirror!frog!lmrc!hassinger
From: hassinger@lmrc.uucp
Newsgroups: comp.sys.amiga.hardware
Subject: Re: (Non) Square Pixels?
Message-ID: <4736@lmrc.uucp>
Date: 9 Feb 90 21:33:00 GMT
References: <4687@lmrc.uucp> <3119@cello.UUCP>
Organization: Liberty Mutual Research Center, Hopkinton, MA
Lines: 78

In article <3119@cello.UUCP>, robin@sabre.uucp (Robin D. Wilson/1000000) writes:
> In article <4687@lmrc.uucp> hassinger@lmrc.uucp writes:
>>A posting today reminded of an issue that I have not seen discussed very much.
>>The issue of the lack of "square pixels" on the Amiga.  I think it was
>>suggested that just getting the right monitor would solve the problem for the
>>case in point.  I don't know that that is a real answer.
>>Bob Hassinger
> 
> The machine doesn't control the shape of the pixel.  It merely sends the signal
> to the montior to GENERATE the pixels.  Only the monitor controls how they are
> shaped.  
> 
> Most monitors (and TV's do not have "square" pixels.  Most have round or oval
> shaped pixels.  In fact, the only monitors that I know to have square pixels 
> are the high-end Sony's similar to the one Apple uses for the MacII.  (The
> Apple color monitor for the MacII is a modified Sony.)  
> 
> Personally I prefer the round pixels, I beleive they show smoother curved lines
> than square pixels.  But then, given a high enough resolution, it wouldn't
> really matter anyway.
> 

Looks like I am going to have to explain further.  I don't care about the shape
of the spots displayed on the screen.  I care about about how many there are.

In the vertical direction you generally get one pixel per scan line (maybe per
two scan lines depending on what you are doing about interlace, etc.).  A
direct integer relationship to the lines.

In the horizontal direction the design of the machine determines how many
pixels will be displayed across the screen.  For example something like 768 for
hi res overscan, or 640 in the smaller, standard display area.  Either way you
get a certain number of pixels per inch on your screen in each direction.

On some computers the number of pixels per inch on the "standard display" for
that system is the same in both horizontal and vertical directions.  If that is
the case, in a program like DPaint you can take a brush of a circle or a
square, rotate it ninety degrees, and it will still be a circle or a square
because it is made up of the same number of dots or pixels in each direction
and when they display they come out taking the same distance on the screen in
both directions.  I think this is the normal case on the Mac for example.

It is NOT the case on the Amiga however, because the horizontal resolution is a
bit too high.  Something like 10 or 20%.  So, if I create a brush in DPaint III
and animate it to rotate, it changes shape as it turns.

The first issue is deciding what the "standard" display width to height ratio
is so I can decide how many pixels I need in the horizontal direction to come
out right.  This is because simply by adjusting the height or width of the
display I can compensate (e.g. the reference to the "right" monitor that comes
adjusted this way).  I contended that on the Amiga the standard should be the
standard used for television.  I beleive this is 4 to 3.  That way, my
animation would look right when displayed on television - a very popular use of
the Amiga - and you have to live with the way all those TV sets are set up, you
don't have the option of changing all of them.

Another place you see this effect is in a desktop publishing system like
ProPage.  It seems to display the page using something like 75 pixels per inch
of the page you are working on, in both directions.  Because of this the
display is distorted from what it will really look like.  If you check the
rulers displayed on the edges you will find the "inches" are different lengths
along the two directions.  You also have to make a compensation when you import
bitmap graphics to compensate for the fact they where generally created to look
right on the non-square pixel display - there are more dots per horizontal inch
than there are per vertical inch I beleive.  You have to stretch the bit map in
one direction by 16% or so to have it come out right when it is printed.

Again, the standard on the Mac seems to be to have square pixels that are
really intended to display and print with the same pixels in the horizontal and
vertical directions.  No distortions, no stretching or fudging.

My question was if anyone else has noticed this, how they work around it, if
anyone cares about it, and what work arounds there are for cases like my
spinning video animation.

Bob Hassinger
508-435-9061
...uunet!ccavax!lmrc!hassinger  or  hassinger@lmrc.UUCP