news:40b73660@news.povray.org...

>   Both of these phenomena may be as bad as a squeezed image because they
> can potentially destroy the look of the image by introducing modelling
> artifacts (lack of proper modelling of the scene) or by leaving out
important
> details of the image.
>
>   Another problem is that the automatic control of aspect ratio can only
> be done by adding/removing extra image vertically. (Making image_width
> and image_height change the 'up' vector does not help: The elongation of
> the image is still done vertically.)

I still don't get it (and I've read the thread posted by Thorsten too).
Unless it's one single colour, a squeezed/stretched image is *** always ***
bad. There's zero point in rendering it, period. So, at its
worst, the automatic trick is the lesser of two evils. And in fact, from a
practical point of view, it's far from being such a problem. During scene
development, I usually cycle through my 4 favourite ratios (1.33, 0.75, 1
and 2.66) and the automatic trick makes it a breeze until I decide on what
ratio I'll finally use for the scene. Then, scene adjustments have to be
made to accommodate the ratio (zooming, moving/adding objects etc.), but
that's basically an artistic choice, as composition is deeply linked to the
aspect ratio.
Simply put, before the trick existed, I had the various ratios x*1.33, x*1
etc. in commented form, so all the drawbacks were already there, but it was
just a real pain as I kept forgetting to uncomment the right one... I'm not
saying that problems with automatic ratio don't exist, but I still have to
see people
complaining about them and I never considered them to be drawbacks at all,
let alone harmful.
However, we still have people puzzling about stretched/squeezed images due
to
the default 1.33.
Professional video output is more a real issue (though a niche one
IMHO) but for this the user could still have the choice of using a
non-square pixel ratio, as it's the case now.

G.

-- 

**********************
http://www.oyonale.com
**********************
- Graphic experiments
- POV-Ray and Poser computer images
- Posters

Post a reply to this message

In article <40b750cb$1@news.povray.org> , "Gilles Tran" <tra### [at] inapginrafr>
wrote:

> However, we still have people puzzling about stretched/squeezed images due
> to the default 1.33.

Well, for a 1024*768 image the difference to a 4:3 aspect ratio is just two
pixels.  I doubt many will ever notice.

> Professional video output is more a real issue (though a niche one
> IMHO) but for this the user could still have the choice of using a
> non-square pixel ratio, as it's the case now.

Not only professional video output: DV is the standard format used for
digital video cameras.  Practically all video editing software supports
exports to DV cameras attached via IEEE 1394 (aka FireWire aka iLink).  And
if you do your own animations on DVDs or just static DVD menu background
images, you also have to deal with non-square pixels.

Oh, and as for inexpensive programs having problems with non-square output,
well, rendering at twice the resolution and then scaling down will be
sufficient regarding quality as far as DCT-based algorithms are concerned
(btw, DCT yields better compression of blurred or anti-aliased images and
video).

    Thorsten

____________________________________________________
Thorsten Froehlich, Duisburg, Germany
e-mail: tho### [at] trfde

Visit POV-Ray on the web: http://mac.povray.org

Post a reply to this message

news:40b75628@news.povray.org...

> Well, for a 1024*768 image the difference to a 4:3 aspect ratio is just
two
> pixels.  I doubt many will ever notice.

Hmm, we have a failure to communicate here ;) This is about the unending
flow of users who try to get a vertical, square or widescreen image and get
a squeezed/stretched one because of the default 1.33 and then complain about
it in newsgroups or to TAG members because they expect the width and height
values to work right away (without requiring another parameter change), as
they do in other renderers and 3D apps (at least all those I'm using). The
automatic ratio trick is popular for a reason.

G.



-- 

**********************
http://www.oyonale.com
**********************
- Graphic experiments
- POV-Ray and Poser computer images
- Posters

Post a reply to this message

In article <40b76476@news.povray.org> , "Gilles Tran" <tra### [at] inapginrafr>
wrote:

> Hmm, we have a failure to communicate here ;) This is about the unending
> flow of users who try to get a vertical, square or widescreen image and get
> a squeezed/stretched one because of the default 1.33 and then complain about
> it in newsgroups or to TAG members because they expect the width and height
> values to work right away (without requiring another parameter change), as
> they do in other renderers and 3D apps (at least all those I'm using). The
> automatic ratio trick is popular for a reason.

Making it automatic will not help.  Just because those who report never read
the documentation do not understand a feature does not imply at all it
should work differently.  There are a million things they won't understand
without reading it.  And it is very likely that changing it to behave the
other way there will be as many people asking the opposite question.
Consequently, we cannot make a program fool-proof, every time we try we just
create better fools or get other fools reporting the opposite behavior as
unexpected.

I bet many CD player manufacturers also got many reports in the early days
that CD did not work because people inserted them upside-down.  Today people
would probably face the opposite problem when dealing with an ancient
playback device.  Either way, one can only figure out the correct usage of
any complex device by reading the manual.  Bad luck if one doesn't...

    Thorsten

____________________________________________________
Thorsten Froehlich, Duisburg, Germany
e-mail: tho### [at] trfde

Visit POV-Ray on the web: http://mac.povray.org

Post a reply to this message

> Are there other
> common circumstances (video output?) where having square pixels by default
> would be problem?


I recently rendered a 720x480 series of images to be edited with video that
had an aspect ratio of 4:3. In addition, occasionally I renderer images
intended to be used as textures for my OpenGL programs; textures must have a
2^n x 2^m resolution, so non-square pixel aspect ratios are necessary if I
plan on drawing the texture at a different size in my program.

 - Slime
 [ http://www.slimeland.com/ ]

Post a reply to this message

Christoph Hormann nous apporta ses lumieres ainsi en ce 2004/05/28 
08:49... :

>
> The most common probably is printing on a printer with anisotropic 
> resolution.
>
> Christoph
>
Well, any decent printer drivers are made to compensate for any 
anisotropy the printer may have. There may be some oddball drivers that 
don't do it automaticaly. The real isue with anisotropic printer is when 
doing a raw transfert bypassing the drivers.

Alain

Post a reply to this message

Gilles Tran <tra### [at] inapginrafr> wrote:
> Unless it's one single colour, a squeezed/stretched image is *** always ***
> bad.

  What do you mean always?
  If I'm making a 320x400 windows startup image, I *want* the image to
still have 4:3 aspect ratio even though the resolution is not. The image
will look stretched when viewed normally, but it's exactly what I want.
  You can't say "always".

> There's zero point in rendering it, period. So, at its
> worst, the automatic trick is the lesser of two evils.

  I wouldn't be so sure that automatic aspect ratio setting is the lesser
of two evils.
  When you render with the wrong aspect ratio, you can at least clearly
*see* that there's something wrong with the image. This is specially
the case when you try to render a scene with a wider aspect ratio
in a default 4:3 resolution (some of the POV-Ray example scenes are
good examples of this): You will *see* that there's something wrong
with it.
  However, if the scenes had automatic aspect ratio correction, there's
a big chance you won't notice that you did something wrong when you
make a 4:3 render of it. You might wonder why there are some oddities
at the top and the bottom of the image (if the author didn't bother
to properly model those parts) but you would simply assume it's a
mistake of the author, not your mistake. However, you will be rendering
the image in a way the original author didn't intend.

  I personally find this specially bothering because the automatic
aspect ratio setting modifies the viewing area *vertically*, which
is the less usual image area change (usually images are made to
be viewed in widescreen, if anything). You may eg. model a landscape
which looks ok no matter where you look at it by rotating the camera
horizontally, but you don't care how the landscape looks below the
camera because it's outside the image... except if someone happens
to render the image with a different aspect ratio than you.

  So my strong opinion is that *no* automatic aspect ratio setting in
POV-Ray by default, and people should really give the "trick" with
a warning of its side-effects. (Something like "be aware that if
you render using a resolution with a different aspect ratio, the
viewing area will be modified in the vertical direction, thus giving
or removing image area from the top and bottom of the image".)

  I would like the trick more if it modified the viewing area horizontally,
but it doesn't.

-- 
#macro M(A,N,D,L)plane{-z,-9pigment{mandel L*9translate N color_map{[0rgb x]
[1rgb 9]}scale<D,D*3D>*1e3}rotate y*A*8}#end M(-3<1.206434.28623>70,7)M(
-1<.7438.1795>1,20)M(1<.77595.13699>30,20)M(3<.75923.07145>80,99)// - Warp -

Post a reply to this message