Le 01/09/2010 17:33, Warp a écrit :
> Le_Forgeron <lef### [at] freefr> wrote:
>> Well, you are going to have issue with the video industry.
> 
>> DVD NTSC: 720 x 480 (3:2) for 4:3 & 16:9, none of the pixel is square.
>> DVD Pal : 720 x 576 (5:4) also for 4:3 & 16:9, not a square pixel in
>> sight either.
> 
>> VCD & SVCD did neither use square pixel.
> 
>   Also, the major problem is: If you wanted POV-Ray to always assume square
> pixels and then someone renders the image with a different aspect ratio
> than what the camera is, what should POV-Ray do? There are quite many
> possibilities:
> 
>   - Always extend or crop the image horizontally.
>   - Always extend or crop the image vertically.
>   - Only extend the image (either horizontally or vertically, depending on
> whether the image aspect ratio is wider or narrower than the camera aspect
> ratio).
>   - Only crop the image (likewise).
>   - Perform a combination of both.
> 
>   The problem with cropping the image is that important parts of the
> scenery might get cropped away, destroying the original image composition.
> Extending the image might be even worse, as unmodeled parts of the scenery
> could become visible, severely degrading the quality of the end result.
> 
>   If POV-Ray supported automatic aspect ratio tuning, the *author* of the
> image should have a saying in it, rather than the decision being solely
> on the person who renders the image (who might not be the same person).
> 
>   I discuss with these issues in more detail and have a suggestion here:
> http://bugs.povray.org/task/85
> 


The more I think about it, the more I think we got it wrong from the
very start.

There is in fact 2 set of related data, for a total of 6:
 view-height, view-width, view-ratio
 image-height, image-width, pixel-ratio

view-height & view-width are usually set in the SDL, (up & right,
angle...) with an implicit view-ratio ( = view-width/view-height)

image-height & image-width are set from the command line, with an
implicit pixel-ratio ( = image-width / (image-height * view-ratio) )

It could be better if instead of -H & -W, the user got to specify either
-H or -W and the pixel-ratio ( -P float ? or -PH integer -PW interger ?).

That way, the view would stick to the intent of the scene's creator (in
unmodified SDL) and the picture would be fine for the display of the
final user.

The issue would be the usage of image_width & image_height in SDL, they
would have to be gone.

If the final user wants to see more that intended by the file's creator,
he would have, with that system, to open & edit the SDL.

Final note: I do not ask for the removal of -W or -H, but that instead
the mandatory missing dimension should be provided by computation based
on pixel ratio and the provided dimension (and it might be convenient to
provide either of them to please the final user).
On the same line, an explicit "view_ratio" in the camera might compute
the up length from the right/angle size (and it might be a clearer clue
for the casual reader:

camera { ...
right x
angle 14
view_ratio 16/9
look_at ...
}

-- 
A good Manager will take you
through the forest, no mater what.
A Leader will take time to climb on a
Tree and say 'This is the wrong forest'.

Post a reply to this message