I've been trying to figure out what the "best" factor to use is for the
camera angle keyword. In the PovRay distribution (3.6.1), the scene files
seem to imply that the authors favor "angle 35" to depict a "normal" view.
I know its a complicated subject, not as simple as one might expect. I've
been surfing for research. I can't find an answer that "I feel comfortable
with". For one thing, humans percieve *binocular* vision (approaching a 190
degree field of view in the x-z plane *if you are young* and not a crusty
old thing like me) - whereas when considering *just one eye* things are
different (where is Popeye when you need him?). And the FoV in a 35mm
camera using a 50mm "normal" lens (which actually approaches what I want to
depict in PovRay - and even then a "normal" lens isn't really "normal") is
different then even just one human eye.

Discussion?

Post a reply to this message

> I've been trying to figure out what the "best" factor to use is for
> the camera angle keyword. In the PovRay distribution (3.6.1), the
> scene files seem to imply that the authors favor "angle 35" to depict
> a "normal" view. I know its a complicated subject, not as simple as
> one might expect. I've been surfing for research. I can't find an
> answer that "I feel comfortable with". For one thing, humans percieve
> *binocular* vision (approaching a 190 degree field of view in the x-z
> plane *if you are young* and not a crusty old thing like me) - whereas
> when considering *just one eye* things are different (where is Popeye
> when you need him?). And the FoV in a 35mm camera using a 50mm
> "normal" lens (which actually approaches what I want to depict in
> PovRay - and even then a "normal" lens isn't really "normal") is 
> different then even just one human eye. 
> 
> Discussion?

My current view on this subject is: the final image should cover the
same angle when viewed.
So, let's imagine a render for your Windows/X/.. wallpaper.
You can measure the traditional distance between your nose/eyes/... and 
the display.
You can measure the width and height of the display.
Then you can do the maths for the "best" value..., at least one that 
would turn out to be believable for the perspective. (if the scene fits!)

And then, there is the artistic effect :-), using a fish eye, or any 
other camera..., or Out of scale objects...

-- 




l'habillement, les chaussures que le maquillage et les accessoires.

Post a reply to this message

Le Forgeron:

Thanks, I did something similar this morning (made a mistake in the
measurements though, I need to do it over again).

I measured (or rather will measure again) my computer room - AND measure the
furniture in it (tall bookshelves, packing boxes, chairs, etc...) - modeled
everything in a POV file (just simple approximate pigments).

I stand in a corner and just look at the room (using only one eye - I HAVE
two eyes of course, but the comparison is meaningless if I use binocular
vision)... note where my perception "falls off" on the walls (I might try
putting marked stickynotes on the walls as markers later).

I render, then keep adjusting the POV camera angle until it seems as
identical as possible to what I perceive (deliberately ignoring too much
peripherial vision). As a check, I derive extremes - I find the "this is
DEFINITELY too much telephoto look", and I find the "this is DEFINITELY too
much wide angle", and then find the midpoint. Etc.

Post a reply to this message

Hi.
A more simple way to determine what would be your "natural focal length" 
could be to have an eye looking at something through a reflex camera equiped 
with a zoom, and the other directly. Zoom in and out so that you cannot see 
any difference between the images of each eye.
I did it when I bought mine and arrived at about 50mm.
Don't know if it's representative, though.


web.4188a6e3e293e888aab3f0cb0@news.povray.org...
> Le Forgeron:
>
> Thanks, I did something similar this morning (made a mistake in the
> measurements though, I need to do it over again).
>
> I measured (or rather will measure again) my computer room - AND measure 
> the
> furniture in it (tall bookshelves, packing boxes, chairs, etc...) - 
> modeled
> everything in a POV file (just simple approximate pigments).
>
> I stand in a corner and just look at the room (using only one eye - I HAVE
> two eyes of course, but the comparison is meaningless if I use binocular
> vision)... note where my perception "falls off" on the walls (I might try
> putting marked stickynotes on the walls as markers later).
>
> I render, then keep adjusting the POV camera angle until it seems as
> identical as possible to what I perceive (deliberately ignoring too much
> peripherial vision). As a check, I derive extremes - I find the "this is
> DEFINITELY too much telephoto look", and I find the "this is DEFINITELY 
> too
> much wide angle", and then find the midpoint. Etc.
>
>
>

Post a reply to this message

regdo wrote:

> A more simple way to determine what would be your "natural focal length" 
> could be to have an eye looking at something through a reflex camera equiped 
> with a zoom, and the other directly. Zoom in and out so that you cannot see 
> any difference between the images of each eye.
> I did it when I bought mine and arrived at about 50mm.
> Don't know if it's representative, though.

No, that is a totally false method. The reason is that most viewfinders 
don't show the full frame but a cropped portion - 90% of the full frame, 
for example. Also, there are differences in the magnification of the 
viewfinder. Then there is also the fact that focal length alone has 
nothing to do with FOV. 50mm lens on 8x10" large format camera has quite 
a different FOV than the same lens on a small sensor P&S digicam. I 
guess you meant a 35mm film (or "full size" sensor) camera?

IMHO, the whole concept of "normal" is quite misleading as human eye 
doens't have a strict field of view but one that degrades gradually to 
borders. Also the eyes are allways moving. The best bet is to do what Le 
Forgeron said: "My current view on this subject is: the final image 
should cover the same angle when viewed." But I don't understand why 
even that should be a goal. Just see what looks nice and use that...

Severi S.

Post a reply to this message

Severi Salminen <sev### [at] NOT_THISsibafi> wrote:
> Just see what looks nice and use that...


That's the best advice I've heard...

The beauty of POV is that you have control over every aspect, so if you want
it to look a certain way, write it that way.  I frequently put parameters
in my camera that have no basis in reality, but give me the view I want.
My "desert" IRTC entry is 1000x600 but my camera statement is using
something else (1000x500 IIRC) because I wanted the wider view. It meant I
had to scale the building slightly to get the proper vertical scale, but
everything else looked right without change.

RG

Post a reply to this message

Hi, All - thanks for replies....

Of course I very much agree, in general (" ... just use whatever looks right
...."). And of course there will be circumstances when you want to use
unusual camera settings.

But as long as I want to try to *determine in theory* what a "normal" camera
{ angle xx.x } is (which I will use *most of the time*), I want to be as
"close to the truth and reality as possible", and I want to standardize...
use it most of the time.

It isn't as simple as it sounds. I brought this up to a friend of mine that
works in a camera store. She basically said try three extremes (from a 35mm
SLR)... a 28mm "wideangle" lens = 75.4 degrees FoV (for an average lens of
this type), a 50mm "normal" lens = 46.8 degrees, and a 70mm
"short-telephoto" lens = 35 degrees; and by the way using an angle of 35
seems to be what the authors of the POV distribution tutorial scenes seem
to prefer for a "standard, normal view".

Now WHY did I enclose "wideangle", "normal", and "telephoto" in quotes in
the previous paragraph? Because, sure, a 28mm lens IS considered to be a
wideangle lens... but apparantly some (most?) professional photographers
believe that the 28mm *most* approximates a human eye (ONE eye - setting
aside binocular vision). If you check medical data on the net, thats true!

If a discrepency exists, it is due to the way that our eyes are *interfaced*
to our brains. If you look thru a 35mm SLR / 28mm Lens, eventhing is
distorted and "tilty" ... it is NOT being processed through any equivalent
of our complicated neural networks, before the camera presents an image to
us. BUT a single human eye IS closer to a 28mm than to a 50mm. ( ... and...
ditto and nod to a comment in a previous reply, perception fades to the
edges if you can keep your eyes "still").

Anyway... I've pretty much decided what I will use... thanks all.

Post a reply to this message