I'm trying to wrap my head around what I should use for camera angle settings
when I render for a screen with 1366 x 768 resolution.  It's unclear to me how
the camera angle coupled with the render size determines what is visible, and
what the "correct" proportion should be.



Also, I've noticed with some of my scene renders that the view looks a little
restricted.  Looking at the real world, I'm able to take in a fairly wide view,
and when rendering something similar, it has that severely cropped feeling.

I've played with the camera angle, and notice that above a certain value (80-90
deg as far as I can tell) I start to get a tunnel-like distortion.

I'm looking to experiment with rendering some scenes that more closely mimic a
"real" field of view, and was wondering if anyone has any recommendations for
giving a better sense of space to the scenes.

http://en.wikipedia.org/wiki/Field_of_view

Humans have an almost 180-degree forward-facing horizontal field of view, ....

.... binocular vision, which is important for depth perception, covers only 114
degrees (horizontally) of the field of vision in humans;[2] the remaining
peripheral 60-70 degrees have no binocular vision (because only one eye can see
those parts of the field of view).

I have the feeling that I'd need to balance image size, aspect ratio, camera
angle, and probably add some focal blurring (which I have yet to even try).

Thoughts?  Suggestions?  Fait accompli examples?

Thanks

Post a reply to this message

Do you know Edouard Poor's 35mm camera macro set? Not sure if it would 
solve your problem but it might help. There also is a nice little macro 
by DrJohn which can correct the 'falling building syndrome'; a bit 
beside the point but still valuable.

 From my personal experience, I often use a camera angle of 80 when 
rendering landscapes, for instance in my Gancaloon scenes. In 
combination with:

right x*image_width/image_height

I feel I get the wide view I expect. Generally I just use the standard 
camera, not the wide_angle or such.

Something that might help too is playing with the camera look_at 
distance in relation to its location.

Zooming in or out of a scene can be achieved by defining a zoom factor 
Zoom, and multiply that with: direction z*Zoom

For more intimate scenes I prefer a camera angle of 50 or even 35.

Hope this helps somewhat.

Thomas

Post a reply to this message

Thomas de Groot <tho### [at] degrootorg> wrote:

> I often use a camera angle of 80 .... In combination with:
> right x*image_width/image_height

So, I guess I was under the impression that the above equation was the default
setting when you declare a camera object?

I guess I'm struggling a little bit - I seem to be at that awkward point where I
know enough to ask questions, but know too little to ask GOOD questions.  :\

all of these attributes seem to be interrelated to one another in a sort of
circular-logic that I can't untangle.   I see you declaring RIGHT - how does
that affect UP - or doesn't it?

> Something that might help too is playing with the camera look_at
> distance in relation to its location.

Hmm.  I'll have to experiment with that, since I didn't realize it had any
effect.  I typically just use it as a camera orientation vector.  I generally
don't care if the vector length is 0.01 or 100,000 so long as they lie on the
same line of sight.  What should I be expecting / looking for?

> Zooming in or out of a scene can be achieved by defining a zoom factor
> Zoom, and multiply that with: direction z*Zoom

The diagram I saw of the perspective camera seemed to suggest that the direction
vector is half the length of the camera to look_at length.  True?
The diagram didn't give a good explanation of right and up in relation to the
image size.

(I also have a Tina-CHeP question for you... how/where to do that?)

Thanks a lot for the suggestions!

Post a reply to this message

Am 28.08.2014 15:44, schrieb Bald Eagle:
> I'm trying to wrap my head around what I should use for camera angle settings
> when I render for a screen with 1366 x 768 resolution.  It's unclear to me how
> the camera angle coupled with the render size determines what is visible, and
> what the "correct" proportion should be.

If you specify a camera angle, it directly specifies the horizontal 
field of view; the aspect ratio (relative length of the right and up 
vectors, which should ideally fit the image aspect ratio) determines the 
vertical field of view.

> Also, I've noticed with some of my scene renders that the view looks a little
> restricted.  Looking at the real world, I'm able to take in a fairly wide view,
> and when rendering something similar, it has that severely cropped feeling.
>
> I've played with the camera angle, and notice that above a certain value (80-90
> deg as far as I can tell) I start to get a tunnel-like distortion.

Try the ultra_wide_angle camera; I consider it far superior to the 
pinpoint projection of the standard perspective camera.

> I have the feeling that I'd need to balance image size, aspect ratio, camera
> angle, and probably add some focal blurring (which I have yet to even try).

One hint for focal blurring: Try to figure out what size an eye's pupil 
would be in scene dimenstions, and set the aperture value to something 
in that order of magnitude.

Post a reply to this message

On 28/08/2014 17:32, Bald Eagle wrote:
> Thomas de Groot <tho### [at] degrootorg> wrote:

> The diagram I saw of the perspective camera seemed to suggest that the direction
> vector is half the length of the camera to look_at length.  True?

No. Look_at is only used for the alignment of the center of the image
with the line going from camera position to look_at position.

Direction can be used to provides some different angle values, as long
as angle is not specified.

See http://wiki.povray.org/content/Reference:Camera#Perspective_projection

> http://wiki.povray.org/content/Reference:Camera#Perspective_projection

When angle is specified, the up & right vectors are internally
recomputed from the direction vector (lengths are adjusted).

> The diagram didn't give a good explanation of right and up in relation to the
> image size.

up & right, in perspective camera, are irrelevant about the visible
field (but if the angle was not specified, the ratio between right &
direction will have an impact).

Ratio of up & right should matches the ratio of the image size time the
ratio of the pixel. (so with traditional square pixel, the ratio of up &
right should be identical to the height & width ratio )... time or
divide, I'm often confused, but square pixel have a nice 1, so it's
irrelevant most of the time for me.

Beware when comparing povray's camera's angle (of perspective camera)
and the photographic angle:
Povray's angle is the one between the horizontal picture dimension and
the opening of the prism, whereas the photographic is along the diagonal
(which means 1:2, 4:3 and 16:9 would have different angle to have the
same object on the left/right border)


> 
> (I also have a Tina-CHeP question for you... how/where to do that?)
> 
> Thanks a lot for the suggestions!
> 
> 
> 
I have seen something like group irtc.general which talks about TC-RTC.

-- 
IQ of crossposters with FU: 100 / (number of groups)
IQ of crossposters without FU: 100 / (1 + number of groups)
IQ of multiposters: 100 / ( (number of groups) * (number of groups))

Post a reply to this message

Am 28.08.2014 17:32, schrieb Bald Eagle:
> Thomas de Groot <tho### [at] degrootorg> wrote:
>
>> I often use a camera angle of 80 .... In combination with:
>> right x*image_width/image_height
>
> So, I guess I was under the impression that the above equation was the default
> setting when you declare a camera object?
>
> I guess I'm struggling a little bit - I seem to be at that awkward point where I
> know enough to ask questions, but know too little to ask GOOD questions.  :\
>
> all of these attributes seem to be interrelated to one another in a sort of
> circular-logic that I can't untangle.   I see you declaring RIGHT - how does
> that affect UP - or doesn't it?

What the up, down and direction vectors do (at least when you also 
specify an angle and look_at point) is simply define the coordinate 
system in which the camera is set up before being rotated to the look_at 
point and zoomed to fit the angle, as well as the aspect ratio of the 
image; these three vectors should be perpendicular to each other, unless 
you are after some special effect (such as a that of a tilt-shift lens); 
the ratio of the right vector's length vs. the up vector's should match 
the ratio of the image width vs. height (unless you use an image format 
that has non-square pixels, but you rarely need those these days). If 
you're not using angle, the ratio of the direction vector's length to 
that of the right vector's also determines the field of view, but with 
angle specified the direction vector's length is entirely irrelevant.


>> Something that might help too is playing with the camera look_at
>> distance in relation to its location.
>
> Hmm.  I'll have to experiment with that, since I didn't realize it had any
> effect.  I typically just use it as a camera orientation vector.  I generally
> don't care if the vector length is 0.01 or 100,000 so long as they lie on the
> same line of sight.  What should I be expecting / looking for?

The distance to the look_at point is irrelevant, the look_at point just 
determines the camera's orientation.

The camera is oriented by first rotating the camera around the up vector 
until the camera looks at the look_at point's projection onto the 
right/direction plane, then tilting the camera up or down to look at the 
actual look_at point.

Post a reply to this message

Thanks everyone.  I'll digest this all, and run some experiments to see what I
get.
Thankfully, most of the scenes I work on are based on real physical objects and
so I usually use a 1 POV-Unit to 1 inch scale, so the pupil size is a snap.

Post a reply to this message

> I'm trying to wrap my head around what I should use for camera angle settings
> when I render for a screen with 1366 x 768 resolution.  It's unclear to me how
> the camera angle coupled with the render size determines what is visible, and
> what the "correct" proportion should be.
>
>
>
> Also, I've noticed with some of my scene renders that the view looks a little
> restricted.  Looking at the real world, I'm able to take in a fairly wide view,
> and when rendering something similar, it has that severely cropped feeling.
>
> I've played with the camera angle, and notice that above a certain value (80-90
> deg as far as I can tell) I start to get a tunnel-like distortion.
>
> I'm looking to experiment with rendering some scenes that more closely mimic a
> "real" field of view, and was wondering if anyone has any recommendations for
> giving a better sense of space to the scenes.
>
> http://en.wikipedia.org/wiki/Field_of_view
>
> Humans have an almost 180-degree forward-facing horizontal field of view, ....
In fact, it's a little larger than 180, closer to 190 to about 200 in 
some persons.


>
> .... binocular vision, which is important for depth perception, covers only 114
> degrees (horizontally) of the field of vision in humans;[2] the remaining
> peripheral 60-70 degrees have no binocular vision (because only one eye can see
> those parts of the field of view).
>
> I have the feeling that I'd need to balance image size, aspect ratio, camera
> angle, and probably add some focal blurring (which I have yet to even try).
>
> Thoughts?  Suggestions?  Fait accompli examples?
>
> Thanks
>
>


When using angle larger that 60, it's often advisable to use the 
ultra_wide_angle camera type. That camera will reduce the perceived 
distortion at the expance of causing straight lines to bend.

panoramic, cylinder 1 or 3 and spherical are also to be concidered.

Post a reply to this message

Le 28/08/2014 17:53, clipka a écrit :
> the aspect ratio (relative length of the right and up vectors, which
> should ideally fit the image aspect ratio) determines the vertical field
> of view.

pinpointing: as long as the expected final pixels are square.

From memory, when doing DVD image (720x576 for Pal, IIRC), the
pictures/movie could be displayed as 16:9 or 4:3 according to the right
bit at the right place. yet, 720x576 is neither 4:3 or 16:9, as it is 5:4.

(for NTSC, the resolution is 720x480, a 3:2 ratio, not a square pixel
either)

So, to avoid distortion, the up & right ratio should be the same as the
viewed picture, and the height & width might be whatever is needed for
the storing media.

-- 
Just because nobody complains does not mean all parachutes are perfect.

Post a reply to this message

On 28-8-2014 18:15, clipka wrote:
> The distance to the look_at point is irrelevant, the look_at point just
> determines the camera's orientation.

Sorry. My bad. Somehow I mixed that up during writing with the focal 
length parameter used in the 35mm camera of Edouard Poor.

THomas

Post a reply to this message