3D broadcsasts

In article , John
Legon writes
R. Kennedy McEwen wrote:
I was really surprised with some shots I took looking across the
Grand Canyon earlier this year. Although the other side is on
average 10 miles away, the 3D effect was really impressive. Of
course the sides of the far canyon walls range from 8 to 15 miles so
there is a lot of depth variation, but I was surprised how well it
worked with only a 75mm baseline. I was really glad I had taken it
along as I only intended to use it for relatively close images.

The difference in parallax from a 75 mm baseline to objects 10 and 15
miles away must be negligible, surely?

You would think so but, as I said, I was surprised that it works.
--
Kennedy

On 03/08/2012 08:15, R. Kennedy McEwen wrote:
In article , John
Legon writes
R. Kennedy McEwen wrote:
I was really surprised with some shots I took looking across the
Grand Canyon earlier this year. Although the other side is on
average 10 miles away, the 3D effect was really impressive. Of
course the sides of the far canyon walls range from 8 to 15 miles
so there is a lot of depth variation, but I was surprised how well
it worked with only a 75mm baseline. I was really glad I had taken
it along as I only intended to use it for relatively close images.

The difference in parallax from a 75 mm baseline to objects 10 and 15
miles away must be negligible, surely?

You would think so but, as I said, I was surprised that it works.

the fuji camera has an auto adjustment to get the separation to look the
best. It is also manually adjustable.

In article ,
John Legon wrote:

The difference in parallax from a 75 mm baseline to objects 10 and 15
miles away must be negligible, surely?

Well, let's see. If the baseline is b and the distance to the object
is d with b d, an object directly ahead of one "eye" will be at an
angle b/d (radians) off the axis for the other "eye". For two such
objects at d1 and d2, the difference in angle will be b/d1 - b/d2. If
the angle of view of the camera is v, this will be a proportion
(b/d1 - b/d2)/v of that angle. If the horizontal pixel count of
the camera is p, the difference in disparity in pixels will be

pb/v (1/d1 - 1/d2)

So for b = 75mm, d1 = 8 miles ~= 13km, d2 = 15 miles ~= 23km, v = 0.3
radians, and p = 4,000 we would get a disparity difference of 0.03
pixels.

I suppose it's just about possible that one can resolve and interpret
such a difference, given that features will be present in many
consecutive scan lines, but it's rather surprising.

Could you (the photographer) measure the difference or post the photo
so someone else can?

-- Richard

John Legon schrieb:
Jim Lesurf wrote:
How far apart are the lenses on the 3D cameras? Presumably that would let
us know the distances at which a point shifts sideways by one pixel in
the
two pixel-arrays.

With the 3D cameras as used in broadcasting the spacing can be varied.
Although it can be the same as for human eyes - about 65 mm - the amount
of parallax I see in some 3D material suggests to me that it is
sometimes much greater.


Well, it depends - the adjustable 3D rigs are using two normal cameras
behind a half mirror box which is bending one view 90 degrees down (for
the right side view). For distant objects the separation of the two
cameras must get higher than normal eye-parallaxe for decent 3D effects.

The professional Panasonic 3D video camera 3DA1 has two lenses
side-by-side 60 mm apart, and this seems to be used at most olympic
venues shown at night. So the wide area shots are less impressive than
the ones close by the camera, and not all camera operators are
experienced 3D "stereographs"...

Klaus wrote:
John Legon schrieb:
Jim Lesurf wrote:
How far apart are the lenses on the 3D cameras? Presumably that would
let
us know the distances at which a point shifts sideways by one pixel in
the
two pixel-arrays.

With the 3D cameras as used in broadcasting the spacing can be varied.
Although it can be the same as for human eyes - about 65 mm - the amount
of parallax I see in some 3D material suggests to me that it is
sometimes much greater.


Well, it depends - the adjustable 3D rigs are using two normal cameras
behind a half mirror box which is bending one view 90 degrees down (for
the right side view). For distant objects the separation of the two
cameras must get higher than normal eye-parallaxe for decent 3D effects.

The type of 3D rig you describe with horizontal and vertical cameras was
shown in a Sky 3D demo (I think) some time ago on Astra 19.2 E.

Do you know roughly what the camera separation might be for the scene I
posted the other day ? I converted this from side-by-side to anaglyph,
so the colour fringes indicate the parallax:

http://www.john-legon.co.uk/temp/jediscene1.jpg

It seems to me that the "baseline" must be something like 50 cm to give
the variation in parallax (negative to positive) from the foreground to
the background, but I don't know whether this is probable or possible.


The professional Panasonic 3D video camera 3DA1 has two lenses
side-by-side 60 mm apart, and this seems to be used at most olympic
venues shown at night. So the wide area shots are less impressive than
the ones close by the camera, and not all camera operators are
experienced 3D "stereographs"...


Thanks for the interesting info...

In article , Gary
writes
On 03/08/2012 08:15, R. Kennedy McEwen wrote:
In article , John
Legon writes
R. Kennedy McEwen wrote:
I was really surprised with some shots I took looking across the
Grand Canyon earlier this year. Although the other side is on
average 10 miles away, the 3D effect was really impressive. Of
course the sides of the far canyon walls range from 8 to 15 miles
so there is a lot of depth variation, but I was surprised how well
it worked with only a 75mm baseline. I was really glad I had taken
it along as I only intended to use it for relatively close images.

The difference in parallax from a 75 mm baseline to objects 10 and
15 miles away must be negligible, surely?

You would think so but, as I said, I was surprised that it works.

the fuji camera has an auto adjustment to get the separation to look
the best. It is also manually adjustable.

That is the *DISPLAY* separation, which adjusts for comfortable viewer
eye convergence. It has a fixed *OPTICAL* separation of 75mm.

It is the optical separation which creates the 3D effect and which is
the subject of the discussion.
--
Kennedy