 |
|
|
|
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Am 12.01.2016 um 22:06 schrieb Stephen:
>>> How are you going to drive your "Cardboard", what software?
>>
>> I don't understand what you mean?
>
> Once you have made your 3d Stereo image. How are you going to use Google
> Cardboard to view it? Or have you not got to that point yet?
> I tried cardbord a few months ago and I think that downloaded a few demo
> apps one had a gallery IIRC. Unfortunatly I've uninstalled tha apps and
> can't remember which ones they were.
Judging from the article description, I /think/ his choice of software
is called "printer driver" ;)
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
On 1/12/2016 4:06 PM, Stephen wrote:
> Once you have made your 3d Stereo image. How are you going to use Google
> Cardboard to view it? Or have you not got to that point yet?
> I tried cardbord a few months ago and I think that downloaded a few demo
> apps one had a gallery IIRC. Unfortunatly I've uninstalled tha apps and
> can't remember which ones they were.
>
I am not using Google Cardboard. I am going to print the pictures and
use this:
http://www.amazon.com/gp/product/B004EDB1S4?keywords=stereoscope&qid=1452568122&ref_=sr_1_90&sr=8-90
Mike
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
> On 1/12/2016 2:17 PM, clipka wrote:
>> By /adding/ a "look_at" statement? ;)
>>
>> For starters you might get away ok without look_at.
>>
>> What you really need to do is translate the camera a bit to the left
>> (for the
>> left eye) or the right (for the right eye) -- which is actually a deal
>> easier if
>> you don't use look_at, because with that statement you'd have to
>> compute your
>> effective left/right axis "manually".
>
> Okay, but I was thinking that rotating by a small amount would be
> better. Otherwise the point of interest gets translated too. Should I
> use real-world measurements for the distance between the "eyes"?
Depending on your scene, and the location of the camera, rotating the
camera can result in uterly unreasonable offsets, like accidently
diverging camera axis or excessive convergence.
It's usualy beter to translate the camera /after/ you set the look_at
point for images like paysages or points of interest a fair distance
from the camera.
If the point of interest is close to the camera, then, it may beter to
translate the camera, then, set the look_at point at the same location
for both views. This correspond to the natural converging of the eyes
when looking to something close.
>
>>>
>>> Also, is there a particular angle of view I should be aiming for?
>>
>> That depends on the (apparent) angle at which the image will be
>> visible using
>> that contraption.
>>
>>
>
> I don't understand what you mean. By angle of view I mean the camera
> angle. Is there some natural angle that is most like human vision?
Yes, the camera's horizontal field ov fiew in degrees.
dependent on how the image is to be viewed.
>
>
> Mike
Alain
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
On 1/12/2016 10:07 PM, Mike Horvath wrote:
> On 1/12/2016 4:06 PM, Stephen wrote:
>> Once you have made your 3d Stereo image. How are you going to use Google
>> Cardboard to view it? Or have you not got to that point yet?
>> I tried cardbord a few months ago and I think that downloaded a few demo
>> apps one had a gallery IIRC. Unfortunatly I've uninstalled tha apps and
>> can't remember which ones they were.
>>
>
> I am not using Google Cardboard. I am going to print the pictures and
> use this:
>
>
http://www.amazon.com/gp/product/B004EDB1S4?keywords=stereoscope&qid=1452568122&ref_=sr_1_90&sr=8-90
>
>
Oh! I saw "cardboard" in the title and jumped to conclusions. That makes
more sense now..
--
Regards
Stephen
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Mike Horvath <mik### [at] gmail com> wrote:
> On 1/12/2016 4:06 PM, Stephen wrote:
>
> I am not using Google Cardboard. I am going to print the pictures and
> use this:
>
>
http://www.amazon.com/gp/product/B004EDB1S4?keywords=stereoscope&qid=1452568122&ref_=sr_1_90&sr=8-90
>
A bit of trivia: There's another stereoscope on that Amazon page called the
'Owl' viewer-- designed by Brian May, the lead guitarist of the supergroup QUEEN
(!) Apparently, he is an avid 3D-stereoscope fan.
About 'convergence' of the two camera views: When I was younger, I did lots of
3D photo experiments, and found that the best 3D effect (for me at least) was to
have both camera views looking off into *infinity* (that is, no 'convergence' of
the cameras onto a nearer object) and to just move one of the cameras laterally
a couple of inches. This produced less eyestrain when viewing the 3D, and didn't
produce any 'odd' convergence-related perspective shifts of other objects in the
scene. In other words, I never converged the two cameras to look at a 'closer'
object, even if that object was the focus of attention.
Just my two-cents worth.
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
"Kenneth" <kdw### [at] gmailcom> wrote:
>
> About 'convergence' of the two camera views: [clip]... I and found that the
> best 3D effect (for me at least) was to
> have both camera views looking off into *infinity* (that is, no 'convergence' of
> the cameras onto a nearer object) and to just move one of the cameras laterally
> a couple of inches. This produced less eyestrain when viewing the 3D, and didn't
> produce any 'odd' convergence-related perspective shifts of other objects in the
> scene...
Let me try and explain that a little better: Converging the two camera views
onto a nearer object can produce unwanted parallax shift of more distant objects
(at least in a real camera lens), the result being that those further-away
objects don't 'line up' correctly (and have unequal 'lens distortion') when
viewed in 3D-- especially when using a wide-angle lens setting. It's an odd
effect to describe, but it results in eye strain.
POV-Ray's typical perspective camera is essentially a 'pinhole' camera-- not
quite the same behavior as a real lens (the lens distortions are... different)
so it might be worth experimenting with 'convergent' views vs. non-convergence
views, to see if there is any real difference. I don't know; strange as it
sounds, I've never made a 3D scene in POV-Ray (!) It's now on my to-do list....
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
> Let me try and explain that a little better: Converging the two camera views
> onto a nearer object can produce unwanted parallax shift of more distant objects
> (at least in a real camera lens), the result being that those further-away
> objects don't 'line up' correctly (and have unequal 'lens distortion') when
> viewed in 3D-- especially when using a wide-angle lens setting. It's an odd
> effect to describe, but it results in eye strain.
>
> POV-Ray's typical perspective camera is essentially a 'pinhole' camera-- not
> quite the same behavior as a real lens (the lens distortions are... different)
> so it might be worth experimenting with 'convergent' views vs. non-convergence
> views, to see if there is any real difference. I don't know; strange as it
> sounds, I've never made a 3D scene in POV-Ray (!) It's now on my to-do list....
>
>
>
>
Having a convergence can be beter, for cases when the center of interest
is close, less than 10 times the spacing of the cameras, and the
background don't contain far objects or is highly out of focus.
Think about 3D macro-photography. Very close, small, objects, tight area
of focal sharpness.
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Mike Horvath <mik### [at] gmailcom> wrote:
> Are there tutorials on how set up POV-Ray to make such images? I'm not
> using the look_at keyword in my camera statement. I am using up, right,
> and direction instead. How can I compensate for this?
I'm not sure if I'm replying to the right post, but anyway. If you have MegaPOV,
I wrote a macro to generate a cross-eye pair image in a single parse and render,
using the camera pigment feature. I've listed it below for interest - if you're
after a parallel pair you can just swap the sign of the CamX vectors in the
location statements of each pigment camera. I can't remember exactly but I think
you just set the macro parameters as if for a single view, call the macro
instead of making a camera, then use double the horizontal resolution for the
render size (I don't really run MegaPOV any more these days so I can't check
quickly).
It generates images like these:
http://news.povray.org/povray.binaries.images/thread/%3C4f0b6ac6@news.povray.org%3E/?ttop=374865&toff=50
If this isn't any use to you, never mind, maybe someone else will find it
interesting!
Bill
----------------------------------------
// Stereo pair camera macro
// needs megapov for camera view pigment
#version unofficial megapov 1.21;
#macro StereoPair(CPos, CLook, CAngle, CRight, CUp, BaseLine, Border)
// camera coord system
#local CamZ = vnormalize(CLook - CPos);
#local CamX = vnormalize(vcross(CUp, CamZ));
#local CamY = vnormalize(vcross(CamZ, CamX));
// replacement camera
camera {
location CPos
up CUp
right x*CRight
angle CAngle
look_at CLook
}
// coordinates for mesh screen
#local XDist = tan(radians(CAngle/2));
#local YDist = XDist / CRight;
#local LeftP1 = CPos + CamZ - XDist*CamX - YDist*CamY;
#local LeftP2 = CPos + CamZ - YDist*CamY;
#local LeftP3 = CPos + CamZ + YDist*CamY;
#local LeftP4 = CPos + CamZ - XDist*CamX + YDist*CamY;
#local RightP1 = CPos + CamZ - YDist*CamY;
#local RightP2 = CPos + CamZ + XDist*CamX - YDist*CamY;
#local RightP3= CPos + CamZ + XDist*CamX + YDist*CamY;
#local RightP4 = CPos + CamZ + YDist*CamY;
// mesh screen
// left
mesh {
triangle { LeftP1, LeftP2, LeftP3 uv_vectors <0, 0>, <1, 0>, <1, 1> }
triangle { LeftP1, LeftP3, LeftP4 uv_vectors <0, 0>, <1, 1>, <0, 1> }
texture {
uv_mapping
pigment {
camera_view {
location CPos + CamZ*1.1 + CamX*BaseLine/2
up CUp
right x*CRight/2
angle CAngle/2
look_at CLook }
}
finish { ambient 1 }
}
no_shadow
no_reflection }
// right
mesh {
triangle { RightP1, RightP2, RightP3 uv_vectors <0, 0>, <1, 0>, <1, 1> }
triangle { RightP1, RightP3, RightP4 uv_vectors <0, 0>, <1, 1>, <0, 1> }
texture {
uv_mapping
pigment {
camera_view {
location CPos + CamZ*1.1 - CamX*BaseLine/2
up CUp
right x*CRight/2
angle CAngle/2
look_at CLook }
}
finish { ambient 1 }
}
no_shadow
no_reflection }
#local R = Border * vlength(RightP2 - LeftP1);
union {
cylinder { LeftP1, RightP2, R }
cylinder { RightP2, RightP3, R }
cylinder { RightP3, LeftP4, R }
cylinder { LeftP4, LeftP1, R }
cylinder { LeftP2, LeftP3, R/2 }
pigment { rgb 0 }
finish { ambient 0 diffuse 0 }
no_shadow
no_reflection
}
#end
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
I still don't understand what is the difference between a cross-eye pair
and a parallel pair? Also, which should I be making for the glasses I
ordered?
http://www.amazon.com/gp/product/B004EDB1S4?keywords=stereoscope&qid=1452568122&ref_=sr_1_90&sr=8-90
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Mike Horvath <mik### [at] gmailcom> wrote:
> I still don't understand what is the difference between a cross-eye pair
> and a parallel pair? Also, which should I be making for the glasses I
> ordered?
>
>
http://www.amazon.com/gp/product/B004EDB1S4?keywords=stereoscope&qid=1452568122&ref_=sr_1_90&sr=8-90
A parallel pair is when you have the left eye image on the left, and the right
eye image on the right. If the separation is small, you can view these by
looking 'through' them and resolving the 'third' image in between to get the
stereo effect. This is what the viewing glasses use. This is also how dot
stereograms or 'magic eye' pictures work.
A cross-eye pair is swapped; the left eye image is on the right, and the right
eye image on the left. These can be viewed by looking 'in front of' them, by
going slightly cross-eyed to resolve the 'third' image in between to get the
stereo effect. This can be easier than the parallel method, and is not
restricted so much by image size / separation. I don't know if any devices exist
to help viewing these.
hope that helps!
Bill
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
|
|
