I find that the 'rotation' method of producing the stereo views can be
confusing because if you have z-clipping on, different atoms get clipped in
the L & R images and you see the corresponding atoms which are still
visible in the other image in mono, while of course all the rest are in
stereo.  This gives me a headache, particularly if I have to look
cross-eyed at the same time!

The solution is to skew, not rotate (this brings to mind from the dim
distant past Gerard Bricogne's manual on his 'skew planes' package called
'The Joy of Skewing' - which of course has nothing whatsoever to do with
producing stereo images).

Assuming the origin is in the centre, the rotation matrix about the
vertical y axis (assuming your monitor is mounted in the conventional way)
looks like this:

[ cos(a)   0   sin(a) ]
[     0       1      0     ]
[ -sin(a)   0  cos(a) ]

where a = +- 3 deg. for the L & R images (I leave it as an exercise for the
reader to work out which is which!).

This matrix gives for the rotated x & z co-ordinates:

x' = x cos(a) + z sin(a)
z' = -x sin(a) + z cos(a)

so z' is different for the L & R images, particularly for atoms with large
|x| near the L & R sides of the screen (because the sine changes sign)
which get clipped differently in the two images.

If I recall correctly the skew matrix looks like this:

[ cos(a)   0   sin(a) ]
[     0       1      0     ]
[     0       0      1     ]

so:

x' = x cos(a) + z sin(a)  (i.e. same as above)
z' = z

So now whereas x' is still different for the L & R images (and gives you
the stereo effect), z' is the same so both eyes see exactly the same view
even if z clipping is on.

You may of course argue that skewing will distort the image whereas
rotation won't.  This is true; however the amount of distortion is tiny and
not noticeable (and certainly well worth the price of making the stereo
image much less confusing).

Cheers

-- Ian

On 18 January 2015 at 19:34, Ronald E Stenkamp <[email protected]>
wrote:

> Are most stereo images now for cross-eyed viewing?  I thought they were
> for wall-eyed viewing.
>
> Perhaps a warning would be helpful for people starting out at looking at
> published stereoviews.  If you look at a stereoview constructed for
> wall-eyed viewing but look at it with crossed eyes, you'll change the
> handedness of the object.  And if you're showing surfaces, they get turned
> inside-out (or something like it).  I usually get a headache soon after
> mixing modes like this and only know that the surfaces are messed up.
>
> Also, in answer to one of Jeorge's questions, the two images in
> stereoviews differ by a small rotation about a vertical axis.  The two
> images are what each of your eyes would see if looking at a single object.
> Because your eyes are separated by about 2.25 inches (I'm a stubborn
> non-metrical American...), the left- and right-eye views differ slightly.
> The amount also depends on how close your eyes are supposed to be from the
> object.  I think long ago things were worked out so the rotation is 6
> degrees and that corresponds to the viewer-object distance being about 30
> inches.  If you place the left eye view on the left, you need to look at
> the two images in wall-eyed mode.  If you place the left eye view on the
> right, you need to cross your eyes to generate the stereo image.
>
> For those of us who can view stereoimages without the assistance of
> glasses or computers, life is good.  I recommend developing the ability to
> do that.
>
> Ron
>
>
> On Sun, 18 Jan 2015, Jim Fairman wrote:
>
>  You can create stereo images for publications in pymol:
>> http://www.pymolwiki.org/index.php/Stereo_ray
>>
>> Adding labels and getting them to "float" at the correct depth within the
>> image can be tricky.
>>
>> As for visualizing the stereo images, you can either practice alot and get
>> good at cross eyed stereo viewing, or you can buy a pair of glasses to
>> assist
>> you in seeing the 3d effect. If you google "cross eyed stereo glasses",
>> you
>> will get many options to purchase. Old chemistry texts used to come with a
>> pair, but I'm not sure that students actually purchase textbooks anymore.
>>
>>
>> On Sat, Jan 17, 2015 at 23:50 PM, jeorgemarley thomas <
>> [email protected]>
>> wrote:
>>       Dear all,
>> First of all sorry to put this off topic and silly question on bb. Can
>> anybody suggest me, how to create a stereo image and how it is different
>> from the normal. How can I visualize it, if anybody has answer for this
>> please suggest me its significance in analysis. Thank you very much in
>> advance.
>>
>> Thanks
>>
>> Jeorge
>>
>>
>>
>> --
>> Sent from MetroMail
>>
>>
>>

Reply via email to