Tutorial: How To Create Stereograms With Computer


TUTORIAL: How to create Auto-Stereograms with Computer

P R E M I S E

A pair of stereoscopic images differs in their relative perspective because the two images
represent different vantage points. This perspective difference results in horizontal shifts
of corresponding near and far points. The horizontal shift, known as horizontal parallax or
disparity, is closely in relation with the width of the image. In order to obtain an optimum
stereoscopic effect within the limits of normal viewing, the maximum horizontal parallax in
the viewed image must be of about 1/24 of image width. Stereoscopic images can be obtained
with a computer with the aid of 3D programs to generate 2 images of the same object or scene,
as the result of differend visual angles. This different perspective can be genearated either
moving the camera, while his focus is fixed on the object, who remains motionless, or keeping
still the camera, while the object is rotated on the axis of its height (usually the y axis).
In this case the rotation of the object must change between 3 or 6 degrees and so also the
difference of the visual degree of the camera in the precedent case.

P R O C E D U R E

The stages for the building of a stereogram using a computer are two. The First Stage is the
correct creation of the two images that will become the stereogram. This creation can be made
with a 3D Program, using renderings of 3D scenes, or with a 2D Program, starting from every
single image. The Second Stage is the Correct Assembling of the two images previously created
so that the stereoscopic view can be led in the observer.

                                 F  I  R  S  T      S  T  A  G  E

                      Creation of the Pair of Images with a 3D  Program

In this first stage of tutorial, in example images,  it was used the software Lightwave 3D ®, 
Copyright © of Newtek Inc.,  but,  really, the expressed here concepts, can be performed with 
many other 3D programs.  Yes,  I really know that in Lightwave,  in the Layout Camera Control 
Panel,  there is  the "Stereoscopic Rendering" and "Eye Separationes" options,  but with only 
these you never  can truly understand  as the Visive Perception Stereoscopic Mechanism works. 
Besides,  the method written here is more accurate  instead of the Camera Automatic Settings. 
So, let's go on!  If you click with your mouse  on the images  of this tutorial,  you can see 
them greater  and  more in detail into a new window.  Warning:  This javascript routine, that 
shows greater images,  when you click with mouse on smaller images,  work well with Nescape ® 
browser, while it could have some malfunction with Explorer ® browser,  also if I've adjusted 
the javascript routine to it,  as the Explorer ® browser does not utilize pure Java language. 
For this,  I suggest you  to use Netscape ® browser to better look and read this my tutorial.

Create  a new scene  with the standard  default  values. 
From 'Objects' Panel  load in the scene all the objects 
you want. Look in the 'Perspective View'  that  all the 
objects were right loaded in your scene.

 Go in the 'Top View'  and  put the Camera  carefully in 
 front of your group of objects.

 Create now  a 'Null 
 Object',  with  the 
 option   'Add  Null 
 Object' and  put it 
 into  or  a  little 
 back  the center of 
 your object.  After 
 this,  go  in  'Top 
 View'  and  also in 
 'Side   View'   and 
 control  that  your 
 Null  is put right.

Select  the Camera  and  choice the 'Null Object' as the 
'Target' of the Camera, or track the Camera at your Null 
Object  with the 'Track to' function,  according  to the 
3D program  you  are utilizing  to make  this operation.

 Control in the 'Camera View'  that your group of objects 
 is well visible in the middle of the scene.

Come back in 'Top View',  select  the Camera and move it 
to left of cm.3,5 on X axis (X=-3,5cm).

 Make the rendering of the scene and after save the image 
 of rendering with the name of 'PhotoEyeSn.xxx'.

Come back in 'Top View',  select  the Camera and move it 
from the previous position  to right  of cm.7 on X axis, 
so that now the X value is positive of cm.3,5 (X=3,5cm).

 Make the rendering of the scene and after save the image 
 of rendering with the name of 'PhotoEyeDx.xxx'.

                      Creation of Pair of Images with a 2D Program

In this other part  of  this  tutorial,  in the example images,  it  will be utilized  the 2D 
software Photoshop ®, Copyright ©  of the Adobe Systems Inc.  This section of the tutorial is 
for many expert users  as  it is possible  to give  only  indicative  instructions  about the 
creation of the Pair of Stereographic Immages, starting from a single 2D image,  that it will 
be subdivided  at first  in several parts,  which will come  then modified,  pixel for pixel, 
bending them  to simulate a 3D perspective,  but making  all this procedure  only in 2D mode. 
Obviously every image  represents a separate history,  therefore its subdivision in parts, to 
put  on various layers  in Photoshop,  it  must  made  according  to the discretion of who is 
building the stereographic image.  For this we can here only suggest to make such subdivision 
thinking at the position in 3D space that those same parts  of image would occupy in the real 
world,  giving then a gradual inclination  and  the same parallactic movement  to those parts 
that  in the real would  be found  on the same horizontal line (Axis X)  in the sense  of the 
depth (Axis Z),  considering unininfluental the vertical axis (Axis Y).  What instead we will 
explain  here  is  like  being able  to bend one  of the parts of the image,  only using a 2D 
program,  in order to simulate  and  to generate the two 3D perspectives.  Repeating then the 
same process for all the several parts of  the image and grouping them then in the respective 
views,  the two images will be obtained and from them it will be possible finally to generate 
a stereogram, as it will be finally explained in the third and last part of this my tutorial.

For this section of the tutorial  we will utilize a part 
of our previous image,  but  we will start just from the 
image  of the frontal view,  using  it  like if it was a 
whichever 2D image  of start.  Therefore  load the image 
with Photoshop.  The part choice  of the image  on which 
you will work  it is the yellow polyhedron on the right. 
Remember  still  then  that  the procedure it would have 
repeated  for every other part  in which  your image had 
been dissected.  In your image moreover, the objects are 
on the same line  of depth,  so  that the inclination of 
the four colored solids,  put  on  ochre parallelepiped, 
will change,  but the horizontal distance  between  them 
will remain unchanged.

Remember instead that,  if these objects were put  in the real world along different lines of 
depth, also their horizontal distance must have varied, according to the following principle:  
if some of the objects  put on one  horizontal line are laterally approached between them, in 
the final  stereoscopic view they will seem  to approach the observer,  while  the ones moved 
away between them will seem to go away from the observer.

 Select  and  cut  with Photoshop your yellow polyhedron. 
 Remeber  to save the selection.  Copy then it in memory 
 with Edit/Copy from menu  or with CTRL+C from keyboard. 
 Paste then it in a new layer  with Edit/Paste from menu 
 or CTRL+V from keyboard.  A new layer will be therefore 
 generated.  Rename it as 'Main Yellow Polyhedron'. Drag
 twice this layer on duplication icon in the layers tool 
 and generate so two layers of copy,  that you will call 
 with  the name  of 'Yellow Polhyedron SN'  and  'Yellow 
 Polyhedron DX'.  Select  only  layer 'Yellow Polyhedron 
 DX' to try  to simulate the three-dimensional position, 
 as it will be from the view of your right eye.

Here it will be showed  only as to simulate  the view of 
right eye.  As the left eye (that  is  the layer 'Yellow 
Polyhedron SN'),  it will be sufficient  that you repeat 
the same operations,  but  giving  an opposite angle and 
inclination.  Well,  then  select only the layer 'Yellow 
Polyhedron DX',  magnified  it  with a zoom of 200%, and 
move  the view  of the polyhedron  to the center of your 
image, using the Tool 'Hand', or pressing the space bar. 
Select Filter/Distort/Shear  and  click  on  the  option 
'Wrap Around',  then move the higher and the lower point 
of the black line,  that  it  is  on  the  outlined grid
within the window,  slighty on the left of the grid,  as 
it is showed in the image here on right. It is necessary 
remember to maintain perpendicular the black line, while 
it is moved on the left. At this point click on Ok.

 Select then the window 'Info', always maintaining active 
 the selection  on the yellow polyhedron.  Click  now  on 
 Edit/Transform/Scale  and  press on the median handle on 
 right side  of the transformation window  and  push with 
 the mouse towards right,  until in the window 'Info' the 
 value of W=106% will not appear.  Give now Ok,  pressing 
 the Return key. Repeat the same operation with the layer 
 'Yellow Polyhedron SN', but give it an opposite angle of 
 inclination  and  you will obtain also the 3D simulation 
 of the view  of the left eye.  If  then  you repeat this 
 kind of operations  for all the parts  in which you will 
 have at the beginning subdivided your image and then you 
 add  and  fuse all the layers with the simulation of the 
 right eye in a first image and all those of the left eye 
 in a second image,  you will obtain  the two eyes views.

These 2 images will be then mounted correctly,  as it will be explained in the successive and 
last part of this tutorial,  in order to generate the final stereogram.  Obviously all the 2D 
technique  of the creation  of the two views,  starting from an one 2D image,  is only enough 
approximate and it can sure be improved.  For example, the new plugins 'Kai Power Tools ® 6', 
Copyright ©  of the MetaCreations Corporation,  and  more exactly  the 'Kpt Goo' and the 'Kpt 
Projector,  seem to be able in helping to improve,  if  adequately employed,  this process of 
simulated creation of the two perspective 3D views,  starting only from a 2D image.  The more 
important thing is to never forget to keep on with experiments...

                              S  E  C  O  N  D         S  T  A  G  E

                      Correct assembly of the Pair of Images with Photoshop

 The procedure is the same,  either for the two really 3D images,  created with a 3D program, 
 or for the two pseudo-3D images,  created starting only from a 2D image. Therefore this part 
 of the tutorial leaves the way  in which you have generated the two images,  that one of the 
 view of the left eye  and  that relative one to the right eye,  apart and is takes care only 
 exclusively  of the final construction of stereoscopic image,  called also Auto-Stereoscopy.

From Photoshop create,  with command 'New', a new image, 
of fair dimension.  Insert into it,  as layers, the left 
and the right image,  first  already made,  in this way: 
one at time, Load the 2 images,  use the command 'Select 
To (the CTRL+A), then 'Copy' (CTRL+C) and finally select 
the new image  and  use  the command  'Paste'  (CTRL+V). 
Rename the two new layers  as 'Image DX' and 'Image Sn'. 
Select the layer 'Image SN',  go  in the window  of tool 
'Info' and select Edit/Transform/Rotate. Point the mouse 
at the handle, low on right of image,  and push it down, 
until the horizon line,  that first it went above,  does 
not become again horizontal. Control the degrees of spin 
in the Info window (In this case A=0.6).  At  this point 
give Ok, pressing the Return key.

 Select  the  layer  'Image DX'   and   repeat  the  same 
 operation,  but  rotating  the  image  with  an opposite 
 angle:  therefore select Edit/Transform/Rotate and point 
 the mouse at the handle,  low on left of image, and push 
 it down, until also in this case the line of the horizon 
 in the image does not become horizontal, that is when in 
 the window  of 'Info' you will see  the value of A=-0.6. 
 Give the Ok, pressing  Return.

Activate  the layers 'Image SN' and 'Image DX' and  then 
select the layer more on,  that is 'ImageSN' and make it 
transparent to 50%,  putting  the Opacity trimmer  to 50 
value.  You  will see  that now  an image enough doubled 
will appear on your screen.

 Select  tool 'Move' (CTRL+V)  and  move the layer 'Image 
 SN' slighty  on right.  Repeat this operation with layer 
 'Image DX', but move it on left. Continue to move either 
 the first, or the second layer until the central objects 
 of the two images nearly succeed in coinciding.

Select still only the above Layer 'Image SN' and make it 
again opaque at 100%. With the tool 'Rectangular Marquee 
Tool' (CTRL+M) create  one  rectangular  selection  that 
includes only the designed parts of the two images,  the 
one Left (Sn)  and  the one Right (Dx). In order to make 
this,  you must before create  the selection only on the 
layer of an image.

 Then choice the other layer, but without coming out from 
 the  already  made  selection.  Make all  the  necessary 
 corrections  on  the selected area,  adding  (SHIFT+Tool 
 Selection)  or  removing  (STOP+Tool Selection)  on  the 
 selected area the parts that you do not need more.  When 
 you have obtained  a selected area  that it includes all 
 the two full designed parts of the two images, save this 
 selection  with 'Select/Save selection'.  Now  selection 
 will appear  in Channels window  with name of 'Alpha 1'. 
 Select the above Layer, that is 'Image SN' and then drag 
 the icon  of Alpha 1  channel  on  the  below button  of 
 Selection (it  has designed  over  an  outlined circle). 
 Select  now  'Select/Inverse' (SHIFT+CTRL+I)  and  after 
 this choose 'Edit/Cut' (CTRL+X).

 Repeat the same operation also for the other layer, the one 'Image Dx', that is move Alpha 1 
 Channel on Selection Button,  then 'Select Inverse' (SHIFT+CTRL+I),  and  finally 'Edit Cut' 
 (CTRL+X).  Therefore you have obtained the two full and of the same width and height images. 
 Select one whichever of the two layers of the images,  the one SN or the one DX, and put the 
 Alpha channel 1 on Selection Button.  Choice 'Image/Crop' and press the mouse to give Ok. So 
 you have eliminated the empty and not necessary  parts from yours two images.

Click on menu 'Image/Canvas Size' and choice the central 
position of the image.  Read the value of Width, in this 
case 540 pixels,  and multiply it for 2.3 approximately. 
Therefore  write  the value  of 1200 pixels.  Go  on the 
value of Height and multiply  it  for 1.2 approximately. 
Therefore you write  500 pixels. Go then in the layer of 
background  and  fill it  with black color (choose black 
color  from palette  and  press togheter ALT+BACKSPACE). 
(Remember that,  if you want,  you can also do not leave 
any space  between the two images,  but you can put them 
one attacked  at  the other).  Select  layer 'Image SN', 
choose tool 'Move' (CTRL+V) and move it  with the mouse, 
or with keyboard arrows,  only horizontally and on left.

Repeat this operation with layer 'Image DX', but this time move it laterally on right, always
using CTRL+V. At the end you will have the two very visible images, one on left and the other
on right, with a short black space around them. So you have created your Auto-Stereogram.
Remeber that, if you want see it correctly, the width of image will have to be of cm.14
approximately, because the distance between our two eyes in the real world is approximately
of cm.6,5 and therefore about the same (or a multiple or a submultiple of it) it must be the
distance between the centers of the two images. Therefore, if you want to print your image
correctly, you must choice the menu 'Image/Image Size', deselect 'Resample Image' and insert
value of cm.14 as Width. The value of Height will be adapted automatically to a new corrected
value, in proportion to the new width inserted. Therefore you will be able to print a really
wide cm.14 image. If, on the contrary, you want view it on screen, you must put attenction to
the settings of resolution of your monitor. For a screen of 800 x 600 pixels, the final image
will have to be of 350 pixels of width. A proportionally optimal dimension is of 350 x 150
pixels. But the consideration of that is optimal, obviously changes from person to person.
For the stereographic images on screen, for example, the optimal dimension I like more is
approximately of cm.9 of width, equivalents at an image of about 230 x 100 pixels, always for
a screen resolution of 800 x 600 pixels. Therefore a good rule, when will show yours
stereograms on screen is that one to introduce, on one screen, the classic version of cm.14
of width and also one proportionally more small. So you will be sure that everyone will be
able to well see yours stereograms. This big tutorial ends here and I, Francesco Franceschi,
the author, invite you to try it and to create yours originals stereogrammi. I hope you can
have a great satisfaction creating them. Good job!

I have written this page in Italian language and after I have translated it in English language.
If you find some my translation's error, excuse me and send me an e-mail with the correct version. Thanx !!
Please Remember That Everything On This Web Site Is Copyrighted.