| Give relief to your images | ||
Requires
Bryce, Photoshop (or any software able to handle RVB channels), glasses
for anaglyphs (with blue and red filters). |
|
|
Please excuse my poor english, I manage to translate this page with
the help of babelfish.altavista,
but I'm pretty sure that it's still not perfect.
|
In the "real" world, our depth perception comes from to the fact that we have two eyes which do not exactly see the same thing. The brain use of the differences between the two images to reconstitute the relief of things, this is the "stereoscopic vision".
With a computer, the easiest way to produce stereoscopic images is to create anaglyphs. An anaglyph mixs into one image the stereoscopic views thanks to the complementarity of colors in the RGB channels. With colored glasses, one can then filter the image and give to each eye the proper informations.
Here I'll only consider the red and blue glasses (blue on the right, red on the left) which are the most current ones and preserve most of the colors (you could find on this site (in french) all the informations to find or manufacture anaglyph's glasses).
Things to take care of
In order to make the relief quite perceptible, it is necessary to follow some rules in the construction of the scene. First of all, you have to multiply the intermediate plans to give depth to the image. Also try to bind the various plans between each other. If they are too independent, there will give the impression to deal with simple frontages rather than to true volumes. Therefore think to add objects or grounds which leave towards the background. Lastly, the choice of the colors (and thus of textures) is a delicate point which conditions the quality of the final result mainly (see further for this particular point).
The snapshots
To create the anaglyph, you will need two slightly shifted images of the same scene.
Once given the location of the camera, render the first image (seen by the right eye). The point is now to know where to position the camera for the second image. It's very important, the quality of the final result depends on it. The drawback is that it requires some calculations.
Start by calculating the distance between the camera and the nearest visible surface. To obtain the coordinates of the camera, simply double-click the ball which controls the movements of the camera, for those of the objects pass by their attributes (Command-Option-E).
To calculate the exact distance, don't forget that Bryce gives the coordinates of the center of the objects whereas one seeks the distance between the camera and the first visible surface. You could create a small object whose center will be position (using the side views) against the first visible object.
Now dig the Pythagora theorem out of your memory to find the magic formula :
xc, yc, zc : camera's coordinate
xo, yo, zo : nearest object coordinate
Once you know the distance, divide it by 30. Why 30 ? Roughly speaking, there is a relation with the average distance between the eyes which is approximately 6,5 cm.
My special Gift : automatic calculation (if Javascript is activated in the preferences of your navigator).
At this point, one can find the point where to place the camera for the second snapshot. It would be possible to calculate exactly the coordinates of this point, but it's long and complex, so I prefer a more empirical method.
| Place a sphere in the center of the camera, the diameter of this sphere should equal to (i. e. 2 times the distance between the 2 positions of the camera). | ![]() |
|
|
Duplicate this sphere
(Apple-D), and convert it into a parallelepiped which you can widen
for more clearness. |
![]() |
|
|
Then you'll only
have to place the camera on the intersection between the sphere and the
axis of the parallelepiped. So you can manage to place the camera in a relatively precise way for the second snapshot. You just have to remove these objects from now on useless. |
![]() |
You don't need to modify the axis of the camera (materialized by the blue line), better results are obtained with parallel axes rather than convergent.The work with Bryce ends with the render of the second snapshot.
Operations on the channels
Open the 2 images in Photoshop (or any sofware that can manage RGB channels), image 1 corresponds to the one seen by the right eye, image 2 is for the left. While playing on the channels, you'll mix these informations to create an unique image (the glasses then allow each eye to recieve the relevant information). Image 1provides Green and Blue channels and image 2 the Red one.
- Select the Red channel of image 2 (Apple-1 and then Apple-A),
- Copy it (Apple-C).
Go to image 1 :
- Select Red channel (Apple-1),
- Paste (Apple-V),
- Deselect (Apple-D),
- Go back to RVB (Apple-0).
An advice : if you use Photoshop 4 or above, you can record the 2 actions for later uses.
With the glasses on the nose, you can now evaluate the result. It could be dissatisfactory ; in this case, you can try to slighty displace the Red channel horizontaly. In the channels palette, choose the red one and click on eye of the RVB channel to see the colored image while only working on the red channel. Then choose the displacement tool (V key).
Problem of the colors
The choice of the colors is very important to achieve the relief effect, you especially have to avoid too strong contrasts and saturations and, particularly too dense reds and blues : these colors are filtered by the glasses, so if a zone has for exemple a strong red dominance, it will only be "visible" for the right eye, therefore it won't achive the relief effect. Anyway, you don't have to focus too many on the colors : the glasses' filters considerably reduce the range of the visible colors. In good anaglyphs, dark greens, yellows and browns often prevail.
While working at the scenes in Bryce, you'll have to keep that in mind, but you also always have the possibility to retouch the images in Photoshop. Particularly, you can often improve the the left image (Image 2) by desaturing it (for example it's possible to convert it in gray levels before to go back in RGB mode).
One last advice : if you save the image in JPEG, don't compress it too much : il would be calamitous for the result.
How
to make 3D-pictures by computer
: a very accurate page on the creation of anaglyphs with computers.
Alpes
stéréo : very interesting site (in french and english) on
photographic anaglyphs.