Stereogram Images and Blender

Ok, I understand the basic thing about These images(also know as “Magic Eye”).

You basically take a single image from two different perspectives(left eye and right eye), then seperate them…

This is done with a mathematic formula and a set thing, so this should be doable for a python script or some other blender thing…

It would be cool to be able to print out 3D images of some meshes and be able to see them in 3D…

Anyone know if this has been done or what?

look at these stereogramme or anaglyphes from me and some with Blender (u may translate with Google) : (Blender) (Blender)

U can find, FreeSoft to do this, on my web in DownloadSection\2D (u must register) ! Or looking for “StéréoPro” and Z’Glyph !!!

Some r using ZBuffer (or BW Blend gradient) and StéréoPro
2 pictures shift of same scene and Z’Glyph !!!

Well I tried your website but after it logged me in, it refreshed to a non-logged in page… my javascript and cookies were both on, and every other website that I visit works fine.

So I couldn’t access anything, I tried both with the google proxy and without…

Thanks for helping though, is there anywhere else to download those programs?

I think they are exactly what i’ve been looking for…

but it work !!!
Try to Refresh simply !!! F5key or !!!

Well ok, after messing around with it I think I am connected, however I cannot translate it.

I downloaded Stereo Pro, and found ZGlyphs in the 2d section.

Anyway, Stereo Pro looks like it is some geological mapping program or something, and not what I wanted…

I want to do stuff like your blue and red StereoGram, what program did you do those with??? I don’t want the glasses picture, just to be able to map out meshes to a design like in your Blue and Red pictures…

Thanks for the help!

from my web



True, U cannot translate pages that need registering !

What I think you are looking into is stuff derived from Single Image Randoim Dot Seterograms. Do a google sort for that or simplty “SIRDS”. They’ve been expanded to not require random dots, but rather the same image tiled horizontally. Basically you take z-buffer type information (not separate images) to adjust how much the pattern is shifted for each copy. So the programs used to generate SIRDS take a greyscale image as input, where the lighter areas are raised, and darker areas are sunken. To create a stuitable image in blender, you could make white objects with no shading, and use a black mist to add the depth shading. Play around with Start and Depth, so the brightest part is white, and the darkest object is not quite 100% black.
You should be able to use that depth information on most SIRDS tools. There are plenty of nately english ones.

If you want to make Red and blue 3D glasses type images, just renter two images with the camera moves a little bit appart. Make one red, the other blue (in the Gimp or similar program), put them in separate layers & merge them together adding the layers.

I agree SpkyElctrc ! i do like this too or like that :

Join all mesh together, Add MATTEX Blend B/W to simulate ZBuffer (in good axis !) (i never use Myst, but I’ll do, that is a good trick :slight_smile: no-join needed)

For other method, u can use SequencePlugin like this

I was just about to mention that one :wink:

I wrote it, so I can help if you’re having problems with it.

Hmm I am having problems… the final sequences I get seem to have a bunch of noise or something in them… I can make out the basic pattern of whatever my mesh is, but it looks like waves or some sort of noise is interfearing…

Like my cube, looked like a bunch of triangles stacked upon one another.

or I had some text, and you could make out the basic shape, but there were wierd lines, bubbles and waves and stuff that seemed to be messing with the output… I checked the Zbuffer image on each, and it looked fine, the contrast was good and it should have outputted clearly.

Any idea’s?

how odd

can you post pics?

Sorry for the really crappy quality but my free image host only allows images of 100k at a time… I could email or something bigger ones if you want to check it out.

This is a bit of text… “This is the part that gets messed up!!!” You can read the first 2 letters but then the rest is messed up.

This is the Z buffer of a cube:

Here is the Sequenced version(sorry for the cropping but I was trying to get the file size down to under 100k) The part with the cube is still viewable.

It looks(to me at least) like a bunch of triangles perpendicular to the camera stacked on one another, approximating a cube shape, but that is hard to see.

Since when viewed at an angle, each slice of a cube would be a triangle, what I think it happening is that somehow it is taking like .25 units at a time, rendering that section, skipping a section, then rendering another… but since you wrote it you probably have a better idea:

I can’t see a thing in either of those stereograms. Could you email the full-size ones to me?

I still don’t see it, but I think I figured out the problem.

it works by changing the distance between repetitions. Obviously, the distance can only change by multiples of 1 pixel.

So effectively, the heightmask looks similar to this:

You can try turning on Interpolation on the plugin to fix it, but be warned, it gives very bad results for noise patterns, and importing patterns is still in the todo list.

ha! i’m glad to see this thread! how uncanny that i have been mulling over this idea for the past few days before i saw this thread!

There are, of course, two ways to get stereo images: those funny dots that I can never see anything in; or the technique my great-grandmother showed me … stereo cards.

They’re pictures taken with a camera having two lenses, spaced the distance of human eyes apart. This can be effortlessly achieved in Blender, e.g. by two cameras, parented to each end of a convenient off-camera “arm” that maintains the proper spacing. Render the image from each camera in turn.