Digital alignment

Digital image pairs can be easily aligned using Adobe PHOTOSHOP (and very likely other image editing softwares). Careful alignment is very important for comfortable 3D viewing, especially in projection. What follows is our method for alignment which we have found to be fast and efficient.

NOTE: Another EASY method for alignment (and many other stereo still processes) is to utilize STEREO PHOTO MAKER.
It's a wonderful FREE utility from Masuji SUTO of Japan.

  1. STEP ONE: Obviously, the stereo pair must be digitized if coming from an analog source such as print or slide. The necessary resolution depends on output- for most applications, 1000-1500 pixel height will suffice.
  2. STEP TWO: Determine the convergence point of the pair. This often involves looking at the image in 3D and asking yourself which point do you want to be at the stereo "window" (the spot along the z axis right exactly at the screen)? Very often it will be the most foreward spot in the image, but if you want something to come through the "window" (out of the screen) you will want to converge on something behind the forepoint.
  3. STEP THREE: Open the left image, select all, and copy. Now open the right image and paste the left image on top of it. You can set the opacity of the top (left) layer to 50%, thereby being able to see both views simulaneously. (Image A) With the "move" tool, you will be able to set the alignment of the image to exactly where you want it... you can even correct for rotation errors (transform, rotate) which should be easily apparent (they sometimes creep in during the scanning process... Keystone distortion can also be corrected with transform/distort, but that's somewhat advanced!)
  4. STEP FOUR: Crop the image. (Image B) Cropping can be a bit tricky... since you've moved the top layer (left image) around, there may be some blank image area- you want to make sure you crop within the image area. (Sometimes turning off the background layer from view helps.) If you are cropping for a particular aspect ratio (such as for a 2XGA image) you will want to use the "Fixed Aspect Ratio" option of the Marquee Tool.
  5. STEP FIVE: Copy the top layer, select file/new- the new image size will be exactly what's on the clipboard (the left image) so click "OK" and paste onto the new canvas, then flatten. Delete the top layer from the right image- you now have an aligned stereo pair, both are the exact same size.

Image A

Image B

CROPPING for a 2XGA image

2XGA images are used to project 3D stills from a laptop to dual digital video projectors

Looking at Image B above, you would shrink the height of the images to 768 each (it's always better to make images smaller than to try and make them larger, so start out with a good size scan!). Since the width of these images is less than 1024, you would go Image/Canvas Size to make the images 1024 wide (black is the best background color). You would then put the two images side by side as one... One way to do this is to take the right image, change canvas size to 2048 (make sure right image stays on the left) and use a different background color (like yellow). Using the Magic Wand Tool, select the new (yellow) background on the right. Go and Select All on the left image, and paste it into the selected (yellow) area of the expanded right image. Flatten. Save as .jpg and chage the extension to .jps if you want the image to be understood as a stereo image by softwares such as Vrex Depthcharge.
Very often you will want to try and use the full 1024x768 image area... to do this, use the "Fixed Aspect Ratio" option on the Marquee Tool, then set the width to 10.24 and height to 7.68. While the image pair is still in layers, select the area you want for your final image. (Hopefully your scan has enough resolution to still provide at least 768 worth of vertical pixels!) Crop, and go Image Size to convert to 1024x768 (changing one number should automatically change the other... be sure to have "Constrain Proportions" checked!). Now you can make your 2XGA image as described above.

The farthest object in a scene can be considered "infinity". The convergence point of a scene has ZERO parallax- it is in the exact same spot in each image of the stereo pair. As you go back in the scene toward infinity, "positive" parallax increases until the maximum is reached at infinity.

The convergence point in the above scene is the standing baseball player, measured at his right elbow, the same number of pixels from the edge of the screen in both images. The infinity point (farthest item) would be the program, which has shifted to the left 25 PIXELS (94-69). If this image were projected on a 7 foot high screen, 25 pixels would equal appx. 2.7 inches. (9.14 pixels = one inch at that size). Since our eyes are just about 2.5 inches apart, this separation on the screen would be comfortable to view. Our eyes do not like to diverge (toe OUT), so too much more "positive" parallax (say, more than 30 pixels) would become uncomfortable to view.

(Our eyes, however, have no problems converging- so items may come OUT of the image ["negative" parallax] quite a bit more than 30 pixels.)

Stereo SLIDE Services / Stereo PHOTOGRAPHY / Stereo VIDEO / COMPUTER Stereo /
ANAGLYPH Stereo / Stereo PROJECTION / Stereo FOR SALE / Ron Labbe / Stereo LINKS