Online CGH Instructions


If you're reading this, you're probably interested in making your own Computer Generated Hologram (CGH).  Fortunately, you're in luck.  This website will easily allow you to compute the interference fringes of an image of your choice, thereby allowing you to skip the messy bits like writing the physics simulation code, figuring out the technical and geometric constraints, and optimizing the binary output mechanism.

You obviously already have an internet connection and access to my website, so if you just have a decent laser (or ink-jet) printer and some transparency paper then let's get to work. 

To create and view your own hologram:
  • Download a source file from the precomputed holograms below, and, if desired, edit it in some graphic program.  Start off with someting simple like the letter "A" in a simple font.  I would recommend no more than a total of 50 black pixels in your early CGH tests, and stick to black and white only since grayscale and color are meaningless here.
  • Read the instructions and launch the program from this page.  Then load your input file and press the "Start Simulation" button.  Gears will spin, smoke will rise, and your hologram will compute.  When it finishes, click the "Display Output" button to see what they look like.  You can use the "Save GIF" command to write the fringes to your local hard-drive.
  • Open your file in a graphics program such as GIMP, and print it on the transparency paper.  Be sure to set the printer resolution correctly.  For example, I usually compute fringes at 600x600 DPI so a 600x600 pixel image should be 1 square inch.  You must turn off interpolation, dithering, and any other printer/driver features that could alter the printout.  We need to be able to place every single dot on the page with precision.  Don't let your software resize the image.  The printout should be *exactly* 1 inch square.  If  it is not exactly this size, then you'll have to figure out where things went wrong.
  • Aim your laser pointer at a smooth white diffuse wall, and secure it so that it stays on and steady without having to hold it.  I know that you're impatient, but you'll be frustrated if you try to hold the laser pointer and hologram simultaneously while getting also viewing the image. 
  • Cut a window in a file folder so it holds your transparency flat.  This will keep that huge transparency from flapping when you're trying to shine a 1 square mm laser beam through a small section of your hologram.
  • Experiment with moving your "mounted" hologram through the path of the laser approximately 1-2 m from the screen while observing the diffraction on the sceen.  If everything worked, then on the screen you should see an upright image on one side of the main beam, and an upside down (aka "conjugate") image directly opposite the upright one.
  • When you understand what is going on, try placing your hologram in the laser path at a 45 degree angle to both the vertical and horizontal axes of the screen.  Although it's a bit awkward, this maneuver effectively raises the printout density from 600 DPI to 848 DPI at the cost of some slight distortion.

Viewing Setup
Here's what the viewing setup looks like.  The laser beam shines through the transparency and projects an image on the wall.  Notice the folder with an aperture which is used to add support to the flimsy transparencies. 

Precomputed Holograms (for the lazy and/or confused)

If you're not sure where to start, download some precomputed interference fringes and try printing them.  Use right-click and "view image" to see them in higher resolution.  The source files are also listed if you want to recalculate them or use them as a starting point for modification.  I really suggest you start off with the diffraction grating, but that's just my two cents.

Diffraction Grating
Single Point
Double Point
Circle (on axis)
Diffraction Grating
Diffraction Grating Diffraction Grating Diffraction Grating
Fringes (CGH)
Fringes Fringes Fringes
Source File
Object Image
Object Image Object Image

Letter "A"
Another Circle
Cube (more)
Diffraction Grating
Diffraction Grating Diffraction Grating Diffraction Grating
Fringes (CGH)
Diffraction Grating Source
Diffraction Grating Source Diffraction Grating Source Diffraction Grating Source
Source File
Single Point Single Point

Letter "H"
A square
Circle (off axis)
Diffraction Grating
Diffraction Grating Diffraction Grating

Fringes (CGH)
Fringes Fringes Fringes
Source File
Object Image Object Image
Object Image XML_file

  • You can use the Source Files above as input to the CorticalCafe CGH Construction Kit, or you can edit these or create your own input files.  But keep the total number of pixels in your image small.
  • The reconstructed images in this technique will always be radially symmetric.  That is, you will see a conjugate image upside down and backwards on the other side of main laser spot.  I took advantage of this with the circle reconstruction since it is naturally radially symmetric.  But the cube reconstruction clearly shows you the 2 images.  Depending on how the CGH is computed and viewed, You may not be able to separate out the desired from the undesired image.
  • Note the random scatter which accompanies the images.  The transparency becomes a bit cloudy as it goes through the laser-printer, and creates this background noise which detracts form the images.
  • The computed holograms sometimes have aliasing and artifacts.  This is a function of using a binary output device, the laser printer.  It looks neat, but undoubtedly causes some issues in the reconstruction which have been addressed in CGH literature.
  • You can't compute the cube yet using my online program, but you can download the fringes and play with it.  In the animation above, you can see that moving the cube through the beam allows you to rotate the projected image of the cube in 3-dimensions.  Groovy.

The Diffraction Grating

A Diffraction Grating is the great grand-daddy of holograms and a good place to understand what is going on here.  It is simply a set of very closely spaced parallel lines.  If you print the gratings on this page, you should see a set of alternating transparent and black lines on your film.  When the laser passes through this pattern, the beam is diffracted, and the light is spread out on the screen.   Ideally, you should see a set of dots which repeat on either side of the main laser beam like this.  If the dots aren't clear, then the light isn't being diffracted efficiently, and the quality of your CGH will suffer big-time.

The 600 DPI grating represents the closest lines that your printer can print, hence, it also represents the largest projected image for this technique.  Your projected CGH's will be this size or smaller.  Of course, when using the diffraction grating, you can increase the distance between the hologram and screen to increase the image/dot size.  But this doesn't really help our CGHs since the hologram calculation requires the inclusion of a focal distance.  And increasing this focal distance makes the object smaller which offsets the size increase.

The 300 DPI grating doesn't push your printer quite so hard as the 600 DPI version.  Because the lines are farther apart, the diffraction spots on the wall are closer together, and an image printed at this reslution will be smaller.

If the 600 DPI grating looks bad, but the 300 DPI grating looks good, then your printer is having difficulty printing at 600 DPI.  In my experience, 600 DPI gratings tend to be "blotchy", perhaps from variability in toner particle size .  You can directly examine the gratings with a magnifying glass to see the quality of your printer in drawing the grating lines.  If you don't see parallel lines at all, but rather some cross-hatch type pattern then your printer settings or driver are problematic and more complicated CGH attempts are almost certain to fail.

About Printers

In the course of my experiments, I've encountered quite a number of printer issues.  We're pushing the limits of printer technology and we care about pixel placement accuracy more than any printer manufacturer ever imagined.  Any of the numerous enhancements that the manufacturer added to the printer firmware because they usually improve image quality are pretty likely to ruin your chances of getting a good reconstruction.  My printer tips are as follows:
  • Just because a printer advertises 1200dpi resolution is no guarantee that every single dot in the image will accurately correspond to a dot in the output.  Don't assume that any resolution mode prints accurately without an experiment to demonstrate it.
  • To quickly detect whether your printer is accurately reproducing your images, try printing a symmetric test pattern (the "double point" pattern above is excellent) at all available resolutions.  You'll have to adjust the page size to ensure that your software doesn't resize the image.  I've seen cases where the fringes print fine at 300dpi  (2in across for the "double point" image) and 600dpi (1in), but at 1200dpi (0.5 in), the pattern was noticeably different.  Some aliasing/moire effects were present and indicate that image reproduction was not accurate.
  • 300dpi is just not fine enough resolution for reasonable size hologram reconstructions.  600dpi is better, but again your reconstructions will be small.  At 1200dpi, you may be pushing the limits of technology so that finge printing is not accurate.
  • The "Diffraction Grating" is the worst case scenario since it alternates light and dark pixels.   Quite frequently the pixels are larger than the blank cells and no light is transmitted through the fringes.  You might also rotate the diffraction grating by 90 degrees, since printer horizontal resolution accuracy and vertical resolution accuracy are controlled by different mechanisms and may not be symmetric. 
  • Add memory to your printer, even if you don't think you need it.  My printer (a brother HL-5340D) would not print at 1200dpi until I added additional memory.  It didn't report any errors, but simply printed in 600DPI mode, despite the fact that the entire page size was far below the 16MB native printer memory.
  • Don't count on the manual or tech support to help.  In my case, Brother technical support was quite wrong about their description of their own product.  From a meaningless explanation of the differences between "1200dpi" and "HQ 1200dpi" (I still don't understand the difference!) to outright misinformation about when additional memory was required, the tech support representative simply didn't know what was going on.  Of course, if the manual explained this information accurately, I wouldn't have wasted any time with tech support. 
  • If your printer advertises "effective" resolution, I suggest you look elsewhere since there's almost certainly some interpolation going on.  
  • Examine the general quality of your printer.  If your images have inconsistent black levels, you have streaking, or some other obvious problem, then you probably aren't going to have much luck.

Problem Solving

Here are some tips:
  • All holograms on this page should print as exactly 1 square inch.  If your images are not this size, then you are rescaling (and almost certainly messing up)the image.
  • Start with the diffraction gratings. Don't try anything more complicated until you know that these work.
  • Transparency film matters. Different brands have different qualities.  Shine the laser through a clear section of your transparency film and observe the beam.  All that noise you see on the screen interferes with nice quality CGH reconstructions.  Unfortunately, you can see that running a transparency through the laser printer changes the optical properties from a nice clear media to a foggy mess. 
  • Printers matter.  Make sure your printer can do 600 DPI.  And don't expect good (or any) CGH results if your normal paper printouts are streaked and dirty.
  • For the cube, I printed the image and then took a picture on 35mm film and used the negative for the laser beam.  This results in a pretty high resolution, high quality CGHs.  All the more power to you if your laser-printer will print at greater than 600DPI.  In these cases for the above examples, the CGH plate will be smaller, but the resulting reconstruction will be bigger and clearer.

Want to help?

Here's my wish list:
  • Any donations toward this project will go for improving computing/output resources and better hosting.  See the page (also for source code, forums, bug tracking, etc).
  • Do you have a good suggestion for efficiently characterizing laser printer output?  It would be great to have a printable template that quickly indicates the quality of an output device.

Hey, did you try my Online Printmaker yet?