New Testingstan

The theory behind holography is super simple, basically just use an interference speckle pattern to reproduce a beam reflected off an object (not just "a beam" but the full wave front) which will appear in our eyes as an exact replica of the actual object. Look it up, it's simple af (as fuck)
The problem is just creating that speckle pattern. One way is to use lasers and mirrors and all kinds of weird photo chemicals but nobody can stand that hassle in this bleak cyberpunk future. We gotta go digital.

There are many ways to compute a hologram pattern, and I even thought of one right now I think will work. It is basically inverted ray tracing, in that for every pixel you record data from rays from all points on the depicted object or scene, instead of just one. The data to store is the resulting intensity when these rays interfere with an imagined reference beam in the plane of the hologram. Basically add the sine of the phase of each ray at that point, which would be the distance (in a ratio of the reference wavelength) the ray travelled from the object to the pixel, plus a "starting phase" for every point on the object, which will be constant for the whole simulation and can be stored. If the reference beam is normal to the plane it will be constant across the surface and can be ignored, but if it is angled in one dimension it will produce a striped pattern when interfering with the other rays, or if arbitrarily angled, a circular striped pattern (it depends on the phase of the coherent beam at each point).
Maybe that won't work but there are also other ways, some using fourier transforms and stuff, that do work. But however we produce our virtual speckle pattern, there exists a problem of getting it out of the computer onto an actual surface. Here are some ways for a poor with no fancy equipment and only ordinary household lasers.

- Use a printer. Normally a computer printer is way too crude for producing hologram level detail. We need detail at least on a 10 micron scale, preferably around 1 micron (or at ideally the wavelength of the light). And printers only print good up to about 300 dpi or so. But it's actually possible to increase the resolution by printing in full detail at 300 dpi and then shrinking the image optically. One way might be to print on a transparent A4-sized film and shine a demagnifying light beam (such as that from an overhead projector, although it would have to be very good lens because we need sharpness) onto a much smaller photo film, 70 or even 35 mm, of good quality. Develop, fix and shine a laser on it and there should be a hologram somewhere nearby.

- Use a CD/DVD burner. Turns out optical discs are really good holographic media that don't even need any chemical treatment and DVD burners are sub-micron laser printers
Sadly it's not simply a matter of calculating the correct data file to write because there is a lot of bullshit added to that data, some specifically to make it seem more noisy. But if we could take direct control of the burner we would have a great hologram printer. We know it is possible because there were those programs that would abuse the burner to burn labels onto the other side of the disc. Some could even apparently manipulate the beam intensity which might allow us to make greyscale holograms that would be impressive indeed. As is appropriate for a cyberpunk future I hear some Japanese scientists have already managed to print holograms in this way.

Share your best holography ideas ITT (in this thread)

I'm not sure anyone can successfully expand on what you've already written.

so is this like how if you can compress digital music with the sine waves then if you just modulated some frequencies it should be easy to synthesize any instrument?

tbh i love me some fm synth sound, +1 :like:

but mang I'm not sure anyone can successfully expand on what you've already written. qft

Why wouldn't you just use a rainbow hologram so you can ditch the laser illumination? That shouldn't be much harder to simulate digitally.

High-end consumer inkjets hit roughly .73 dots per micron, though this is elliptically distorted (1.44 dpµ horizontally, 0.36 vertically). Of course, this is still probably too crude given that the very high end only use seven colors (black, and two shades of red/cyan/magenta) or vary ink intensity, and these are still $700+ printers.

I did find a Japanese optical-disc hologram printer. http://www.researchgate.net/publication ... _holograms However, after registering an account, it actually was purpose-built- the CD burner used by "Sakamoto et al" is no longer on the market. It is optically simpler to build than a comparable laser printer, but the issue is control systems for modulating the laser output and the speed (2 hours to produce a 25-cm square hologram, while good, isn't exactly in the range for the gloomy cyberpunk future). Also, the LD might be a tad bit on the expensive side. In theory, though, you could make one by tearing a laser printer apart and rigging it to work with their rotating drum+objective lens system.

Interestingly, you can print holograms with commercial printers, they're just too narrow-angle to actually view (1200 dpi gets a 1.7 degree viewing angle), but in theory, if you printed a series of them and glued them on a curved surface, you could increase the effective viewing angle. Of course, the issue is the cost of printing 40 holograms and something that can differentiate them enough to get a 70-degree effective viewing angle.

Bakustra wrote:Why wouldn't you just use a rainbow hologram so you can ditch the laser illumination? That shouldn't be much harder to simulate digitally.

A version of my post got eaten by the one-hour relogin, but I had a thought it should be possible to make rainbow holograms by, for each line of pixels in the hologram, only tracing a corresponding plane of points on the 3D object. I've had some more time to think about it now, and indeed this is how it should work. Rainbow holograms are made by shining the object and reference beams through a slit, removing one axis of parallax (for fun, try shining a laser pointer through a spreading lens at a rainbow hologram, and you'll notice only a horizontal line is reflected - this is the view through the slit). So it should actually cut down the computations exponentially (or alternatively a higher quality simulation could be achieved for the same cost).

As an aside, it seems interesting to me that rainbow holograms are sort of half-way between holograms and photographs. If you add another, orthogonal, slit when exposing the hologram you would get a pinhole, and it would be a regular camera.

High-end consumer inkjets hit roughly .73 dots per micron, though this is elliptically distorted (1.44 dpµ horizontally, 0.36 vertically). Of course, this is still probably too crude given that the very high end only use seven colors (black, and two shades of red/cyan/magenta) or vary ink intensity, and these are still $700+ printers.

There's also the question if they are printing accurately at those resolution or if it's an "effective resolution" where they use funky interpolation that might ruin our interference patterns.

I did find a Japanese optical-disc hologram printer. http://www.researchgate.net/publication ... _holograms However, after registering an account, it actually was purpose-built- the CD burner used by "Sakamoto et al" is no longer on the market. It is optically simpler to build than a comparable laser printer, but the issue is control systems for modulating the laser output and the speed (2 hours to produce a 25-cm square hologram, while good, isn't exactly in the range for the gloomy cyberpunk future). Also, the LD might be a tad bit on the expensive side. In theory, though, you could make one by tearing a laser printer apart and rigging it to work with their rotating drum+objective lens system.

Interestingly, you can print holograms with commercial printers, they're just too narrow-angle to actually view (1200 dpi gets a 1.7 degree viewing angle), but in theory, if you printed a series of them and glued them on a curved surface, you could increase the effective viewing angle. Of course, the issue is the cost of printing 40 holograms and something that can differentiate them enough to get a 70-degree effective viewing angle.

There is also this, which I was referring to: http://cgh.ist.hokudai.ac.jp/en/researc ... -themes08/
It seems to me they are probably using something similar to the "LabelFlash" or "DiscT@2" technologies. I haven't looked into fully whether DiscT@2 was implemented in the driver or hardware, and likewise if the error correction and byte modulation stuff when writing discs are in software or hardware (open-source cdr code is not exactly pretty). If it is in software, it would be fun to do some experiments. Especially what could be done with layered DVDs and Bluray discs.

adr wrote:so is this like how if you can compress digital music with the sine waves then if you just modulated some frequencies it should be easy to synthesize any instrument?

kind of. The holographic projection of a single point on a flat screen would be a circular sine-like pattern, and an object would be a composition of many such sine waves. There is a connection between recorded sound and holograms, in that a sound recording is also a full wave front, but through a single point. A microphone and speaker can be seen as a point in a wall that blocks all sound; the sound coming through the hole is what you hear through the speaker.


Also, I figured it might be useful to have a "hologram decoding" utility, which doesn't seem to exist for some reason. So you can check a digital hologram from various projections and wavelengths to see if it "works" etc. A first step I think should be to calculate diffraction patterns, so I started making a little thing to do that for various configurations using wave simulations rather than analytical solutions. For the next step I want some gui to test if it's doing what it should, but of course as usual there are only terrible options for gui libraries. Anyway here is something it produced (a 5 cm screen is going back and forth from 0.5 to 1 metre from a 50 micrometer double slit - don't know if it's actually correct, guess I could use the analytical solution to check...)

I'd need quite a few schematics to fully understands what's being talked about here, but, still, I like where this is going.

Wikipedia has pretty good starting diagrams.

Also I had accidentally made the screen and aperture twice as big in the diffraction program so there was an error of a factor of about 4 when I compared to where the maxima in a double slit interference pattern should be (as can be calculated here).
This is more like it

(units are micrometers, wavelength is 605 nm (k is wave number, 2*pi/wavelength))