3D-Printed Anaperture Single-Shot Anaglyph Aperture

Oleg L K wrote:

Well, to maintain interest to single-lens-3d, I uploaded lots of my photos to:

https://flic.kr/s/aHsmXbd33R

I included different photos, all considered by me as good, but not necessarily best, so that the method could be judged.

I remind that I glued color filters to the rear side of the lens - a DIY take on Q-DOS lens.

They don't look bad (actually, it's better than the Q-DOS does), but the shape of the left/right bokeh needs to be the same for the effect to be really nice. For example, this is a particularly good example of what's wrong with just covering opposite halves of the aperture with different color filters. Here's a crop of one OOF PSF from that image:

Basically, in stereo viewing the bokeh shape ends up overlaid half circles, which looks very unnatural.

Just to be clear, the Q-DOS is far worse because it places an optical seam in the middle of the PSF, seriously degrading focus quality -- which is actually quite sharp in non-Q-DOS mode. Here's a quick example:

At 135mm in Normal mode on A7RII

And the same shot in Q-DOS mode:

At 135mm Q-DOS mode on A7RII

Notice that the anaglyph effect is quite good (very little vignetting), but there's nothing sharp in the Q-DOS mode.

Oleg L K wrote:

ProfHankD wrote:

They don't look bad (actually, it's better than the Q-DOS does), but the shape of the left/right bokeh needs to be the same for the effect to be really nice. For example, this is a particularly good example of what's wrong with just covering opposite halves of the aperture with different color filters. Here's a crop of one OOF PSF from that image:

Basically, in stereo viewing the bokeh shape ends up overlaid half circles, which looks very unnatural.

Maybe one can add an opaque mask with max-diameter circle inside each halve? E.g. glue to the rear-side the filters and the mask?

I prefer front-mounted filters, but the same thing can theoretically be done in front, behind, or at any of several places within -- there are various points where appropriately-sized openings will behave as apertures. The trick is correctly sizing the two circular holes.

Notice that the anaglyph effect is quite good (very little vignetting), but there's nothing sharp in the Q-DOS mode.

Q-DOS lens is 70-210mm. With such long focal lenses vignetting gradually went away in my experiments too. Though your advice on vignetting mask was always beneficial.

Well, yes and no. Fast normals and wider often vignette a lot, but so do many telephoto lenses. The Q-DOS says it only works in the 135-210mm range, and that's largely because the lens doesn't vignette much at all in that range. Most telephoto lenses I've tried still need to be stopped down a bit to avoid vignetting.

Oleg L K wrote:

And maybe such two-hole aperture could be used for optical printing of interlaced image for viewing through lenticular lens? Is the idea completely idiotic?

Let's start with the parallax-barrier based idea described by Mr. Turbguy1.

But instead of moving the barrier relative to the paper, make the two exposures through two holes in front of the enlarger lens. Bokeh / OOF shape shouldn't matter, since the paper is flat.
Distance between the holes could be obtained with enough precision on 3d-printer.
The parallax barrier could be printed on photo-plotter to properly match lenticular-lens LPI.

Could it have a chance?

I would expect some variation of that to work quite well... at least for monochromatic lenticular prints.

3D Gunner wrote:

Those are interference filters used for passive 3D (cinema) projection, such as those used by the Dolby 3D system, or those invented and produced by INFITEC®.

They're just crossed polarizers (and depend on a screen that preserves polarization).

Commercial sensors don't generally come with differently-oriented polarizers over pixels, so not useful for capture. It's also worth noting that you'd actually have a little issue in that the same object seen by differently polarized left/right views can look very different; you'd probably also need a half-wave plate (as used to make a circular polarizer) to essentially randomize the polarization before separating-out the linearly-polarized left/right views.

There are tricks one can play with LCLVs (Liquid Crystal Light Valves, as I discuss in Programmable Liquid Crystal Apertures and Filters for Photographic Lenses ). In theory, you could use LCLVs to independently electronically shutter a pair of left/right apertures such that you could get separate exposures for left/right by timing the aperture openings against the progression of a slightly-unusual rolling electronic shutter that does something like interlacing. I.e., open left, scan even lines, close left, open right, scan odd lines, close right. That might be do-able on some Canon EOS models using Magic Lantern, because apparently some Canon EOS models have had two independent channels for readout of alternating lines). Easier, just do a two-exposure burst with left aperture, then right aperture.

Anyway, generally not real helpful for capture.

BTW, the real way to do display would be lightfield displays using tech like what Misapplied Sciences Parallel Reality is doing, although they use the lightfields to have many viewers -- without glasses -- simultaneously see entirely different things on the same display (here's a geekwire about it ). (Disclaimer: my brother, Paul Dietz, was actually Chairman and CTO of Misapplied Sciences, although he is no longer involved in running it.)

3D Gunner wrote:

ProfHankD wrote:

3D Gunner wrote:

Those are interference filters used for passive 3D (cinema) projection, such as those used by the Dolby 3D system, or those invented and produced by INFITEC®.

They're just crossed polarizers (and depend on a screen that preserves polarization).

No, they are interference filters and their use does not depend on the type of screen.

The only "interference filter 3D" I know of is this , and it isn't a filter in front of a projector but basically notch filters to distinguish two different bands within each of R, G, and B. I suppose they could be used in front of a projector, but with major light loss. Honestly, it would make a lot more sense to just do LED lights, since they have naturally very narrow emission bands.

Anyway, on capture they don't help much because of the ambiguity of separating-out 6 color channels using a 3-color CFA on the sensor....