Blog Post

The Technical Challenges of Turning Stereophotography into 3D StereoGIFs

The Technical Challenges of Turning Stereophotography into 3D StereoGIFs

As a Student Assistant at Queen's University Library, I work on a project to digitize late 19th- and early 20th- century stereographic view cards. These cards have 2 photographs (stereographic photographs) that, when seen through a viewer like the one below, appear to be 3D. Many libraries and museums have digitized and display images of the stereoview cards but few projects attempt to replicate the images' 3D effect. But the 3D effect was stereographic photography's key distinguishing feature and, at the Special Collections Library, we decided to preserve it in our digitization.

I've come across 2 ways of doing this. One is to make "wigglegrams" of the GIFs: by alternating rapidly between each image, you can trick the eye into seeing a single 3D image. The second technique comes from Patrick Feaster, the author of this blogpost, who developed a new technique to avoid the jarring sensation of wigglegrams. This technique uses Adobe Photoshop and "tweening"—an animation technique that computer-generates extra frames between two keyframes—to smoothen the motion between the two photographs. As the GIF below explains, the process seems relatively simple. But there are a number of technical challenges that go into making GIFs from historical stereoviews.

 

 

How You See in 3D

Before going into the technical challenges, it's worth discussing how a stereoview "works" in the first place. There are several cues your eyes and brain use to create the illusion of three-dimensionality. These include cues that you can see in photography and other 2D visual media (e.g. near objects appear larger than far objects, near objects can block or "occlude" far objects).

Stereographic photograph relies on a depth clue called stereopsis. Each eye receives slightly different views of an object because each eye is in a slightly different position in your head. Through the magic of human perception (and my explanation's gross oversimplification), these images combine so that you perceive a single 3D image. The stereoGIF (as I call it) still relies on stereopsis as the basis for its 3D effect but it also relies on another important cue: motion parallax. Motion parallax refers to the concept that, when the observer moves from side to side, near objects appear to move less and move slower than objects that are far away.

http://upload.wikimedia.org/wikipedia/commons/a/ab/Parallax.gif

Image by Nathaniel Domek, available on WIkimedia Commons

 

Characteristics affecting StereoGIFs

1. The photographs themselves                                                               

The 3D effect is much stronger with some stereoviews than with others. For example, the effect is rather strong with the fur parlour GIF but there are other cases where photos look like they're morphing into one another rather than swaying from side to side. And in still other cases, things appear move up or down, tilt, enlarge or get smaller (or turn in the case of the moon).

My guess for why this happens has to do with the photographs themselves, which never seem to match exactly even when accounting for the left-right disparity of each eye (pictured below). For example, you can see from the photograph's caption that one image is slightly higher than the other. Writing in Realism, Photography, and Nineteenth-Century Fiction, Novak mentions that, with the collodion process and mass-production, "Studios came to resemble factories, using unskilled labor to cut and mount mass-produced prints, in order to keep up with demand." Because of this, it's not all that surprising that they're slightly off. I don't know if there's a workaround for this but I couldn't find one. 

2. The Speed (number of frames) of the stereoGIF

How fast the GIF appears to be moving depends on the number of frames—the more frames in the animation, the slower (and smoother) the motion. Personally, I've found that the fewer frames there are, the stronger the 3D effect. But there's a tradeoff: the fewer the frames, the more disorienting or jarring the final GIF looks.

At one end of this spectrum is the wigglegram with only 2 frames (one for each photograph). But going too far in the other direction has its own consequences. When you tween two keyframes, Photoshop fades one image gradually into another. Too many frames and it no longer looks like things are moving in 3D space. Instead, it looks like exactly what it is -- 2 images fading in and out. Not what we want.

The sweetspot seems to be about 12-18 frames total (usually 13-15). In his example, Feaster has 13 frames total so I'm assuming he likewise experimented and found the speed he liked.

3. Motion and "motion blur"

Feaster calls this "jumping" in his blog post and likens the effect to lenticular images (which I will forever associate with agenda covers from grade school). I like to think of it as "motion blur"—that is, things seem to blur as they move from side to side. Motion blur makes the image appear artificial. To a point, even the motion itself is artificial. The moon, for example, appears to be rotating from one position to the next. This is a far cry from the stereoview, where the moon is still and appears to be convex.

4. Reference point and distance from reference point

When I say "reference point," I mean the point where I match the two images up when I superimpose them. Its position is a matter of choice. As noted earlier, the photographs never match perfectly when superimposed.

Choice of reference point also has implications for motion blur: the farther away from the reference point, the blurrier the object looks as it moves from side to side. Pick a point too far from the centre and the stereoGIF becomes disorienting since most of the image will be blurry. For that reason, I chose the reference point to be generally at or near the stereoGIF's centre.

But choosing a reference point can also be an interpretive decision in another sense. Obviously, parts of the photo that are closer to the reference point will be less blurry and therefore easier to see than parts that are farther away. In a stereoGIF like this one of the "Khodinsky Plain Panic," choosing a reference point that is "farther back" would make individual faces (that are nearer to the viewer) harder to see. The Khodinsky Plain Panic—or Khodynka Tragedy as it's more commonly called today—was a human stampede that resulted in the tragic loss of life. Judging from these contemporary news articles from Australia and New Zealand, "tragic" was probably how contemporary audiences would have described it. Ensuring the faces of the victims are clearly visible would push towards a more affective, sympathetic representation and that's what I chose to do.

Conclusion

Wading through all these technical complications really made it clear to me that, even as we are aware of what new things technology allows us to do, we should also be aware of its limitations. No technology or tool is perfect and it may have unintentional side effects despite the best of intentions. At its worst, some of these effects could even be called misleading (consider the moon stereoGIF, where the final result is quite different than that of the stereoview). The stereoGIF is an approximation or simulation of the 3D effect of its predecessor—not a full-blown replacement. As a PhD student working in DH at my university once said, "technology is 'in addition to,' not 'in lieu of.'"

So the disadvantages have to be weighed against the benefits. For Patrick Feaster, motion blur is not a "fatal flaw" because it reminds him of lenticular prints, another popular 3D form. I don't consider the motion blur a fatal flaw either but my reasons have to do with another point that Feaster makes.

For me, the issue boils down to the fact that the stereoGIF at least allows you to experience the object in (some semblance of) 3D space. This, I would argue, is "the point" of the stereoview and stereoscopic photography more generally. As Feaster notes, this 3D experience is otherwise unavailable to those who don't have access to some type of viewer or who don't know how to "free view" stereographs (free viewing is a technique that takes time and practice to learn). The stereoGIF allows you to see something that you would otherwise be unable to access even if the process and its results are imperfect.

73

7 comments

Hi Tiffany,

I found this post fascinating! My own research involves the 3D reocnstruction of 'legacy' photographic data form underwater archaeological excavations (1970s black and white negatives). Although this may be very different to your own topic presented here, I am very interested in your conclusions with technology and its limitations.

 

Cheers,

Maddy

93

Thanks for your comment, Maddy. Your research with 3D modelling also sounds fascinating.

Perhaps you have seen/read this before, but UCL has a chapter on 3D recording in one of their books. It sounds somewhat similar to what  I read from your blog and might interest you: http://blogs.ucl.ac.uk/dh-in-practice/chapter-5/

I hope your project is going well!

78

That is fantastic thanks Tiffany - I'll definitely follow up reading that one.

Cheers,

Maddy

68

Really fascinating work, Tiffany! it was a pleasure to read your blog.

 

95

Thanks for reading!

68

Tiffany,

Our Indiana University Hastac workshop just had a lecture by Patrick Feaster who is doing a lot of interesting work with sound and image morphing. Are you familiar with his techniques?

 

78

Hi Olga,

I am familiar with his blog post  and in fact, his blog is the basis for the technique I'm using to make my GIFs. I'm also familiar with the follow-up he did using FotoMorph software (which gives some really good results!) but it looks a bit too time intensive to be practical for my project.

88