AI ANIMATION from 100-year-old 3D Images

With the invention of the stereo camera around 1850, it became possible for the first time to capture the world in 3D and reproduce it with a sense of spatial depth.

Until around 1930, negatives and positives consisted of small glass plates, lending them a distinctive materiality and sense of value today. These delicate objects have preserved moments from over a century ago —times that predate two world wars. Consequently, there is a high probability that any surviving stereograph is a unique original—such as the image of a young woman experimenting with her stereo camera during the Victorian era, taking selfies, or capturing her own reflection in a mirror.

Their tactile quality and uniqueness make such photos the exact opposite of the AI-generated digital creations that are currently flooding our visual world — images that lack any physical counterpart or connection to a tangible memory. This is not to say that I do not experiment with AI myself. However, I do not use it to fabricate the moment or the subject, nor to animate the person as if a motion-picture camera had been used. Instead, I use AI to generate the space extending beyond the original frame while maintaining a strong link to the original; I leave the subjects’ facial expressions and physical presence exactly as I found them — frozen in a split second of time. Only the camera itself is subsequently granted dynamic freedom of movement, circling the subjects in a 360-degree sweep. This reveals every aspect of the location and the setting beyond the original stereoscopic frame.

Ultimately, I am left with the analogue, tangible glass original alongside the digital, disembodied AI animation — a combination that evokes a completely different kind of imagination than an AI animation standing alone without the original.

Girl in a mirror as a stereoscopic glass positive ©Martin Blum MIND.WORK
mind work stereoscopic selfie made in the Victorian age. Copyright by MIND.WORK®

4 comments

  • Animating century-old stereo images raises a fascinating provenance question. Keeping the original pair visible beside the animated interpretation would let viewers enjoy the sense of depth while still seeing which details came from the archive and which motion was created later.

    • Thanks for this feedback and have a look at the result :-). I believe there are clear rules and timeframes regarding the artistic use of photographs that are more than 100 years old. What I am less certain about is the question of provenance when it comes to the AI-generated part—since it is virtually impossible to determine where, or using what specific source material, an AI model was trained. Perhaps you have more knowledge and information on that?

  • Generating only the space beyond the original frame while leaving the subjects’ expressions untouched is a respectful way to use AI with stereographs. Moving the camera rather than the person keeps the link to that split second from more than a century ago.

    • Thanks for your comment and these stereographs, in their original form, truly possess an immense presence when you hold them in your hands.

Submit a comment