3 D Photography vs Machine Learning - Use case issues

I have been permanently captured by 3 D experiences right from the 1970s, when I was introduced to the View Master and tens of Reels (7 pairs of photos set in a cardboard wheel). In the early 90s, I was introduced to a binocular slide viewer by a BITS college mate, which used same methodology with better clarity, as it was on 35mm film with great colour saturation. It was also a method of preserving own photos & memories, in 3D. Since then I had captured lot of photos on diapositive film (slides) in pairs, in order to store and enjoy the memories.

Ganapathi sculpture on rock, Lepakshi, AP

The advent of further VR technology helped in moving to the digital landscape with even better experiences & details that could be preserved - later enjoyed at leisure, sitting at home. If only I could have foreseen this present tech in the 90s, I would have taken more pictures on regular nagative films & prints. Such pics are now easily scanned into 3D pairs!

Mobile Phone Gallery

Now getting into the title of this post - Machine Learning Use case issues - my phone suggests that I have about 272 dupliacte images in my gallery (see screen snapshot of my mobile phone). Unfortunately, 90% of them are not duplicate. They are left and right images that help create the 3D image. The slight difference in angle helps create the 3 D Experience that one gets in the real locations.

By the way, our eyes & brains together work more wonders in location, that can never be recreated due to limitation of presenting a camera point of view. Our dynamic movement of head, eyes, walk around, etc., in micro seconds create the immense physical reality. In my opinion, that is the reason Tourism can never be replaced by VR, YouTube videos, Tiktok, Instagram, or any other tech.

Back to the issue of Use cases. The algorithm that seems to have been put in place in these mobile devices processes similar images, with say 90% match, as potential duplicate. The use case is - wish to help people clean-up space for our ever increasing appetite of taking pictures (and not taking a second glance for the rest of our life in 98% of the times, in my opinion). This algorithm goes against the use case of images taken for purpose of 3D photography & will knock off one of the images, if its suggestion is used carelessly. I have included a 3D Pair image of Belur Chennakeshava temple's 37th Salabhanjika (Madanika) - a drum dancer sculpture (how to experience 3D - search for Stereoscopy to look up instructions).

Drum Dancer - Belur, Karnataka

These types of conflicting use cases that occur everywhere in our lives, can never be resolved by a 3rd party, leave alone a machine! It always seems to lead back to the individual to decide on what is best for themselves. Dependence on others for decision making - be it machine or sentient beings - seems to bring regrets at some point or other!

PS: This idea / view, can be extended to other areas in life too. My role-model's use cases are not a 100% fit to my life. My astrologer's points of focus need not be aligned with my apprehensions or anxieties about the future. I may not be 100% aligned with my coach due to slightly different priorities, biases, experiences & expectations. This does not mean that Astrology, Coach or Role-models are wrong. We have to observe and utilize the necessary from each, and leave out the chaff (relative to our principles, priorities, and life journey).

Similarly "Dharma" is very subtle - every use case needs to be analyzed in its context. When humans are not able to grasp the subtleties over thousands of years, can machines work them out in a few years?

Mandapam at Varadaraja Temple, Kanchipuram, TN

PPS: My collection of such 3D photographs is available on Wikimedia Commons (Wikipedia's Media collection website), organized in this Stereoscopy category.

These images uploaded in Wikimedia contain information on how to view them too.