Our increasingly AI (Artificial Intelligence)-driven future remains a subject of great concern, given the inherent biases that are reflected in so many approaches, and the outcomes that seem to emphasize the worst in human nature. With that caveat, here’s something positive to look forward to, the enhancement and improvement of historical materials. Denis Shiryaev used a set of AI tools to both smooth and upscale Louis Lumiere’s 1896 footage of a train arriving at La Ciotat to the modern standards for phone video (4K, 60 frames per second).

 

A second tool, “De-oldify” was applied to add color to the footage.

As an old friend pointed out, this makes one of the most-mocked scenes in science fiction suddenly a lot more credible.

David Crotty

David Crotty

David Crotty is a Senior Consultant at Clarke & Esposito, a boutique management consulting firm focused on strategic issues related to professional and academic publishing and information services. Previously, David was the Editorial Director, Journals Policy for Oxford University Press. He oversaw journal policy across OUP’s journals program, drove technological innovation, and served as an information officer. David acquired and managed a suite of research society-owned journals with OUP, and before that was the Executive Editor for Cold Spring Harbor Laboratory Press, where he created and edited new science books and journals, along with serving as a journal Editor-in-Chief. He has served on the Board of Directors for the STM Association, the Society for Scholarly Publishing and CHOR, Inc., as well as The AAP-PSP Executive Council. David received his PhD in Genetics from Columbia University and did developmental neuroscience research at Caltech before moving from the bench to publishing.

Discussion

13 Thoughts on "AI Upscales 1896 Movie to High Definition"

I contest the idea that these new films “enhance” anything. These digital versions have fundamentally replaced the actual with the palatable. These people from 1898 have much in common with us, but they did not occupy the same physical world in most ways. Their social, moral, political, historical perspectives are far removed from ours.

Is it really so burdensome to accept primary sources as primary sources? “Fixing” the past is intellectually lazy and historically dishonest. Of course, I majored in history in college and have spent my career in STM publishing, so I have a substantially above average interest in such matters.

That’s a really interesting response, and one I will have to think more about. My initial thoughts are that one doesn’t have to treat something like this as a “replacement”. It’s not a zero-sum game, and there are different values to be found in different types of artifacts. The original certainly gives a much better representation of film technology of the time, but perhaps you’re interested in the details of clothing manufacturing, so being able to see more finely the outfits being worn would be helpful.

This is a travesty, not “something to look forward to.” A film such as this is a historical artefact. Do you want someone painting Greek statues all the colors of the rainbow and embedding LED lighting in them? Film historians study the quality of the photography of early films, comparing it to later advances. Watching a film like this gives the viewer a sense of how it was viewed at the time, what people actually saw. Why homogenize everything so that it all looks like a perpetual, eternal, has always been with us and always will present, erasing every historical marker of the time and the medium and the viewing experience. These techno geeks should find some other trifle to occupy their idle time.

Again, I fail to understand the outrage here. This is a non-destructive process. No one is painting an original Greek statue nor embedding lighting in them. The original film is just as available and pristine as ever. To me, this is the equivalent of doing a CAT Scan analysis on an Egyptian mummy, or an X-ray analysis on a painting to understand the underlying layers and the painting process used. Why is it a travesty to provide a different way of looking at or thinking about a historical artifact? Should we also consider this recent effort to be tragic:
https://www.newscientist.com/article/2230840-listen-to-the-groaning-voice-of-a-3000-year-old-egyptian-mummy/

Because this is what college instructors will show to students and what people will find on YouTube, and it is all they will come to know of the original document. It is a visual document, with its own characteristics and mysteries to be studied and appreciated. Not everything has to look like a contemporary smartphone video.

A quick search on YouTube for “Lumiere La Ciotat” in an anonymized browser brings up the original as the first two, and four out of the top five results. The new high res one is in at number 3, so again, it’s not like the original has been downgraded or made to disappear. Why do you think a college instructor would show this new version to students (assuming the instructor is most likely showing it to talk about the history of film)? Perhaps it would be a great visual aid in a class about artificial intelligence. What about the young person who sees this new version and is fascinated and follows up by learning more about the Lumiere brothers? Is that a bad thing?

In other shocking news, sometimes Shakespeare’s plays are performed set in modern times, or the language is adjusted for modern audiences. Sometimes they even let women play the female roles…

Is repainting a replica to show what the original looked like a problem as well?

Ancient Greek statues were in fact brightly painted. The pale marble we know and love is itself part of an historical process of those colors fading away over time. There is ample evidence of this.

Here’s another way of looking at it, the AI version looks ‘normal’ but that’s because we’ve been trained to see it as film or video, not as real life, even at 60-fps. Remember the ‘soap opera’ effect when 120 MHz tv sets came out, which were good for sport but made drama look weird even though we can’t explain why? Although it’s close and makes it seem more normal, I think we still have a long way to go before we can say it’s as realistic as being there. (which gets to another question I’ve often wondered about music. Wouldn’t it be easier to train a computer to play the instruments on some of these old recordings rather than clean up the recordings which gives them an artificial air? If we did that, could we really claim that we’re hearing it as it was originally played?)

Think of it as a new item, not a “historical reconstruction.” Keep the old stuff; archives are great things (and praise be to archivists). But let’s also explore the new and novel. The major parts of our history are still ahead of us. I’m on the side of the thoughtful geeks.

Comments are closed.