Random Access Memories
Вставка
- Опубліковано 26 вер 2024
- Let me present to you likely the first AI-to-film transfer in the world. What you can see below is the result of an AI film sent to a film lab, projected to Kodak film stock, have it developed, and scanned back to digital format in 5K resolution. I think this is the first AI film to do this process, ever. You can get a sense of the process by this workflow I’ll update this with pictures of the actual machines later: hellolaco.io/fi...
From a technical standpoint I wanted to bring together the two far ends of the spectrum of filmmaking: analog celluloid film - let's call it organic - and AI generated footage - let's call it organised. You can check a single frame from the full 5K overscan result: hellolaco.io/fi...
As for the film type I’ve chosen the Kodak Vision3 500T which is the same film stock that was used in movies like Inception, The Dark Knight Rises, La La Land, etc. I will try to have a more comprehensive write-up about the technical details, but for now I will put out one before-after comparison here: hellolaco.io/fi.... Here you can see that the film transfer not just solve some of the main problems with AI footage, but the process gives the image a very nice texture and the colors. Of course some of these can be emulated with digital tools but it will not be the exactly same. Also when it comes to film, everything that happens on film only happens once, this reminds me of Norman Jean Roy’s article about how digital photography kills unique imperfections: www.thecut.com... . As for the original AI fottage, everything is Runway Gen3 Alpha text-to-video only(!), the only effects are the camera shake at the motorcycle scene and the end title had to be retouched because there was an extra letter.
From a story standpoint the video explores the idea of sentient machines: how will it feel for a genAI machine to create the images and videos we prompt? Will it have fear to be turned off? Can it be that exactly this fear will be the pledge of catharsis? In the video this “conscious-fluid” is giving emotions to the lifeless machine.
At the end Plato’s cave is mentioned. This allegory depicts prisoners chained in a cave facing a wall for their whole life. Their perception of reality and the outside world is only through the movements of shadows on that single wall. Doesn’t that remind you how we are training these AI models? For me this mirrors how AI models comprehend the world through training datasets, both presenting limited and maybe distorted views of reality. Just as the cave dwellers form their understanding from indirect representations, AI models develop their "knowledge" from imperfect data, they don’t have direct access to the world. But just like the end of Plato’s story, the prisoners are freed and brought outside their cave. What would be the analogy for a machine in this case? How will be they freed?
As you can see in the beginning the shots are very natural, and you might mistake the narrator for a human soul facing euthanasia. But in the middle we can clearly see things are changing as we see more and more surreal images, as if the machine is trying to explore more of it’s power to change the world he previously thought is fixed and unchangable. Is this the moment for becoming self-aware?
This video is part of my journey to live these questions now without worrying about the answers, just like Rilke wrote:
“Be patient toward all that is unsolved in your heart and try to love the questions themselves, like locked rooms and like books that are now written in a very foreign tongue. Do not now seek the answers, which cannot be given you because you would not be able to live them. And the point is, to live everything. Live the questions now. Perhaps you will then gradually, without noticing it, live along some distant day into the answer.”
There are some easter eggs, as for the unicorn, I will give away the reference for the Blade Runner fans: hellolaco.io/fi...
What a great experiment! Where'd you send it to, just a local film lab?
this made me feel so empty
Very interesting. Without reading the description I could tell that analogue film was part of the project. The whole AI element is intriguing, but lacks a certain human quality that is hard to put my finger on
@@noahnims3942 i think we are on the edge. There are a lot of things I don’t like and it screams AI. But if you compare to videos from last year, I think in 1-2 years we won’t be able to tell the difference between generated and shot footage.
Technically interesting, but has all the AI cliches: unrelated clips held loosely together with a ponderous AI voice narration. Mandatory burning object included.
Unrelated clips?? You can create anything you want. They are only unrelated if you make it so. You sound like a fun time
Unrelated as in even with endless possibilities, the creator fails to maintain a real human connection. The Kuleshov effect and intuitive editing exude meaning whereas this is just a montage of Terrence Malick rip-offs.
@@AdeebFarooqi for example which shot do you feel unrelated to the story?
@@AdeebFarooqi lol
I don't know about you but arson isn't exactly a common theme in my favorite memories. I have nothing against your artistic journey, just want to express a weakness in the video, my friend.
clicked for daft punk, left disappointed 😔
@@HXKUSAI you are not alone!
Just splendid ❤
What AI software did you use? All the footage is generated or did you film some of it and mixed intakte AI in the same scene?
@@elPistoleroFilms all shots are generated with Runway Gen3, text to video only
Que belleza❤
Do you think it would not have been possible with a filmic filter?
@@handfuloflight no. It can get close, but not like this! I will try to do a comparison soon on Linkedin!
@@hellolaco6636 This is pretty incredible man. A cost breakdown would be awesome to see!
this has not the amount of views it deserves. Well Done!
@@jakobwagner3476 thank you!
Boycott AI art.
mind-blowing
This is absolutely stunning. What a cool process. I love it.
AI generated videos are shockingly good now and will get much better in the near future. This can be quite a powerful tool for filmmakers who lack the budget for certain shots or realize during the editing process that they should've filmed more. Once editing softwares start implementing AI video generation features that can easily match the style of existing footage, the possibilities will be endless.
@@NJ-jc9iq yes, but editing wasn’t the pricey/hard part, it was always the location, actors, dresses, equipment.
@@hellolaco6636 Yeah I meant once the software can utilize the already filmed footage to generate new videos that fit in seamlessly. Currently, AI generated footage stands out and is hard to integrate with actual clips since the actors and location will look totally different.
this would be terrible for filmmakers. studios and producers will prefer the low-cost ai images over hiring actual talented crew. ai generated images lack something very important to art: a soul.
@@sierratokl Yeah, I honestly agree, it's terrible for filmmaking overall. It would be like the current reliance on CGI for cost cutting purposes, but considerably worse since not a single person would need to be involved in the process.
Amazing and excellent use case! 🙌
Amazing!
Maravilloso 💎 congrats!!
I'm very inspired with this project. My main videos made with AI follow analogue features but all made digitally. This opened my eyes for the future possibilities
AI steals from real artists. Boycott it
this is breathtaking, thanks for your innovation
the best AI movie I've seen so far.
absoluty amazing wow!