What's extra weird (which you do explain) is that "Trailer for X if it were made in year Y" has been a YouTube cliche for at least a decade. It's AI solving a problem we very much do not have. As you say, it must come more from "well I made this content, now how can I stitch it together?"
Brilliant post, summing up what I’ve experienced as well. I dove into AI filmmaking earlier this year, took the two Curious Refuge classes, made few trailers and spec spots, and then now have just stopped for a hot minute. This is not because I’m not wildly excited about the technology - it’s just that it’s only so interesting after seeing 1,000 cool shots of people standing there sort of moving a bit. Taking a moment to write, specifically stories that can be adapted into AI shorts once world building and consistency is a bit better. Appreciate the thoughts and knowing I’m not alone in having this experience.
We're in total agreement. Fake trailers are a great on-ramp into the space. And Curious Refuge does a spectacular job teaching people how to make those and bringing them in to the space. We need somewhere to go once we're onboarded though. That's what I'm canvassing the internet for.
Let me know if find anything. Feel like eventually it will be ready for interesting experimental narrative work, but not sure if that is a few months or a few years away. The biggest issue for me right now is the bad lip sync. I love doing dialogue and currently it just stops everything dead it’s so bad.
I'm in on alpha testing for some new developers, and... you won't believe what things are going to look like in just a few months. Like, tools that are already developed but not public yet are jaw dropping.
We agree, and use tools like Runway and ElevenLabs where it makes sense, but our Short Film includes something most don't that you're getting at: story, characters, and actual dialog: https://youtu.be/urXLdGGq5lQ
We'll use mocap and video game engines to work around our limited budgets and resources, and hope the next generation of tools like Sora will help us achieve films with talking and character consistency that AUDIENCES CHOOSE TO WATCH - not for AI gimmicks and virality but because it's enjoyable and "a good movie".
That's the ticket! AI can help with story, character, and dialog (www.writeonsaga.com)
AI is useless for real movies, to the points in your article - precisely we use Unreal Engine 5.4 with Metahumans and ElevenLabs for voices - between 2 of us we act/film all scenes and replace with Virtual Production sets, props, costumes, and characters as we make scene after scene for our Feature Film: https://www.youtube.com/playlist?list=PLjsAdQ8VbAN4DEKyWUlGYGn7w27__lCV5
Your concerns about the limitations of AI Video will all soon disappear. AI Video is here to stay, has improved tremendously in a just a year and will continue to evolve. It democratizes filmmaking and empowers the creativity of the masses. In a few years, there will be millions of filmmakers globally. That doesn't mean any of them might become the next Kubrick or Hitchcock. It's just a tool and a very powerful one at that.
Every single tech that has come out for films has had limitations which have helped shape the styles of film for decades. Gen AI, with all its stochastic quirks, is just firming The Sixth Great Age of Filmmaking.
Two things will speed up the process. 1. AI learns on its own. 2. The competition among Video Generation companies. When Samsung and Apple become the dominant players in Smart Phones, every day, the camera improved every year.
I read the article, and you make some good points. However, I think you may be missing the point that the reason that there have not yet been great AI feature-length films yet is because there hasn't been enough time for anyone to make one yet.
I am two-thirds of the way through production on what is likely to be one of the most complex, intricate AI-powered feature films yet created.
The Wizard and The Scholar (or, a Rapscallion Runs Amok in Rye), set in Rye, England in the year 1156, is a buddy story and historical drama of two amazing women who join forces with Robin Hood to solve a mysterious crime, and save the city from an even more egregious threat.
The film is due to premiere on 29 October, in theaters and online. The script and trailers are already winning awards at film festivals and award shows around the globe.
And those of us producing feature-length AI-driven films have, so far, just released trailers so far because, even with AI, films still take a year or so to make, and there has barely been a year since gen AI even started coming close to being capable of pulling this off. (And for those of us attempting to do every single job in a movie studio by ourselves, they also serve as a "proof of concept" that we can actually pull this off). lol. :)
Even huge megastudios using old-school tech only release trailers for the first year or so that they are working on a film, because even with hundreds of millions of dollars and a gajillion people working on it, trailers are all that can be made of a film before the film is ready.
And here we are, trying to make films with technology that has barely been invented, in a style no one has yet developed, while doing everything from writing to foley work to color grading and promotion. Give us time. Amazing stuff is around the corner. :)
Here is a real trailer for the film I am working on, and if you are interested, I will be happy to share a rough cut of the first act with you if you are interested. :)
I think there are still a lot of problems that make the content hard to watch for two hours. Those limitations will be cleared eventually. I will be very interested to watch your film. Please feel free to send the full thing to me.
...Reflections of Evil (by Packard) was a dvd we used to share and stay up all night to watch...interesting to see he is still out there pushing boundaries and working to disturb...
I'm curious to hear any impressions on the movie. I watched some of online but haven't seen the whole thing. My impression is he loves creating warped worlds. AI is great at that.
...yeah it is worth a watch, it is basically a DIY twisted fantasy into one man losing it (him)...he is one foot and tongue in cheek and could be seen as akin to the whole Heidecker-verse for sure (but with Lynchian tendencies)...inspirations would be found footage festival, public access TV, and decaying mindsets....for original gen A.I. video this seems to be best usage imo...i don't know if they coined a name for the phenomenon yet, but I get physically ill looking at A.I. human facsimiles...honestly can't make it more than a minute or two into any of the links (even though I am into really weird shit)...that will improve over time no doubt...i'm not so sure that this will democratize anything as the other commenters implies, but see lots of folks parroting that mindset...so far a.i. video and art has increased the speed of meme production though...and replaced stock services...i can see animation and sfx being great bedfellows with it moving forward as well...
Great article Mike! Wondering if you've seen the Unanswered Oddities series from Neural Viz? https://www.youtube.com/watch?v=YGyvLlPad8Q&t
I saw some of this on reddit, stripped of any context. This is great. I'm glad to see the source. They do a great job making it work as a narrative.
What's extra weird (which you do explain) is that "Trailer for X if it were made in year Y" has been a YouTube cliche for at least a decade. It's AI solving a problem we very much do not have. As you say, it must come more from "well I made this content, now how can I stitch it together?"
Also 1950s Leia looks like Vivien Leigh to me.
Brilliant post, summing up what I’ve experienced as well. I dove into AI filmmaking earlier this year, took the two Curious Refuge classes, made few trailers and spec spots, and then now have just stopped for a hot minute. This is not because I’m not wildly excited about the technology - it’s just that it’s only so interesting after seeing 1,000 cool shots of people standing there sort of moving a bit. Taking a moment to write, specifically stories that can be adapted into AI shorts once world building and consistency is a bit better. Appreciate the thoughts and knowing I’m not alone in having this experience.
We're in total agreement. Fake trailers are a great on-ramp into the space. And Curious Refuge does a spectacular job teaching people how to make those and bringing them in to the space. We need somewhere to go once we're onboarded though. That's what I'm canvassing the internet for.
Let me know if find anything. Feel like eventually it will be ready for interesting experimental narrative work, but not sure if that is a few months or a few years away. The biggest issue for me right now is the bad lip sync. I love doing dialogue and currently it just stops everything dead it’s so bad.
I'm in on alpha testing for some new developers, and... you won't believe what things are going to look like in just a few months. Like, tools that are already developed but not public yet are jaw dropping.
Can’t wait!
We agree, and use tools like Runway and ElevenLabs where it makes sense, but our Short Film includes something most don't that you're getting at: story, characters, and actual dialog: https://youtu.be/urXLdGGq5lQ
We'll use mocap and video game engines to work around our limited budgets and resources, and hope the next generation of tools like Sora will help us achieve films with talking and character consistency that AUDIENCES CHOOSE TO WATCH - not for AI gimmicks and virality but because it's enjoyable and "a good movie".
Yes, that is the solution to all the problems, isn't it? Really good writing :)
Did you use Meta-human or a game engine for the talking heads in this film? They stay quite consistent.
That's the ticket! AI can help with story, character, and dialog (www.writeonsaga.com)
AI is useless for real movies, to the points in your article - precisely we use Unreal Engine 5.4 with Metahumans and ElevenLabs for voices - between 2 of us we act/film all scenes and replace with Virtual Production sets, props, costumes, and characters as we make scene after scene for our Feature Film: https://www.youtube.com/playlist?list=PLjsAdQ8VbAN4DEKyWUlGYGn7w27__lCV5
It's amazing how far AI video has come in such a short amount of time. Thanks for the update!
Your concerns about the limitations of AI Video will all soon disappear. AI Video is here to stay, has improved tremendously in a just a year and will continue to evolve. It democratizes filmmaking and empowers the creativity of the masses. In a few years, there will be millions of filmmakers globally. That doesn't mean any of them might become the next Kubrick or Hitchcock. It's just a tool and a very powerful one at that.
I agree AI video is here to stay. I’d like to see it applied in more formats and more experimentally.
I’m not concerned with the limitations, but I do recognize the current limitations and how they’re shaping the creative energy in the space currently.
Every single tech that has come out for films has had limitations which have helped shape the styles of film for decades. Gen AI, with all its stochastic quirks, is just firming The Sixth Great Age of Filmmaking.
Two things will speed up the process. 1. AI learns on its own. 2. The competition among Video Generation companies. When Samsung and Apple become the dominant players in Smart Phones, every day, the camera improved every year.
Mike,
Your stuff is fun to read, I hope you plan on creating more articles, sorry, refuse to call creativity "content" ::>_<::
Hi Mike:
I read the article, and you make some good points. However, I think you may be missing the point that the reason that there have not yet been great AI feature-length films yet is because there hasn't been enough time for anyone to make one yet.
I am two-thirds of the way through production on what is likely to be one of the most complex, intricate AI-powered feature films yet created.
The Wizard and The Scholar (or, a Rapscallion Runs Amok in Rye), set in Rye, England in the year 1156, is a buddy story and historical drama of two amazing women who join forces with Robin Hood to solve a mysterious crime, and save the city from an even more egregious threat.
The film is due to premiere on 29 October, in theaters and online. The script and trailers are already winning awards at film festivals and award shows around the globe.
And those of us producing feature-length AI-driven films have, so far, just released trailers so far because, even with AI, films still take a year or so to make, and there has barely been a year since gen AI even started coming close to being capable of pulling this off. (And for those of us attempting to do every single job in a movie studio by ourselves, they also serve as a "proof of concept" that we can actually pull this off). lol. :)
Even huge megastudios using old-school tech only release trailers for the first year or so that they are working on a film, because even with hundreds of millions of dollars and a gajillion people working on it, trailers are all that can be made of a film before the film is ready.
And here we are, trying to make films with technology that has barely been invented, in a style no one has yet developed, while doing everything from writing to foley work to color grading and promotion. Give us time. Amazing stuff is around the corner. :)
Here is a real trailer for the film I am working on, and if you are interested, I will be happy to share a rough cut of the first act with you if you are interested. :)
https://youtu.be/LDIhZUGsdek
Enjoy!
I think there are still a lot of problems that make the content hard to watch for two hours. Those limitations will be cleared eventually. I will be very interested to watch your film. Please feel free to send the full thing to me.
Hey Mike:
I’m still putting the third act together, about 15 more scenes to produce.
I’d love to hear your thoughts on it when it is finished, and hear what you think.
Probably five to six weeks.
James
...Reflections of Evil (by Packard) was a dvd we used to share and stay up all night to watch...interesting to see he is still out there pushing boundaries and working to disturb...
I'm curious to hear any impressions on the movie. I watched some of online but haven't seen the whole thing. My impression is he loves creating warped worlds. AI is great at that.
...yeah it is worth a watch, it is basically a DIY twisted fantasy into one man losing it (him)...he is one foot and tongue in cheek and could be seen as akin to the whole Heidecker-verse for sure (but with Lynchian tendencies)...inspirations would be found footage festival, public access TV, and decaying mindsets....for original gen A.I. video this seems to be best usage imo...i don't know if they coined a name for the phenomenon yet, but I get physically ill looking at A.I. human facsimiles...honestly can't make it more than a minute or two into any of the links (even though I am into really weird shit)...that will improve over time no doubt...i'm not so sure that this will democratize anything as the other commenters implies, but see lots of folks parroting that mindset...so far a.i. video and art has increased the speed of meme production though...and replaced stock services...i can see animation and sfx being great bedfellows with it moving forward as well...