I generated this one using Comfy UI to get the look of the BW photo (with a prompt from Chat GPT) then used Runway to “render the animation”.
I generated this one using Comfy UI to get the look of the BW photo (with a prompt from Chat GPT) then used Runway to “render the animation”.
6 Likes
And rampant theft.
10/10
Maybe it would be more difficult if they removed the runway watermark.
This makes me think of this documentary about Ray Harryhausen and James Cameron is in a segment and he’s saying, if Ray were making films today he would absolutely be excited about this new technology, the world of CG, he’d be on the forefront, that’s what creatives are all about yada yada yada, or something like that. And then it cuts to Harryhausen directly saying something along the lines of “No, I wouldn’t be using computers. I really enjoyed the physical process, working with the materials, the closeness of it, etc. I don’t think I’d be satisfied sitting at a computer all day.” I love that the Harryhausen doc directors kept Cameron’s conjecture in there and then just cut to Harryhausen directly, unambiguously refuting it LOL.
5 Likes
Now he can have de-aged AI DiCaprio fix the “Lake Wissota” line in Titanic.
This is pretty incredible: An AI generated podcast on VFX preproduction.
I think captures my sentiment on the industry pretty accurately…
A lot of people have asked me if I suspect there will be a .com style bust. Other than that it would be called the .ai bust, I think there’s a distinct possibility. However the blast radius will be much larger, well beyond the tech industry (in contrast to 2000).
Imagine you were pitching someone a new post house, but your best current workers can only generate 15% of what it takes to run the place (based on # in article). But you’re convinced this is better than sliced bread? “The numbers will get better, trust me, just can’t give you specifics. To make it happen we need you double our cash position though.”
Most sane people would respond with some variation of the letter F to that. Unless they’ve lived in the tech-bro Silicon Valley echo chamber for far too long.
Not making this up:
OpenAI is booming, with almost $4 billion annual recurrent revenue. However, they are close to losing $5 billion this year alone, remaining highly unprofitable and forcing a new investment round for $6 billion in ‘cash equivalents’ (cash or compute credits) plus $5 billion in debt.
The four big data center providers (Google, Amazon, MSFT, Meta) run at $210B annual capex just for GenAi related build outs.
For a combined projected revenue run rate of $20B (10%). OpenAI accounts for another $4B and the rest is breadcrumbs.
You won’t find a single essay more often cited in Silicon Valley than The Bitter Lesson by Rich Sutton. It’s a short read, but the summary is quite simple: nothing matters beyond compute, and every single data and algorithmic breakthrough leads to AI progress insofar as it allows for greater compute allocation at scale.
In layman’s terms, the answer to current AI limitations is, well, increasing the rate of investment in chips and energy. As mentioned, this view is ubiquitous in the industry, but few essays represent it better than Sam Altman’s recent ‘The Intelligence Age’ blog post.
Long story short, it’s a cry for help to unlock more compute disguised as a beautiful allegory about the future of the world thanks to AI, which leads us to an unequivocal future of accelerated AI spending to make the already-huge investment hole much, much larger.
Silicon Valley’s spending frenzy, fueled by the belief that we have solved everything and need to spend more money on compute, is exceptionally arrogant for two reasons: technological and energetic limitations .
Remember, on a simplified 50,000ft level - aside from research breakthroughs, the majority of this AI is supposed to solve efficiency bottle necks in current human workers. AI is not generating new value, it changes how much existing value costs for the most part. So all this insane amount of money ultimately has to be extracted out of existing businesses that save by eliminating people and other otherwise slow work. And that is assuming that all those AI tools work perfectly before long.
The Capex of the four big ones equals the GDP of Greece. What if they would need to remove more cost than there is in the first place. At some point the math gets very strange.
Exception handler: Process killed. Division by zero error.
Enjoy…
PS: The only smart people in this rodeo are Jensen Huang and his employees. They cash in on the GPUs. They don’t own or run the fabs that make them (all made by TSMC). They don’t own the data centers and the nuclear power plants that run those GPUs. All they own is the design and the relationships. When the final accounting happens, they just shrink back a smaller shop and enjoy the loot floating on a yacht in the Mediterranean. And a few more leather jackets.
That’s a good business plan if you ask me.
3 Likes
Can you please change the clothing of this fashion spot?
3 Likes
2 Likes
Amount of fresh water used to generate a video of a koala riding a surfboard: more than we care to think about.
Video of a koala riding a surfboard that gets several thousand likes on Instagram: priceless.
3 Likes
I’ve been using Illustrator since it was called Illustrator 88. This new feature is bending my mind.
3 Likes
The whole channel is great.
1 Like
1 Like
It’s more likely to be some kind of terminative.
2 Likes
The new Runway Act-One basically replaces Pixar. This isn’t CG, its all generative ai video.
Someday, Nvidia will make a GPU with enough VRAM to generate more than a head and shoulders!