Is this a common approach to reducing weights with "no loss in output quality", assuming this is true? Seems almost too simple to work. If this is doable, would this be applicable to LLMs as well?
Neat with native frame-to-frame generation, but wonder how easy it is to "link" together clips at the intersection, typically the models kind of lose the "momentum" across these stiches, being able to merge things with frame-to-frame between clips might help with this it feels like.
I remember a paper which was posted on HN a few weeks ago where somebody implemented KAN networks in FPGAs, since those can readily be approximated as LUTs.
Is MiniMax H3 capable of logical / technical reasoning, or is it purely art oriented?
I have not tested yet, but I suspect that LLMs with a harness that can execute code can do SPICE simulations rather ok these days? I have seen MCPs for measurement equipment also, maybe they can even close the physical loop?
What are those things exactly? AFAIK, everything we can "know" can be written down, one way or another, even analog circuits.
Also, what SOTA LLMs are you referring to? GPTs been handling analog circuits fine for quite some time, I want to say for at least one year? I've been "pair programming" a bunch of working circuits with GPT models since probably GPT 5 or so.
As far as I can tell, the current ComfyUI nodes don't even do compilation, and I haven't looked into what attention mechanism they're using, but I'm sure with time these durations will come down even more.
Pretty cool.
But assuming you have a 16GB 3060, how long would it take to generate a 15 second clip?
The only one that looks "off" is the beverage ad video during the can opening clip, it still has that "AI smoothening" effect. Good thing this can be done pretty well using traditional rendering.
I feel like for a good while now we'll transition into a process that uses traditional "close-up" rendering/shots + AI generated wide-shots or quick cuts.
Exciting, but also troubling. This being open-weights is a massive win for the community though.
But it represents a coming price pressure that will face the leading foundation models. Open source will prevent runaway costs.
Moreover, it prevents the hair-trigger platform safety checkers from shutting down creative work. Video models are notoriously bad at shutting down a huge number of requests.
Creatives will prefer to work on cloud or private GPU clusters. Waiting 10 minutes for a few seconds of 480p is unacceptable. Hobbyists will have fun, but most actual production work is happening in the cloud.
Artist's time is worth money, and they like to spin up dozens of concurrent generations at a time to more quickly explore the generation state space and make progress on completing work.
I suspect it'll be quite a while until AI gets a good enough aesthetic sense to do this, as even with static HTML websites humans can easily see that it's AI slop.
But as far as your composition tools, they are already available.
There is some debate on the license for those in the US, UK, EU, plus… no comment other than whew those samples though!
You just have to pinkie promise you won't make disney mad and they will send you a licence https://huggingface.co/MiniMaxAI/MiniMax-H3/blob/main/docs/Q...
I remember reading a report where people running AI-model instagram account were using insanely long and detailed prompts about the setting, lighting, makeup, pose, disposition, clothing, etc. about their models. Presumably with some reference image of the face / body to remain consistent across images.
It‘s not clear to me whether a sufficiently detailed prompt can generate actually interesting video with a natural ”texture” (for lack of a better word).
> TRANSITION: a violent WHIP PAN off the rooftop that SMEARS the floating words away with it, motion-streaked —
And the video just didn't do any of that transition at all, it just replaced it with a cut. If you look at the rest of the prompts, you'll find similar lines that are just totally ignored. Except maybe the mouse one, I didn't see anything wrong with that off the bat.
This is AGI.
I see a future here for digital advertising, for design mock-ups, for political propaganda, for robotics model training/world simulation, for fan movies, etc.
But the film industry is struggling to get eyeballs and butts in seats already; using a tool that most people inside-and-outside the industry loathe isn't the answer to fixing that.
At most, I could see it used for pre-viz/pre-production work in Hollywood.
As with the arts, 99.9% of people can't use these models to express vision, get attention, or achieve distribution.
The game is the same as it has always been. You still need hard work, taste, something important to say, the ability to articulate it, good timing, and luck.
Nothing has changed. We can just build faster.
What this does enable is for more to be created that caters to a wider variety of interests. It disrupts existing structures of capital allocation, production, and distribution and gives new players a chance to reshape the game.
The bar will rise and people will still be running at the same pace on the treadmill. There will be more to see, but less time to see it.
And it's literally the first day. Someone will make a distilled 4-8 step LoRA and we're off to the races.