@KenKirtland17 What we will have, eventually, is a combo of conventional and neural ISPs and to think otherwise just seems counter productive.
English
Mathieu Tuli
10.1K posts
Mathieu Tuli
@TuliMathieu
post-training image/video models @meta
San Francisco Katılım Temmuz 2022
888 Takip Edilen986 Takipçiler
@TuliMathieu People keep making this well ackshully explanation.
No. This is not the case.
Digital photography assigning pixel values to sensor reading is not the same as having a neural net draw pixels based on its training etc.
English
I'll never understand these "features" the entire premise of photography is to capture a real state of the universe
由依🍥@youy1qwq
华为长焦拍飞机被识别成鸟了
English
it's not really meant to be an ackshully, more just a future potential - like we can obviously not enable this level of zoom at all if we want, or we can try to build tech that takes low level signals and regenerate the right image. there is early success, but like I said, not perfect yet.
English
@TuliMathieu @KenKirtland17 no, actually, raw files are the actual accounting of photons on each photosite on the sensor. it's the processing afterwards where things are "faked".
English
@hankgreen Or, I believe according to my unvetted AI math, a hockey net is 1.56x larger relative to the puck than a net is to a soccer ball.
English
This guy got 11,000 likes telling me that it’s harder to score in soccer than hockey because soccer has goalies.
yedoye@yedoye_
well you see they also have a guy in there
English
Watched Project Hail Mary (I liked the book). What immediately struck me was how stupid and useless the ship's computer was. LLMs have probably already broken the vast majority of sci-fi worldbuilding done over the last thirty years. Probably explains some of the author ire.
English
People were just not this mentally ill ten years ago
Josh Ferme@JoshFerme
The Odyssey releases on the 17th of July, and I do not want a single one of you watching it. Christopher Nolan has desecrated one of Europe’s foundational myths. This film deserves to fail. Do not give it your money!
English
Fable but cheap and fast and on my phone.
Who’s building this?
English
chicken church, meet turkey library
allison harvard burke@alliharvard
Mysterious abandoned ‘Chicken Church’ built in the Indonesian jungle by man who had a vision from God.
English
@Rahll i mean technically they were always there, but I tend to agree - that being said there are good uses, like low-powered HDR, digital zoom (although this is an extreme example), and auto color grading. it's still new.
English
It's insufferable how companies have been integrating AI into their phone camera apps.
International Cyber Digest@IntCyberDigest
Huawei’s telephoto camera turned a twin-engine jet into a bird.
English
We started Thinking Machines a year and a half ago with a couple of instincts: that people should have much more ability to customize models and do research on them, and that even as AI becomes more autonomous, there's a lot more to build to make humans and AIs work well together.
A lot has happened since then, especially the massive progress in agents, so we wanted to revisit those instincts in light of everything we've learned, argue about them, and write down what we actually believe now.
This is where we landed after a lot of debate. I'm happy with it!
Thinking Machines@thinkymachines
We're building AI that people and organizations can shape and make their own. AI should extend our will and judgment instead of neglecting it; enabling that is the technical challenge we are working to solve. thinkingmachines.ai/blog/the-futur…
English
You can't be offside if you bring the defender along
English
We are... It’s better than human hands in many ways.
Bernt Bornich@BerntBornich
Tomorrow we’re unveiling the most advanced robotic hand in human history
English
@willdepue totally. I think there's two camps really - the video gen as an intermediate and the video gen as a product. in either case, modelling is fundamentally the same, it's more a function of data. one is practical and serves real-world, the other is artist and serves culture.
English
video generation and world understanding for robotics go hand in hand. compression is prediction is generation is understanding. physical world ai will have an unbelievably positive impact on the world, video gen is a side effect
Tim Kostolansky@thkostolansky
i don’t see many positive things coming from lifelike video generation, ie it will mostly result in ensloppifying the digital experience further
English
Preview of one of the main breakthroughs that convinced me that @1x_tech was on the clear path to winning 👀
Stay tuned!!
Bernt Bornich@BerntBornich
Tomorrow we’re unveiling the most advanced robotic hand in human history
English
@linguinelabs trick question no one who uses vscode has had sex
English
Also excited for Muse Video!
The agentic loop feels even more important for video: not just making a single clip, but iterating toward an idea with planning, tools, memory, and taste. We are just getting started!
Alexandr Wang@alexandr_wang
4/ previewing muse video too. competitive on prompt adherence, visual fidelity, temporal consistency. coming to meta ai soon.
English