${USER}

Denis Villeneuve reflects on Interstellar

Awesome moment of FSD sentience today with an immediate reaction to a change in situation and then following a human cue (flashing brights) to immediately proceed. 🤯

@DotCSV Pronto podrás controlar qué quieres ver con lenguaje natural, imagino que podrás tener un tab en tu feed con todo el tema de IA fácilmente. For me: algoritmo típico que te enseña con lo que interactúas. Followers: feed cronologica de la gente que sigues Feed custom de XYZ ...

Totalmente cierto. Antes de la compra de Musk, Twitter era una de las mejores herramientas que tenía para documentarme e investigar. El algoritmo mostraba muchos papers, intercambios de ideas entre investigadores, blogs, etc. Todo eso se ha perdido. Mucho ruido y poca señal 🤷‍♂️

‘Pirates of the Caribbean’ director Gore Verbinski says that some CGI in movies now is worse because of over-reliance on Unreal Engine “People started thinking maybe movies can also use Unreal for finished visual effects. So you have this sort of gaming aesthetic entering the world of cinema” (Source: butwhytho.net/2025/11/gore-v…)

Anduril CEO @SchimpfBrian’s biggest lesson from Palantir: “This focus on talent. Everyone says it. In practice, it’s a lot more annoying than you’d think. Often the talented people are super opinionated, kind of aggressive, tell you you’re dumb. It doesn’t always feel good.” “What Karp really recognized and drove at Palantir was this notion of: let’s have really brilliant people that are highly empowered to own problems. It’s probably the number one lesson in how we run things here.”

@mrhyrum I felt absolutely disgusted when watching that scene. Such a good job conveying that yucky-body-shaking feeling, congrats to the team

More fun eating worms! You’ll see that my original version was much longer for this shot. It was the right call to trim it, but I loved how uncomfortably long it was. #disney #disneyanimation #animation #zootopia2

This might be the wildest clip from a @wholemars video ever. I can’t stop laughing 😂😂🤣

Replacing my MacBook geohot.github.io//blog/jekyll/u…

accidentally discovered the website for Departure Mono, made by the makers of Phosphor Icons. what a beauty🧡

This is an incredible video I was just sent from a follower who was using @Tesla FSD V14.1.4 today. His car was going around a construction crew, but then a bus started driving head-on toward the car. It had no time to go around, so FSD reversed & got out of the way. h/t @DylanW913

Nuestro amigo @ControlMision nos pide el fichero STL de #MIURA5 para imprimirlo. Habrá que compartirle un fichero y que lo imprima para que lo tenga en su plató. @PLD_Space

PewDiePie went from Minecraft let's plays to building 10xGPU rigs, training 120B LLMs, distributed vLLM inference, coding webapps, RAGs, speech recognition pipelines, linux sysadmin.. Millions of kids are watching him compile CUDA now. 110M+ subs. WHAT AN ABSOLUTE LEGEND!!

Tony Hawk as Larry David is still the best Halloween costume.

Here’s what we are working on at Adobe Layered image editing 🤯 the future of AI image editing

The single biggest difference between students who succeed at getting a job and people who never enter the industry is how much time they spend creating 3d projects. I’ve never seen anyone who puts in only the assigned school work and gets a job out of university.

Yes and FINALLY an example of AI actually doing something useful for a change Would be curious to know more about this. There's not a lot of details and would like to know what the actual "AI" component of this is and how well it works in practice. We will keep monitoring 🫖

If you are a freelancer in 3D and VFX please read this, the logic applies almost perfectly to our field

🚀 Update Next Scene V2 only 10 days after last version, now live on Hugging Face 👉 huggingface.co/lovis93/next-s… 🎬 A LoRA made for Qwen Image Edit 2509 that lets you create seamless cinematic “next shots” — keeping the same characters, lighting, and mood. I trained this new version on thousands of paired cinematic shots to make scene transitions smoother, more emotional, and real. 🧠 What’s new: • Much stronger consistency across shots • Better lighting and character preservation • Smoother transitions and framing logic • No more black bar artifacts Built for storytellers using @ComfyUI or any diffusers pipeline. Just use “Next Scene:” and describe what happens next , the model keeps everything coherent. 🧩 Try it directly in ComfyUI, or check the thread to launch it on @fal . Open-source, no restrictions, made for filmmakers, animators, and dreamers. @ComfyUI #AIcinema #LoRA #Flux #Qwen #ComfyUI #AIart #GenerativeVideo you can test on comfyui or to try on fal.ai, you can go here : fal.ai/models/fal-ai/… and use my lora link : huggingface.co/lovis93/next-s… start your prompt with "Next Scene:" and lets go !!

Our GPU stack for both NVIDIA and AMD, aside from minimal pieces of signed firmware, is 100% open source and pure Python except for the compiler. It's not using vendor drivers, frameworks, or libraries. That's why it's so easy to make it work on Mac. For compilers, on AMD, we use upstream LLVM, and on NVIDIA, we use the NAK compiler from the MESA project. We plan to replace the compiler with pure tinygrad in a year or two as well. With RANGEIFY merged, our lowering stuff now matches the state of the art, TVM style. We're studying ThunderKittens and TileLang for speed at that level, and should have all this stuff ready in 200 days for the due date of our AMD Llama 405B training contract. Due to tinygrad's small size and pure Python nature, it's the easiest ML library to make progress on, aka fastest slope of improvement. With Megakernel style for scheduling, MODeL_opt style for planning, and E-graph style for symbolic, we should blow past the state of the art in PyTorch and JAX speed. If we do that, NVIDIA's moat is over. It's 1000 lines at most to add a new accelerator to tinygrad. And I don't mean to add a new accelerator with help from a kernel driver, compiler, and libraries. Just 1000 lines of software for the *whole* accelerator speaking right on the PCIe BARs, like what tinygrad is doing with the NVIDIA and AMD GPUs now.

There is much data and lessons to gain from Starship Flights 10 & 11 with the heat shield tile experiments and results, but just like we saw in WWII, the initial reactions to aircraft battle damage assessments, it seems many posting about the results are drawing the wrong conclusions and missing the real engineering conclusions. In WWII, seeing battle damage on returning aircraft, the initial, well-meaning reactions were to redesign the areas damaged to address what was perceived to be design deficiencies. However, this was the wrong conclusion. It was pointed out that aircraft damaged on other areas DID NOT survive and it was those areas that were critical. This was the right conclusion! @SpaceX has learned far more about the survivability of the basic design and areas where damage is not a critical threat to the ship as well as where the truly critical areas are than I think many realize. I also think the real path forward will be continuing iteration and initially some refurbishment between flights, gradually approaching faster turnaround times. Fully rapidly reuse is likely not the near term goal, rather the end state and this will not deter @Space from pushing forward. Yes, much work remains. Yes, it may take a little longer than some want. Yes, this is the right path and Yes, I believe they will ultimately succeed! @ellieinspace @MoonManX22 @EzekielOverstr1 @FelixSchlang @elonmusk