Mion. retweeted
Fable 5.2 did not cheat on this one!!!
This is 100% code and let me explain how this amazing video of the life in the eyes of a Squirrel was built:
- Everything is built, animated and rendered by Python scripts driving Blender Cycles; the sound is synthesized with numpy. No text-to-video, image-to-video or any other generative model was used.
- The one non-procedural input is CC0 photoscans from Poly Haven, downloaded by script: a forest HDRI, leaf-litter and bark textures, and log, stone and twig models. I counted those as "code only".
Time and cost:
- $34.76 in API equivalent
- Total time: 4h15m (of which 3 hours were spent rendering the movie)
Mion. retweeted
For many on the left, SpaceX exists in a bizzare quantum superposition of cognitive dissonance.
SpaceX is simultaneously both a shameless valueless con job AND critical national infrastructure so essential that it must be nationalized.
Mion. retweeted
Come, our friends, time to seek a newer world together!
World Labs is now officially part of @AMD.
Mion. retweeted
Just testing the latest beta of world camera node inside KF Canvas. It's a super easy version of Blender if you don't connect an spz file to it.
Now you can plan and animate the precise camera move and blocking for your gens.
v0.8 will bring Splats and primitives together.
Mion. retweeted
JUST IN: 88% of Nvidia employees voted for Jensen Huang to stay as CEO.
Readers added context they thought people might want to know
88% of Nvidia employees did not vote; the figure is from a Blind survey of 232 verified employees (self-selected, out of ~42,000 total staff) where respondents were asked about Huang remaining CEO.
teamblind.com/news/jensen-hu…
fourweekmba.com/nvidia-employe…
aiweekly.co/alerts/nvidia-…
Mion. retweeted
Character consistency in AI video feels mostly solved.
Location consistency? Still a bit of chaos. New angle, and suddenly the jetty has moved to the other side of the pond. 🙃
So I've been experimenting with an idea: what if I give the model a floor plan?
Here's what I'm testing:
1. Generate the set from a 90° nadir angle. Camera straight down, wide lens, trees and walls leaning out at the edges.
2. Add a compass with N, E, S, W in a corner.
3. Use it as a location ref next to the character sheet.
4. Place every camera by direction in the prompt: "from the north shore looking south", "from the jetty looking east".
So far, the model seems to read it like a map. Still early and still testing, so it's not perfect yet. But it's the most consistent location I've gotten so far.
Would love to hear if you try it, or if you've found something that works better.
🤖 Made with AI
Mion. retweeted
Introducing Blackmagic Ambisonic Microphone! The world’s first 7th order ambisonic microphone with 108 MEMS microphones and a built in spatial encoder, allowing capture of spatial audio over a single USB-C cable! Learn more at bmd.link/K9HUX3
Ilya Sutskever and Sam Altman explain the exact state of AI right now
AI is getting more expensive to build and easier to copy at the same time. here’s how they break it down:
• “the effort it takes to produce one such neural net keeps increasing”
• “the gap between open source and private models may even increase”
• “it’s easy to copy something once you know it can be done”
and the line that explains every AI company today:
“the moat is figuring out what comes next”
watch this before your next AI take ↓
Mion. retweeted
with Blender 5.3 gaussian splat support, we are one step closer to realizing an insane one-of-a-kind world model workflow
During pre-release testing, Claude Fable 5.2 and Claude Mythos 5.2 worked on ambitious projects in many different fields. Here are a few.
From a one-paragraph prompt, Fable 5.2 designed three working electronic devices, including this home voice assistant hub.
Mion. retweeted
15 voices, drums, rain and thunder, all made by air moving through simulated throats. github.com/jamesondh/chimera…
"Le lion est mort ce soir", à la Pow woW. Engine by @jamesonhodge, arrangement by Claude. 🦁
#chimera































