@Vynqorxe the insurance call is the interesting one, though it may show up as downtime coverage first, since a stopped cell is the loss that actually hurts
@0xHoogarden kangaroos can seriously hurt people, so an untested robot walking into that is bolder than it sounds,curious what the fallback was if things went wrong
A FORMER MICHELIN CHEF JUST WATCHED TESLA'S ROBOT CRACK AN EGG AND WENT QUIET
he said he spent 6 years learning to crack an egg with one hand without leaving a single shell fragment in the bowl
optimus did it on camera, with fingers that move like a human's
no spill
no crushed shell
no second attempt
it passes the egg from one hand to the other, cracks it over the bowl, and then walks over to the gym and does squats like it's leg day
think about what this means
an egg is the hardest test for a robot hand
too much force and it breaks
too little and it drops
if a machine can handle that, it can handle dishes, laundry, groceries, and anything else you do at home
the robot isn't coming for factories first
it's coming for your kitchen
@AIBoticssq argon's cherry blossom glow at sunset versus sol's moodier purples with the torii gate, two completely different moods from the same prompt is wild to see side by side
🚨Lovart with GPT Image 2.5 turned 5 plain app screens into a full App Store set
Someone dropped five UI screenshots onto a canvas and pasted one prompt. The headlines and 3D objects for each screen came from the AI.
The video opens on a monitor full of finished posters: navy and purple backgrounds, a phone in the center, "Your Money, Simplified" on the first one. Then it goes back to https://t.co/1g4IhQrrca. GPT Image 2.5 Flare Max is picked in the model list. Five files go in: Home, My Card, Statistic, Transfer and Upload Photo ID. The agent writes a plan for each screen, then renders the posters. The "Fast Verification" headline gets recolored right on the canvas. At the end, one poster becomes a vertical video with floating Finpay cards.
Tool: Lovart, model GPT Image 2.5 Flare Max
Input: five UI files from a fintech app called Finpay, and the app screens stay untouched
Headlines written by AI: Your Money, Simplified / Smart Card Control / Spending Insights / Instant Transfers / Secure Verification
Editable layers: Roboto Bold 133, fill color 9F9FFF, export as Download or Export PSD
Video: Seedance 2.5, 9:16, 5 seconds, 1080p
App Store screens used to be built by hand. Now they are ordered.
Video below ↓
@0xvosss bimanual coordination has quietly ended more pilots than any hardware limit, if the prediction between arms is genuinely unscripted here that's the actual news buried under the folding demo
TWO ROBOT ARMS JUST FOLDED A CARDBOARD BOX WITHOUT A SINGLE HUMAN TOUCH
no joystick no teleop no person nudging it along the label says autonomous and the arms just start working
one holds the flap steady the other folds the next one in
watch how they take turns one waits its beat while the other repositions then they swap who's doing the holding and who's doing the folding
this is the hard part nobody talks about with robots
picking something up is easy two robots working together where one has to react to what the other just did that's a completely different problem
the box doesn't care if a robot built it a human has to fold thousands of these a day in warehouses same motion same flaps same tape line over and over
here it's two machines doing it without stepping on each other
one robot can be a worker two robots that coordinate can be a team
what happens to warehouse work once teams like this scale to hundreds of robots?
@0xHoogarden 4 hours to go from zero to reading angles and picking shots is nuts, if that number holds up billiards might really be the easy version of the physical world problem
@haldenn7 coordinating two arms in real time is basically a distributed systems problem wearing a robot costume, one arm messes up and the other has to know before it becomes part of the problem
TWO ROBOT ARMS JUST FOLDED A CARDBOARD BOX WITHOUT A SINGLE HUMAN TOUCH
no joystick no teleop no person nudging it along the label says autonomous and the arms just start working
one holds the flap steady the other folds the next one in
watch how they take turns one waits its beat while the other repositions then they swap who's doing the holding and who's doing the folding
this is the hard part nobody talks about with robots
picking something up is easy two robots working together where one has to react to what the other just did that's a completely different problem
the box doesn't care if a robot built it a human has to fold thousands of these a day in warehouses same motion same flaps same tape line over and over
here it's two machines doing it without stepping on each other
one robot can be a worker two robots that coordinate can be a team
what happens to warehouse work once teams like this scale to hundreds of robots?
THIS ARM PICKS UP A DIE AND CATCHES A BALL AND NOBODY TOUCHED A CONTROLLER
benjamin choi built it for $300 on a ping-pong table in his basement during the pandemic
no implant just a headband sensor on his forehead and one on his earlobe reading brain waves over bluetooth
his sister's $75 3d printer his own code 75+ versions before this one worked
978 pages of math 23000 lines of code and he taught himself python and c++ in ninth grade
the commercial version of this tech needs open brain surgery and costs hundreds of thousands of dollars his hits 95% accuracy on a headband sensor that's noisier than any implant
what you're watching in this clip isn't a party trick close the fist open the hand grab a die move a ball across a table all just from thought
mit ended up funding the project after seeing it an amputee is now helping him design what comes next
a $300 arm on a ping-pong table just did what used to require a hospital and a hole in someone's skull
@haldenn7 the recovery is the flex, not the jump, getting 80kg off the floor into guard that fast needs way tighter control than launching off leftover momentum