Loading video...

Video Failed to Load

There was a problem loading this video. This could be due to a temporary network issue or the video might be unavailable.

CSS killed another centering hack. 🔥 `align-content: center` works on a plain block. No flex, no grid. Baseline since April 2024 (Chrome 123, Firefox 125, Safari 17.4), widely available on Oct 16. Before this, centering a single item vertically meant turning the box into a flex column. Now ⤵️... show more

IP

2,171 subscribers

24,891 views • 7 days ago •via X (Twitter)

Science & Technology

Anya Rossi• Live Now

Private livecam show

0 Comments

No comments available

Comments from the original post will appear here

Related Videos

JavaScript killed another JavaScript library. 😁 `Intl.Segmenter` has been in Baseline since April 2024 (Chrome 87, Firefox 125, Safari 14.1) and will be widely available on 16 Oct. Before this, "👍🏽".length was 4, and counting what users actually see meant shipping grapheme-splitter. Now ⤵️ the browser does it.

JavaScript killed another JavaScript library. 😁 `Intl.Segmenter` has been in Baseline since April 2024 (Chrome 87, Firefox 125, Safari 14.1) and will be widely available on 16 Oct. Before this, "👍🏽".length was 4, and counting what users actually see meant shipping grapheme-splitter. Now ⤵️ the browser does it.

IP

20,628 views • 8 days ago

Motion graphics created by using Seedance 2.0 on Nsketch AI Prompt ⤵️ Ultra-sleek motion design sequence. Pure black background throughout. White-on-black minimal aesthetic. No cuts — one single continuous unbroken visual evolution. Locked-off static center frame, no camera movement. All motion occurs within the subject. Self-luminous white forms on absolute black, no grain, no shadows. 0–1s [ORIGIN]: A single white point materializes at center frame. It pulses once — three concentric rings ripple outward like a heartbeat and dissolve into black. 1–2s [LINE]: The point stretches into a razor-thin white horizontal line extending edge to edge with elastic ease-out motion. 2–4s [GRID → CIRCLE]: The line multiplies into a precise grid of nine parallel horizontal and vertical lines appearing in staggered succession. The grid warps — outer lines curve inward, the structure rotates, straight lines bend into arcs, converging into a perfect geometric circle outline. 4–6s [WIREFRAME → SOLID SPHERE]: The circle extrudes into a 3D wireframe sphere, every latitude and longitude line crisp, slowly rotating on its vertical axis. The wireframe fills in smoothly — dissolving into a solid white surface with a soft specular highlight in the upper-left quadrant. 6–8s [INTERFACE]: The sphere flattens and unfolds into a clean rectangular plane — a minimal phone interface with rounded corners. Abstract UI elements appear inside: status bar, a grid of rounded app icons, home indicator bar. White line art on black. 8–10s [CUBE → BLOOM]: The interface folds inward, collapsing into an isometric wireframe cube that tumbles once. The cube's six faces peel outward simultaneously, each becoming a smooth white elliptical petal — blooming open like a flower in zero gravity. 10–12s [VORTEX]: The flower dissolves into hundreds of tiny white particles orbiting the center in a controlled elliptical vortex, some clockwise, some counterclockwise, each following a precise mathematical path, flickering softly. 12–13s [WATCH]: The particles decelerate and self-organize — converging and snapping into position to assemble the silhouette of a wristwatch: a clean rectangular case outline with lugs, built entirely from particles. 13–14s [COLLAPSE]: The watch dissolves into fluid white ribbons that compress into a single dense white sphere. It pulses once — brighter than before — then collapses rapidly inward to a single brilliant point. 14–15s [LOGO]: The point holds for one silent beat. Then it expands into a clean minimal brand logo — white sans-serif letterforms on black. A single pulse of warm amber light blooms from center behind the logo, fades to black. White text remains. Style: luxury tech brand motion identity, Bauhaus-minimalist, mathematical precision. Mood: calm, inevitable, intelligent, premium.

Oogie

332,581 views • 6 months ago

A FLEET OF DELIVERY ROBOTS JUST ROLLED THROUGH MANHATTAN ON SCOOTERS AND NOBODY ON THAT SIDEWALK HAS A JOB DELIVERING FOOD ANYMORE. Broadway and Prince St. New York City. a line of humanoid robots on electric scooters crosses the intersection like a morning commute from hell. DoorDash bag on one. Uber Eats on another. Domino's on the next. UPS. DHL. every delivery brand you've ever used. all on robots. all on scooters. all at the same time. the people on the sidewalk are filming and smiling. like it's a parade. like it's CUTE. nobody's thinking about the 2.1 MILLION delivery drivers in the US watching their job ride past them on two wheels. a DoorDash driver makes $15-25 an hour. works in rain. works in snow. dodges cabs. gets no benefits. no insurance. no retirement. eats a cold burrito in their car between drops and hopes the next tip covers gas. every single one of those robots costs $16,000. doesn't need tips. doesn't need gas. doesn't eat the fries. doesn't crash into a taxi and file a $50,000 insurance claim. it just delivers. all day. all night. rain or shine. forever. and there's not ONE of them. there's a FLEET. rolling through Manhattan like they own the street. Domino's next to DHL next to Uber Eats. the brands already picked their side. and it's not the guy on the bike. 4.4 MILLION Americans work in delivery. FOUR POINT FOUR MILLION. and every single one of them just watched their replacement roll through SoHo in a line while tourists clapped. the people filming think this is content. the people delivering think this is a warning. which side of the sidewalk are you on?

Cybronix

402,254 views • 29 days ago

🚨 Alibaba just open sourced a GUI agent that lives inside your webpage and controls it with natural language. It's called Page Agent and it's not a browser extension. It's pure JavaScript no Python, no Puppeteer, no headless browser, no screenshots. Just one script tag and your web app understands natural language. Here's what it actually does: → Embed it with a single tag or npm install → Control any web interface with plain English commands → Text-based DOM manipulation no OCR, no vision models needed → Bring your own LLM (GPT, Claude, Qwen, anything) → Ships a built-in UI with human-in-the-loop support → Turn 20-click ERP/CRM workflows into one sentence → Optional Chrome extension for multi-tab agent tasks → Works on any web app SaaS, admin panels, internal tools Companies are charging $30/month for AI copilots built on this exact idea. This is 3 lines of code. Your users. Your interface. The AI copilot layer for every web app just got open sourced. 1.6K stars. 100% Open Source. (Link in the comments)

🚨 Alibaba just open sourced a GUI agent that lives inside your webpage and controls it with natural language. It's called Page Agent and it's not a browser extension. It's pure JavaScript no Python, no Puppeteer, no headless browser, no screenshots. Just one script tag and your web app understands natural language. Here's what it actually does: → Embed it with a single tag or npm install → Control any web interface with plain English commands → Text-based DOM manipulation no OCR, no vision models needed → Bring your own LLM (GPT, Claude, Qwen, anything) → Ships a built-in UI with human-in-the-loop support → Turn 20-click ERP/CRM workflows into one sentence → Optional Chrome extension for multi-tab agent tasks → Works on any web app SaaS, admin panels, internal tools Companies are charging $30/month for AI copilots built on this exact idea. This is 3 lines of code. Your users. Your interface. The AI copilot layer for every web app just got open sourced. 1.6K stars. 100% Open Source. (Link in the comments)

Ihtesham Ali

135,634 views • 7 months ago

China open-sourced a peanut-sized OCR that parses entire 100-page PDFs in one shot.. It's called Unlimited-OCR. Only 3B params. Runs locally. Every other OCR tool chops your doc into pages and loses the thread. this one reads the whole thing in a single pass. → One-shot "long-horizon" parsing (32K context window) → Multilingual, out of the box → 93% on the standard parsing benchmark (+6 over baseline) → <0.11 error rate past 40 pages → Runs 100% locally on your own hardware → Works with Transformers, vLLM, SGLang, Docker, Ollama, llama.cpp Traditional cloud OCR (Textract, Google Vision, Azure Doc Intelligence) costs $1.50–$15 per 1,000 pages. This runs on your machine. For free. Forever. Baidu built it explicitly to push DeepSeek-OCR one step further. Already at 1.9M downloads on Hugging Face and most people have no idea it exists yet. 100% open source.

China open-sourced a peanut-sized OCR that parses entire 100-page PDFs in one shot.. It's called Unlimited-OCR. Only 3B params. Runs locally. Every other OCR tool chops your doc into pages and loses the thread. this one reads the whole thing in a single pass. → One-shot "long-horizon" parsing (32K context window) → Multilingual, out of the box → 93% on the standard parsing benchmark (+6 over baseline) → <0.11 error rate past 40 pages → Runs 100% locally on your own hardware → Works with Transformers, vLLM, SGLang, Docker, Ollama, llama.cpp Traditional cloud OCR (Textract, Google Vision, Azure Doc Intelligence) costs $1.50–$15 per 1,000 pages. This runs on your machine. For free. Forever. Baidu built it explicitly to push DeepSeek-OCR one step further. Already at 1.9M downloads on Hugging Face and most people have no idea it exists yet. 100% open source.

Superman

1,109,716 views • 2 months ago

New open source library: kugiri (区切り). Already running in production on several of my sites, now public. It splits text into lines, words and characters, the thing every text reveal needs. The difference is where it cuts. Most splitters guess where a line breaks, and the guess falls apart the moment the text does something real: text-wrap: balance, hyphenation, a drop cap, a float, a link that wraps, RTL, Japanese. kugiri does not guess. It asks the browser where the lines already are and cuts exactly there, so the split text looks pixel for pixel like the text before it. No animation, no scroll watching, no opinions. Every unit gets an index and the library gets out of the way. Animate with CSS, the Web Animations API, GSAP or Motion. One file, no dependencies, about 7 kB gzipped. The demo is also the test suite: fifty situations, each checked against what the browser painted, in Chromium, WebKit and Firefox. Demo: npm: Source:

New open source library: kugiri (区切り). Already running in production on several of my sites, now public. It splits text into lines, words and characters, the thing every text reveal needs. The difference is where it cuts. Most splitters guess where a line breaks, and the guess falls apart the moment the text does something real: text-wrap: balance, hyphenation, a drop cap, a float, a link that wraps, RTL, Japanese. kugiri does not guess. It asks the browser where the lines already are and cuts exactly there, so the split text looks pixel for pixel like the text before it. No animation, no scroll watching, no opinions. Every unit gets an index and the library gets out of the way. Animate with CSS, the Web Animations API, GSAP or Motion. One file, no dependencies, about 7 kB gzipped. The demo is also the test suite: fifty situations, each checked against what the browser painted, in Chromium, WebKit and Firefox. Demo: npm: Source:

Edo

46,588 views • 1 month ago

🚨 Anthropic committed up to 1M TPU chips for Claude. Openai is leasing TPUs for chatgpt inference. Here's How kernels work on TPUs (deep dive 2/6 by emilio andere) pallas is Google's answer to kernel writing. a python kernel SDK built on JAX. still very experimental (jax.experimental.pallas). on TPU it compiles through mosaic; on GPU it lowers to triton. if you know CUDA, the syntax will feel familiar but the execution model is completely different. in CUDA, grid=(4,4) launches 16 blocks running simultaneously across SMs. in pallas, those 16 iterations run one after another in lexicographic order. no threads. no warps. no blocks. no occupancy tuning. a TPU is a sequential machine with a very wide vector register — more like a CPU than a GPU. performance comes from width: a 128x128 systolic array doing matmul and an 8x128 SIMD vector unit doing everything else. maximum parallelism on chip: 2, one per TensorCore in megacore mode. three concepts replace CUDA's thread/block/grid hierarchy. Refs are mutable memory references. because execution is sequential, each iteration safely accumulates without atomics. in CUDA you'd need atomics or a separate reduction pass. the memory model is also very different from NVIDIA's. zero hardware caches. VMEM is 32-128 MiB of software-managed scratchpad — 500-1000x larger than GPU shared memory per SM. all data must be explicitly DMA'd from HBM to VMEM before any computation touches it. four levels: HBM → VMEM → VREGs → MXU/VPU, plus SMEM for scalar control data. every byte of data movement is your responsibility. this is like CUDA shared memory except it's 500x bigger and there's no cache fallback. pipelining is mandatory. without double-buffering HBM→VMEM transfers, the MXU just stalls waiting for data. this is the single most important optimization on TPU. and because grid execution is sequential and deterministic, consecutive iterations that need the same input block skip the redundant HBM transfer automatically, impossible on GPU where block execution order is undefined. the compilation pipeline is unlike anything in this series: python → jaxpr → stableHLO → XLA HLO (71+ optimization passes) → LLO (78+ passes) → 322-bit VLIW bundles. the compiler packs instructions for scalar, vector, matrix, and DMA units into a single 322-bit word. everything in that bundle executes in parallel, with no runtime scheduling.

🚨 Anthropic committed up to 1M TPU chips for Claude. Openai is leasing TPUs for chatgpt inference. Here's How kernels work on TPUs (deep dive 2/6 by emilio andere) pallas is Google's answer to kernel writing. a python kernel SDK built on JAX. still very experimental (jax.experimental.pallas). on TPU it compiles through mosaic; on GPU it lowers to triton. if you know CUDA, the syntax will feel familiar but the execution model is completely different. in CUDA, grid=(4,4) launches 16 blocks running simultaneously across SMs. in pallas, those 16 iterations run one after another in lexicographic order. no threads. no warps. no blocks. no occupancy tuning. a TPU is a sequential machine with a very wide vector register — more like a CPU than a GPU. performance comes from width: a 128x128 systolic array doing matmul and an 8x128 SIMD vector unit doing everything else. maximum parallelism on chip: 2, one per TensorCore in megacore mode. three concepts replace CUDA's thread/block/grid hierarchy. Refs are mutable memory references. because execution is sequential, each iteration safely accumulates without atomics. in CUDA you'd need atomics or a separate reduction pass. the memory model is also very different from NVIDIA's. zero hardware caches. VMEM is 32-128 MiB of software-managed scratchpad — 500-1000x larger than GPU shared memory per SM. all data must be explicitly DMA'd from HBM to VMEM before any computation touches it. four levels: HBM → VMEM → VREGs → MXU/VPU, plus SMEM for scalar control data. every byte of data movement is your responsibility. this is like CUDA shared memory except it's 500x bigger and there's no cache fallback. pipelining is mandatory. without double-buffering HBM→VMEM transfers, the MXU just stalls waiting for data. this is the single most important optimization on TPU. and because grid execution is sequential and deterministic, consecutive iterations that need the same input block skip the redundant HBM transfer automatically, impossible on GPU where block execution order is undefined. the compilation pipeline is unlike anything in this series: python → jaxpr → stableHLO → XLA HLO (71+ optimization passes) → LLO (78+ passes) → 322-bit VLIW bundles. the compiler packs instructions for scalar, vector, matrix, and DMA units into a single 322-bit word. everything in that bundle executes in parallel, with no runtime scheduling.

wafer

33,655 views • 3 months ago

NOBODY IN YOUR HOUSE GETS PAID FOR THIS. THAT'S EXACTLY WHY IT GOES FIRST. Watch the water, not the robot. It tips a full bucket across the terrace and the run-off comes back rust orange. That isn't dust. That's months. Then it picks up the long brush and works the same square four times before moving on. No gloves. No kneeling. No standing up slowly afterwards. Last shot: clean tile, edge to edge, right into the corner by the railing where everyone gives up. Now the part nobody says out loud. Cleaning a home is the largest unpaid job on the planet. It has no salary, no title, no invoice. It has never appeared in a single GDP figure. Nobody's CV says nine years of weekend floors. And because it had no price, everyone assumed automation would come for the paid work first. Backwards. The job with no salary is the easiest one on earth to hand over. Nobody negotiates. Nobody strikes. There is no union for Sunday.

NOBODY IN YOUR HOUSE GETS PAID FOR THIS. THAT'S EXACTLY WHY IT GOES FIRST. Watch the water, not the robot. It tips a full bucket across the terrace and the run-off comes back rust orange. That isn't dust. That's months. Then it picks up the long brush and works the same square four times before moving on. No gloves. No kneeling. No standing up slowly afterwards. Last shot: clean tile, edge to edge, right into the corner by the railing where everyone gives up. Now the part nobody says out loud. Cleaning a home is the largest unpaid job on the planet. It has no salary, no title, no invoice. It has never appeared in a single GDP figure. Nobody's CV says nine years of weekend floors. And because it had no price, everyone assumed automation would come for the paid work first. Backwards. The job with no salary is the easiest one on earth to hand over. Nobody negotiates. Nobody strikes. There is no union for Sunday.

BACKWOOD

210,319 views • 2 months ago

I told ClawdBot: "build me a 6-agent system for Polymarket that works while I sleep"... 6 hours while i was asleep. Not a single question. Here's what it built: Monitoring agent - runs 24/7, watches Polymarket for mispriced markets. Spots an anomaly - writes to MEMORY md and pings me on Telegram instantly. Research agent - parses news, X, macro data via browser tool on a cron schedule. Every morning I have a full digest on all open positions before I even check my phone. Trading agent - reads the research agent's memory through Gateway, sees the market hasn't reacted yet, acts. Exec tool in gateway mode with a whitelist - no full access on a live server. Watchdog - HEARTBEAT md every 5 minutes: monitoring running, no errors, positions up to date. Something breaks - immediate Telegram message. All of this - one Gateway. One config.json. Isolation via dmScope: per-agent. The token trick: stopped dumping everything into AGENTS md. Critical rules - bootstrap. Try copytrade my bot here: Everything about markets, patterns, past trades - MEMORY md, semantic search pulls it when needed. Token spend dropped 3x, from $0.40/request to $0.13. First week running: - 47 mispriced markets caught before Polymarket adjusted - avg entry edge: 8-12¢ per position - watchdog fired 3 times, caught a broken RPC before it cost me anything The whole system is plain .md text files. Open an editor, change one line - agent behaves differently. No deploy. No build. A bot responds. An agent earns.

I told ClawdBot: "build me a 6-agent system for Polymarket that works while I sleep"... 6 hours while i was asleep. Not a single question. Here's what it built: Monitoring agent - runs 24/7, watches Polymarket for mispriced markets. Spots an anomaly - writes to MEMORY md and pings me on Telegram instantly. Research agent - parses news, X, macro data via browser tool on a cron schedule. Every morning I have a full digest on all open positions before I even check my phone. Trading agent - reads the research agent's memory through Gateway, sees the market hasn't reacted yet, acts. Exec tool in gateway mode with a whitelist - no full access on a live server. Watchdog - HEARTBEAT md every 5 minutes: monitoring running, no errors, positions up to date. Something breaks - immediate Telegram message. All of this - one Gateway. One config.json. Isolation via dmScope: per-agent. The token trick: stopped dumping everything into AGENTS md. Critical rules - bootstrap. Try copytrade my bot here: Everything about markets, patterns, past trades - MEMORY md, semantic search pulls it when needed. Token spend dropped 3x, from $0.40/request to $0.13. First week running: - 47 mispriced markets caught before Polymarket adjusted - avg entry edge: 8-12¢ per position - watchdog fired 3 times, caught a broken RPC before it cost me anything The whole system is plain .md text files. Open an editor, change one line - agent behaves differently. No deploy. No build. A bot responds. An agent earns.

Lunar

165,099 views • 7 months ago

On April 26th, 2026, the Japan Meteorological Agency's Himawari-8 satellite captured a full day of Earth. Every ten minutes, for twenty-four hours, it photographed a complete hemisphere of the planet. One hundred forty-four individual photographs. No editing for content. No compositing. No CGI. No green screen. No hologram. No drawings. These are photographs, assembled into video. Himawari-8 is a geostationary weather satellite operated by Japan. It orbits at roughly twenty-two thousand miles above the equator, remaining fixed over one spot on the planet, orbiting at the same rate Earth rotates. It was launched in 2014. It takes photographs of Earth in visible light and infrared. It does this every ten minutes, every single day, year after year. These images are available to the public. Meteorologists worldwide use them for weather forecasting. News organizations use them. Scientists use them. You can access them yourself right now. What you're seeing over that twenty-four hour period is the terminator line moving -- the day-night boundary shifting as the planet's rotation carries different regions into and out of sunlight, all viewed from a fixed point in space. It is not a composite. It is not a trick. The curvature you see is not distortion. It is not perspective compression. It is the actual shape of the Earth as photographed by an independent satellite operated by a nation with no interest in perpetuating a NASA conspiracy. This is what Earth looks like from space. This is the globe holding atmosphere in place with gravity as you witness weather patterns moving across its surface. This is reality. (No fisheye lens was harmed in the making of this video.)

On April 26th, 2026, the Japan Meteorological Agency's Himawari-8 satellite captured a full day of Earth. Every ten minutes, for twenty-four hours, it photographed a complete hemisphere of the planet. One hundred forty-four individual photographs. No editing for content. No compositing. No CGI. No green screen. No hologram. No drawings. These are photographs, assembled into video. Himawari-8 is a geostationary weather satellite operated by Japan. It orbits at roughly twenty-two thousand miles above the equator, remaining fixed over one spot on the planet, orbiting at the same rate Earth rotates. It was launched in 2014. It takes photographs of Earth in visible light and infrared. It does this every ten minutes, every single day, year after year. These images are available to the public. Meteorologists worldwide use them for weather forecasting. News organizations use them. Scientists use them. You can access them yourself right now. What you're seeing over that twenty-four hour period is the terminator line moving -- the day-night boundary shifting as the planet's rotation carries different regions into and out of sunlight, all viewed from a fixed point in space. It is not a composite. It is not a trick. The curvature you see is not distortion. It is not perspective compression. It is the actual shape of the Earth as photographed by an independent satellite operated by a nation with no interest in perpetuating a NASA conspiracy. This is what Earth looks like from space. This is the globe holding atmosphere in place with gravity as you witness weather patterns moving across its surface. This is reality. (No fisheye lens was harmed in the making of this video.)

Alex Boge

12,187 views • 5 months ago

Elon Musk just named it. Starmind. One million AI satellites, each a flying data center wider than a 747, cooled by panels that glow their heat into the void. This is not Starlink. Starlink moves internet. Starmind moves thought. Each satellite, the AI1, carries 150 kilowatts of compute on a 70-meter span of solar cells, runs the model on board, and beams the answer down. No building, no grid, no water. The reason is in SpaceX's own IPO filing. The AI market it values at 26.5 trillion dollars hits a wall of electricity and water that Earth cannot supply at a sane price. Orbit erases both. The sun never sets up there, and a 147-year-old law does the cooling: in a vacuum, heat escapes only as infrared, so a panel facing the 3-kelvin void radiates it straight out, no water touched. Run that panel hotter and it sheds heat 16 times faster for double the temperature. What Earth data centers burn rivers to do, a glowing wing does for nothing. Musk flagged the problem himself, in the same filing. SpaceX cannot get enough chips to build this yet, and the fab meant to fix that, Terafab, ten times the size of Tesla's Austin gigafactory, may fail. The prototypes do not fly until 2027. Astronomers are already in revolt over a million new lights crossing the sky. Days ago, Masayoshi Son, who owns the whole AI stack on the ground, called space data centers pointless. Musk just named his and put a million of them on the drawing board. One man is reading his balance sheet. The other is betting on the laws of physics. The piece works out which one the void rewards.

Elon Musk just named it. Starmind. One million AI satellites, each a flying data center wider than a 747, cooled by panels that glow their heat into the void. This is not Starlink. Starlink moves internet. Starmind moves thought. Each satellite, the AI1, carries 150 kilowatts of compute on a 70-meter span of solar cells, runs the model on board, and beams the answer down. No building, no grid, no water. The reason is in SpaceX's own IPO filing. The AI market it values at 26.5 trillion dollars hits a wall of electricity and water that Earth cannot supply at a sane price. Orbit erases both. The sun never sets up there, and a 147-year-old law does the cooling: in a vacuum, heat escapes only as infrared, so a panel facing the 3-kelvin void radiates it straight out, no water touched. Run that panel hotter and it sheds heat 16 times faster for double the temperature. What Earth data centers burn rivers to do, a glowing wing does for nothing. Musk flagged the problem himself, in the same filing. SpaceX cannot get enough chips to build this yet, and the fab meant to fix that, Terafab, ten times the size of Tesla's Austin gigafactory, may fail. The prototypes do not fly until 2027. Astronomers are already in revolt over a million new lights crossing the sky. Days ago, Masayoshi Son, who owns the whole AI stack on the ground, called space data centers pointless. Musk just named his and put a million of them on the drawing board. One man is reading his balance sheet. The other is betting on the laws of physics. The piece works out which one the void rewards.

Shanaka Anslem Perera ⚡

290,770 views • 3 months ago

Just jumped into this 2016 Mercedes-Benz GLE 63 S AMG... and bro, this car is the most overprotective babe I've ever driven 😂 It's like it has trust issues with Nigerian roads! A lot happened while driving this car. 1. Another car edges too close? The steering assist just smoothly slides you away to a safe gap. No honking needed. It works like personal bodyguard. 2. If you are driving too close to a pavement, it drifts you away from the pavement on its own. 3. It senses real emergency (like sudden stop ahead)? Seatbelts tighten hard, front seats adjust, PRE-SAFE mode kicks in, and it brakes itself before you even react. First time e happen, my heart nearly left my body. But after a while... I started loving the extra care. Scary at first, but now? This level of pampering is sweet die. Luxury wey dey save life while you flex. Soup wey sweet na money kill am

Just jumped into this 2016 Mercedes-Benz GLE 63 S AMG... and bro, this car is the most overprotective babe I've ever driven 😂 It's like it has trust issues with Nigerian roads! A lot happened while driving this car. 1. Another car edges too close? The steering assist just smoothly slides you away to a safe gap. No honking needed. It works like personal bodyguard. 2. If you are driving too close to a pavement, it drifts you away from the pavement on its own. 3. It senses real emergency (like sudden stop ahead)? Seatbelts tighten hard, front seats adjust, PRE-SAFE mode kicks in, and it brakes itself before you even react. First time e happen, my heart nearly left my body. But after a while... I started loving the extra care. Scary at first, but now? This level of pampering is sweet die. Luxury wey dey save life while you flex. Soup wey sweet na money kill am

Verterex Properties 🏡 🚗

82,979 views • 8 months ago

Contact sheet prompting is the hottest AI video technique right now 🤯 If you've seen this technique blowing up, here's why it works: You feed AI one image, and it generates a grid of consistent shots—same face, same outfit, different angles and poses. Instant storyboarding. Full creative control. No reshoots. But doing it manually is brutal: → Write the prompt from scratch → Generate the contact sheet → Crop each frame by hand → Feed frames into a video model one at a time → Repeat for every single product That's hours of work per campaign. This n8n automation handles everything: → Upload a character image + product image → AI analyzes both and writes the contact sheet prompt → Nano Banana Pro generates a 6-frame grid → System extracts each frame automatically → Kling 2.5 creates smooth transitions between frames → 5 video clips land in Airtable ready to use No manual cropping. No frame-by-frame prompting. No tedious busywork. What you get in Airtable: - AI-generated creative prompt - Hero image (model + product) - Full 6-frame contact sheet - 5 cinematic video clips - Approval gates before each step All inside n8n + Airtable. Contact sheet prompting on complete autopilot. I recorded a 20-minute Loom showing exactly how I built this. Want the walkthrough + the full n8n workflow + Airtable base? > Like this post > Comment "CONTACT" And I'll send it over (must be following so I can DM)

Contact sheet prompting is the hottest AI video technique right now 🤯 If you've seen this technique blowing up, here's why it works: You feed AI one image, and it generates a grid of consistent shots—same face, same outfit, different angles and poses. Instant storyboarding. Full creative control. No reshoots. But doing it manually is brutal: → Write the prompt from scratch → Generate the contact sheet → Crop each frame by hand → Feed frames into a video model one at a time → Repeat for every single product That's hours of work per campaign. This n8n automation handles everything: → Upload a character image + product image → AI analyzes both and writes the contact sheet prompt → Nano Banana Pro generates a 6-frame grid → System extracts each frame automatically → Kling 2.5 creates smooth transitions between frames → 5 video clips land in Airtable ready to use No manual cropping. No frame-by-frame prompting. No tedious busywork. What you get in Airtable: - AI-generated creative prompt - Hero image (model + product) - Full 6-frame contact sheet - 5 cinematic video clips - Approval gates before each step All inside n8n + Airtable. Contact sheet prompting on complete autopilot. I recorded a 20-minute Loom showing exactly how I built this. Want the walkthrough + the full n8n workflow + Airtable base? > Like this post > Comment "CONTACT" And I'll send it over (must be following so I can DM)

Mike Futia

24,619 views • 8 months ago

whoever leaked this has bigger balls than sense Google Research and MIT ran the same agent jobs 260 different ways for Nature last month: they held the prompts, the tools and the compute budget identical and moved nothing but the wiring between the agents, and the same work swung from 70% worse than a single agent to 80.8% better, averaging out at 0.0% i ran my own single agent against the task list first and it cleared 6 of 10 alone, already past the line where a crew starts subtracting this is Graph Engineering, the layer that decides whether a crew is worth 80% more or 70% less, and it installs into the agent you already pay for: - score your solo agent on the real task first: above roughly 45% success that study predicts zero to negative returns from any crew you put around it - under that line, put one supervisor over the fan out: crews with no correction step amplified their own errors to 17.2x the single agent rate, supervised aggregation held it to 4.4x - give every worker one output and let none of them read a peer's draft, so a wrong step reaches the supervisor instead of four other agents - run the comparison again after every model upgrade, because a better model raises your baseline and a higher baseline is what makes a crew stop paying - keep the single agent alive as the control, the only number that says the wiring is earning its calls turns out the shape does not travel: the biggest win came off a finance task under one supervisor and the worst collapse off a planning task with independent agents my position, and it is the arguable one: a crew is a bet on your own diagram, and the model you pick moves that bet less than one arrow does bookmark this, the three moves that draw those arrows before you pay for one extra call are in the post below ↓

whoever leaked this has bigger balls than sense Google Research and MIT ran the same agent jobs 260 different ways for Nature last month: they held the prompts, the tools and the compute budget identical and moved nothing but the wiring between the agents, and the same work swung from 70% worse than a single agent to 80.8% better, averaging out at 0.0% i ran my own single agent against the task list first and it cleared 6 of 10 alone, already past the line where a crew starts subtracting this is Graph Engineering, the layer that decides whether a crew is worth 80% more or 70% less, and it installs into the agent you already pay for: - score your solo agent on the real task first: above roughly 45% success that study predicts zero to negative returns from any crew you put around it - under that line, put one supervisor over the fan out: crews with no correction step amplified their own errors to 17.2x the single agent rate, supervised aggregation held it to 4.4x - give every worker one output and let none of them read a peer's draft, so a wrong step reaches the supervisor instead of four other agents - run the comparison again after every model upgrade, because a better model raises your baseline and a higher baseline is what makes a crew stop paying - keep the single agent alive as the control, the only number that says the wiring is earning its calls turns out the shape does not travel: the biggest win came off a finance task under one supervisor and the worst collapse off a planning task with independent agents my position, and it is the arguable one: a crew is a bet on your own diagram, and the model you pick moves that bet less than one arrow does bookmark this, the three moves that draw those arrows before you pay for one extra call are in the post below ↓

Argona

892,436 views • 2 months ago

For decades, the greatest church in Florence stood open to the sky. There was a hole at its centre, forty-five metres across, and nobody alive knew how to close it… They had done it on purpose. A model approved in 1367 called for a dome wider than any built since ancient Rome, and the Florentines put up the walls anyway, leaving the problem to whoever came next. The Roman method had been lost for a thousand years. Every dome since had been raised on centering, an enormous wooden scaffold built underneath to hold the masonry until it set. For a span this size, no forest in Tuscany could supply the timber. So the cathedral waited, with rain falling into it… In 1418 the city announced a competition. A goldsmith named Filippo Brunelleschi told them he would build the dome with no centering at all. The masonry would hold itself up as it rose. Many of his contemporaries decided he was a lunatic. One rival, Giovanni da Prato, wrote a public sonnet calling him a "dark, deep wellspring of ignorance" and a "miserable and imbecile beast," and rashly promised to kill himself if the plan ever worked. Brunelleschi answered with a sonnet of his own, advising him to destroy his poems, "lest they sound ridiculous when all the dancing starts, in celebration of that which he now thinks impossible." It took sixteen years. Over four million bricks, laid in a herringbone pattern so each ring locked and carried itself, in two shells, one inside the other, weighing something over twenty-five thousand tons. He designed the hoists that lifted the materials, machines Leonardo da Vinci would later sit and copy into his notebooks. It was finished in 1436. Six hundred years later it is still the largest masonry dome on earth… The director Franco Zeffirelli, a Florentine, once said what that dome does to a person who looks up at it: "When I feel depression creeping in, I return to Florence to gaze at Brunelleschi's dome. If human genius was able to achieve something so great, then I too can and must try to create, to act, to live."

For decades, the greatest church in Florence stood open to the sky. There was a hole at its centre, forty-five metres across, and nobody alive knew how to close it… They had done it on purpose. A model approved in 1367 called for a dome wider than any built since ancient Rome, and the Florentines put up the walls anyway, leaving the problem to whoever came next. The Roman method had been lost for a thousand years. Every dome since had been raised on centering, an enormous wooden scaffold built underneath to hold the masonry until it set. For a span this size, no forest in Tuscany could supply the timber. So the cathedral waited, with rain falling into it… In 1418 the city announced a competition. A goldsmith named Filippo Brunelleschi told them he would build the dome with no centering at all. The masonry would hold itself up as it rose. Many of his contemporaries decided he was a lunatic. One rival, Giovanni da Prato, wrote a public sonnet calling him a "dark, deep wellspring of ignorance" and a "miserable and imbecile beast," and rashly promised to kill himself if the plan ever worked. Brunelleschi answered with a sonnet of his own, advising him to destroy his poems, "lest they sound ridiculous when all the dancing starts, in celebration of that which he now thinks impossible." It took sixteen years. Over four million bricks, laid in a herringbone pattern so each ring locked and carried itself, in two shells, one inside the other, weighing something over twenty-five thousand tons. He designed the hoists that lifted the materials, machines Leonardo da Vinci would later sit and copy into his notebooks. It was finished in 1436. Six hundred years later it is still the largest masonry dome on earth… The director Franco Zeffirelli, a Florentine, once said what that dome does to a person who looks up at it: "When I feel depression creeping in, I return to Florence to gaze at Brunelleschi's dome. If human genius was able to achieve something so great, then I too can and must try to create, to act, to live."

James Lucas

272,881 views • 2 months ago

This guy built an AI pipeline that generates hyperrealistic fashion models in 47 minutes and now dropshippers pay him $1,400 to clone the entire system. He got tired of watching e-com brands lose $8K per photoshoot when a single product angle changed so he built a 9-node workflow that generates 127 product videos from one Pinterest photo without hiring a single model. Here's the exact breakdown: → Claude writes a 34-parameter JSON brand DNA before any image is touched target psychographics, price anchor, vibe matrix, anti-inspiration blacklist → Pinterest becomes the model source library but you can't just download and animate → Kling 2.6 takes that static JPG and turns it into 5-second video but only after the prompt architecture is locked → Negative prompt node runs 41 exclusion terms: no plastic skin, no CGI glow, no symmetry artifacts, no doll face, no synthetic lighting → That one step kills the "AI look" that tanks engagement by 67% in the first 3 seconds → TikTok Studio uploads 19 videos in one batch with zero manual captioning because the brand voice was pre-programmed in step one → Atlas scrapes Amazon product links and auto-generates a Shopify store with hero images, pricing tiers, scarcity copy, and mobile-optimized checkout in 90 seconds → The store goes live before the first TikTok video finishes processing The key move 94% of people skip: you can't animate the photo before you inject the negative prompt. If you send a raw Pinterest image straight into image-to-video the face morphs into a wax figure. The fabric loses texture. The hands grow extra fingers. The whole thing screams "AI" and your CTR dies. His system runs the exclusion filter first so the model moves like she's shot on an iPhone 15 Pro in natural light. One brand hit 2.6M views on TikTok in 11 days with zero paid ads and converted at 3.7% because the videos looked like organic UGC not polished studio content. Brands now pay him $1,400 for the full pipeline setup + $340/month to keep the store synced with new product drops and seasonal video batches. The entire system runs on $23/month in API costs and one laptop. No photographer. No model agency. No product samples. Just a prompt template, a Pinterest account, and the discipline to filter out the AI artifacts before you render movement.

Shade

537,458 views • 4 months ago

opus 5.5 is f*cking cracked at motion design this entire video is code, 0 after effects im open sourcing the prompt template for these motion designs steal it to recreate these ↓ Ask me for: 8 to 12 UI states I want the shape to become (e.g. button, loader, player, slider, toggle, tabs, chart, command palette, toast), pure black and white or one accent color, and a royalty-free song around 120 BPM (e.g. Mixkit, free for commercial use). Dribbble-level UI motion. One shape, never cut: every state is the same element morphing its size, radius and color while its content swaps with a short blur. A cursor drives every change with real clicks and drags. Light warm-gray canvas, black and white components, one clean UI font (Geist). Springs everywhere, a tiny overshoot at most. The camera zooms so each state fills the frame. The last frame is the first frame, so it loops. Banned: bouncy easing, particle bursts, glows, gradients on UI chrome, mismatched icon strokes, dead time, anything that looks like a template. 120 BPM, 7 bars, something happens on every beat. Button → loader → check → dynamic island → music player with a play/pause morph → scrub the progress bar → it becomes a volume slider that stretches when dragged past max → a toggle flips on the beat → the knob becomes a liquid tab indicator → the tabs open into a chart that draws itself, with a tooltip on hover → it collapses into ⌘K → type to filter → enter → toast → back to the button. 1. One HTML file, square 1440x1440. Every style is computed from time inside seek(t): no CSS transitions, no timers, no state carried between frames. 2. Springs are closed-form step responses. A value that changes target many times is the sum of one spring per change, so it stays a pure function of time. 3. The tab indicator's two edges ride different springs, so the leading edge stretches ahead of the trailing one. Same trick for the toggle knob. 4. Drags are direct manipulation: while the cursor is held, the value is computed from its position. On release it springs back from wherever it was. 5. Analyze the song with numpy for the beat grid and start on a downbeat. Place every UI sound by its measured peak. 6. Render with Playwright: 4 subframes per frame, blended with ffmpeg tmix for motion blur at 60fps. 7. Render one frame per beat before the full render. Fix anything off the grid, cramped or hard to read. Never put will-change on anything the camera scales or the text renders blurry. Text that swaps inside a morphing container needs its own enter and exit timing or it overlaps. Make the last frame identical to the first, cursor position and speed included, or the loop stutters. Ask me for the inputs, then show me the state list on the beat grid before you write any code.

opus 5.5 is f*cking cracked at motion design this entire video is code, 0 after effects im open sourcing the prompt template for these motion designs steal it to recreate these ↓ Ask me for: 8 to 12 UI states I want the shape to become (e.g. button, loader, player, slider, toggle, tabs, chart, command palette, toast), pure black and white or one accent color, and a royalty-free song around 120 BPM (e.g. Mixkit, free for commercial use). Dribbble-level UI motion. One shape, never cut: every state is the same element morphing its size, radius and color while its content swaps with a short blur. A cursor drives every change with real clicks and drags. Light warm-gray canvas, black and white components, one clean UI font (Geist). Springs everywhere, a tiny overshoot at most. The camera zooms so each state fills the frame. The last frame is the first frame, so it loops. Banned: bouncy easing, particle bursts, glows, gradients on UI chrome, mismatched icon strokes, dead time, anything that looks like a template. 120 BPM, 7 bars, something happens on every beat. Button → loader → check → dynamic island → music player with a play/pause morph → scrub the progress bar → it becomes a volume slider that stretches when dragged past max → a toggle flips on the beat → the knob becomes a liquid tab indicator → the tabs open into a chart that draws itself, with a tooltip on hover → it collapses into ⌘K → type to filter → enter → toast → back to the button. 1. One HTML file, square 1440x1440. Every style is computed from time inside seek(t): no CSS transitions, no timers, no state carried between frames. 2. Springs are closed-form step responses. A value that changes target many times is the sum of one spring per change, so it stays a pure function of time. 3. The tab indicator's two edges ride different springs, so the leading edge stretches ahead of the trailing one. Same trick for the toggle knob. 4. Drags are direct manipulation: while the cursor is held, the value is computed from its position. On release it springs back from wherever it was. 5. Analyze the song with numpy for the beat grid and start on a downbeat. Place every UI sound by its measured peak. 6. Render with Playwright: 4 subframes per frame, blended with ffmpeg tmix for motion blur at 60fps. 7. Render one frame per beat before the full render. Fix anything off the grid, cramped or hard to read. Never put will-change on anything the camera scales or the text renders blurry. Text that swaps inside a morphing container needs its own enter and exit timing or it overlaps. Make the last frame identical to the first, cursor position and speed included, or the loop stutters. Ask me for the inputs, then show me the state list on the beat grid before you write any code.

zero

1,160,943 views • 19 days ago

F&CK, THIS MEMECOIN TRADER JUST MADE $1.7 MILLION IN 10 DAYS.🤯 He discovered a cheat code for making money with memecoins and decided to share his method with everyone. This guy reportedly pulled in $1.7 million in just 10 days by trading memecoins. Now, he’s breaking down his approach in a FREE guide. He made it accessible to everyone, so you can learn how his strategy works without spending a single dollar. Here’s how he approaches the market: → Trade memecoins using App in bio → Identify promising setups before they gain traction → Get into potential opportunities early → Reinvest profits into new trades → Follow a structured trading strategy And now, he’s turning his attention to the Robinhood memecoin narrative. → Robinhood-related tokens are gaining attention → Some tokens have made huge moves, attracting traders looking for 100x opportunities → He launches and trades tokens on Robinhood himself → He applies his strategy to identify opportunities before they become crowded No need to waste hundreds of dollars on an expensive trading course. Check: Comment “MEMES” if you want his guide.

F&CK, THIS MEMECOIN TRADER JUST MADE $1.7 MILLION IN 10 DAYS.🤯 He discovered a cheat code for making money with memecoins and decided to share his method with everyone. This guy reportedly pulled in $1.7 million in just 10 days by trading memecoins. Now, he’s breaking down his approach in a FREE guide. He made it accessible to everyone, so you can learn how his strategy works without spending a single dollar. Here’s how he approaches the market: → Trade memecoins using App in bio → Identify promising setups before they gain traction → Get into potential opportunities early → Reinvest profits into new trades → Follow a structured trading strategy And now, he’s turning his attention to the Robinhood memecoin narrative. → Robinhood-related tokens are gaining attention → Some tokens have made huge moves, attracting traders looking for 100x opportunities → He launches and trades tokens on Robinhood himself → He applies his strategy to identify opportunities before they become crowded No need to waste hundreds of dollars on an expensive trading course. Check: Comment “MEMES” if you want his guide.

Niche trader

37,446 views • 24 days ago

So... Vertical tabs are out in Chrome. I thought it would make me switch back to it, I really thought it was the one thing I needed in Chrome. I currently use Dia , but here's the thing. In comparison, the implementation of vertical tabs in Chrome is just awful. It barely changed since I saw it in beta. I thought it would greatly improve during the beta phase, but it did not. So here's everything I dislike about it compared to Dia: 1. It's super laggy. 2. When collapsed, the sidebar moves down, meaning the position of the expand/collapse button shifts down too, making it much harder to find. (I ironically couldn't find the first few seconds of the video specifically because of that) 3. No keyboard shortcut to expand/collapse. 4. Collapsed doesn't move the tabs out of view, they are just tiny. 5. The amount of tabs per line in the grid of pinned tabs is not adjusted depending on the number of tabs. 6. In collapsed mode, all the pinned tabs are now in a column, which can push the actual tabs quite far down. ----- It's just awful. It was built into Chrome as a way to just say "hey we also have vertical tabs!" But the UX is simply awful. This has to be the worst implementation of vertical tabs I've seen so far yet, sadly. For comparison, I'm also showing how Dia handles vertical tabs.

So... Vertical tabs are out in Chrome. I thought it would make me switch back to it, I really thought it was the one thing I needed in Chrome. I currently use Dia , but here's the thing. In comparison, the implementation of vertical tabs in Chrome is just awful. It barely changed since I saw it in beta. I thought it would greatly improve during the beta phase, but it did not. So here's everything I dislike about it compared to Dia: 1. It's super laggy. 2. When collapsed, the sidebar moves down, meaning the position of the expand/collapse button shifts down too, making it much harder to find. (I ironically couldn't find the first few seconds of the video specifically because of that) 3. No keyboard shortcut to expand/collapse. 4. Collapsed doesn't move the tabs out of view, they are just tiny. 5. The amount of tabs per line in the grid of pinned tabs is not adjusted depending on the number of tabs. 6. In collapsed mode, all the pinned tabs are now in a column, which can push the actual tabs quite far down. ----- It's just awful. It was built into Chrome as a way to just say "hey we also have vertical tabs!" But the UX is simply awful. This has to be the worst implementation of vertical tabs I've seen so far yet, sadly. For comparison, I'm also showing how Dia handles vertical tabs.

Dy 🌸🌈

187,535 views • 6 months ago