Video wird geladen...

Video konnte nicht geladen werden

Beim Laden dieses Videos ist ein Problem aufgetreten. Dies könnte an einem vorübergehenden Netzwerkproblem liegen oder das Video ist möglicherweise nicht verfügbar.

How to resolve Draven combats step-by-step Relevant Rulings 322: Cleanup 444.1: Combat Cleanup 376.3.b: Simultaneous triggers 444.2.b: Conquering 448.1.b.2: Conquer scoring at 7 Spiritforged FAQ (winning combat):

TSS houses 🥷🟢🟡 ➡ CCS Houston

3,636 subscribers

26,189 Aufrufe • vor 7 Monaten •via X (Twitter)

Anya Rossi• Live Now

Private livecam show

9 Kommentare

Profilbild von Alex "MaskedSwan" Swan ➡️ RQLA

Alex "MaskedSwan" Swan ➡️ RQLAvor 7 Monaten

Super well made video, clear and to the point and explained so that anyone familiar with the game can understand!!

Profilbild von Sebrex

Sebrexvor 7 Monaten

Is it the same in case of trinity and skyfall of aerion combo?

Profilbild von Charm

Charmvor 7 Monaten

Remind me yugioh

Profilbild von TSS houses 🥷🟢🟡 ➡ CCS Houston

TSS houses 🥷🟢🟡 ➡ CCS Houstonvor 7 Monaten

Very similar to yugioh yes, although main difference is that you can respond to each chain link VS yugioh it all resolves with no response

Profilbild von UberGuitarDude

UberGuitarDudevor 7 Monaten

Super informative! Love this stuff.

Profilbild von Otter 🏳️‍⚧️🏴🇵🇸

Otter 🏳️‍⚧️🏴🇵🇸vor 7 Monaten

This is incredibly helpful, thank you so much for posting this!

Profilbild von Feng

Fengvor 7 Monaten

What if player 2 were to draw in to reaction speed removal. Would that change anything here?

Profilbild von TSS houses 🥷🟢🟡 ➡ CCS Houston

TSS houses 🥷🟢🟡 ➡ CCS Houstonvor 7 Monaten

If opponent draws gust from sentry? No because they will not have priority before draven trigger goes on the chain

Profilbild von TSS houses 🥷🟢🟡 ➡ CCS Houston

TSS houses 🥷🟢🟡 ➡ CCS Houstonvor 7 Monaten

i thought i had subtitles working but apparently not.. ill get it fixed next time

Ähnliche Videos

It's only after you step onto a longship in Assassin's Creed: Valhalla that you realize just how spoiled Black Flag: Resynced and Odyssey made us. No naval combat. No epic sea battles. Just... sailing from point A to B. It's a surprisingly lonely experience. ⛵️

It's only after you step onto a longship in Assassin's Creed: Valhalla that you realize just how spoiled Black Flag: Resynced and Odyssey made us. No naval combat. No epic sea battles. Just... sailing from point A to B. It's a surprisingly lonely experience. ⛵️

𝕯𝖗𝖊𝖆𝖉𝖗𝖔𝖇𝖊𝖗𝖙𝖘

234,753 Aufrufe • vor 1 Monat

🇺🇸🇮🇷 America’s 30,000 Pound “Bunker Buster” Has One Combat Victim — Iran 👀 The GBU-57 was built to destroy what lies deep beneath the earth, and so far, only one country has faced it in combat. The GBU-57 Massive Ordnance Penetrator is America’s massive conventional bunker buster, designed to attack heavily buried underground targets. 🇮🇷 Iran is the only country publicly confirmed to have been struck by the weapon in combat. 💥 2025: Operation Midnight Hammer 🔰 7 B-2 stealth bombers 🔰 14 GBU-57s dropped 🔰 Fordow and Natanz nuclear facilities targeted 🔰 Isfahan separately struck with Tomahawk missiles 🔥 2026: Operation Epic Fury 🔰 U.S. B-2 bombers reportedly used GBU-57s again 🔰 An underground IRGC compound was reportedly targeted 🔰 Other alleged strikes remain unconfirmed 🇺🇸 One country. Two U.S. campaigns. One terrifying weapon. At roughly 30,000 pounds, the GBU-57 is carried by the B-2 Spirit stealth bomber. Built for the targets buried too deep for ordinary bombs. 💣

🇺🇸🇮🇷 America’s 30,000 Pound “Bunker Buster” Has One Combat Victim — Iran 👀 The GBU-57 was built to destroy what lies deep beneath the earth, and so far, only one country has faced it in combat. The GBU-57 Massive Ordnance Penetrator is America’s massive conventional bunker buster, designed to attack heavily buried underground targets. 🇮🇷 Iran is the only country publicly confirmed to have been struck by the weapon in combat. 💥 2025: Operation Midnight Hammer 🔰 7 B-2 stealth bombers 🔰 14 GBU-57s dropped 🔰 Fordow and Natanz nuclear facilities targeted 🔰 Isfahan separately struck with Tomahawk missiles 🔥 2026: Operation Epic Fury 🔰 U.S. B-2 bombers reportedly used GBU-57s again 🔰 An underground IRGC compound was reportedly targeted 🔰 Other alleged strikes remain unconfirmed 🇺🇸 One country. Two U.S. campaigns. One terrifying weapon. At roughly 30,000 pounds, the GBU-57 is carried by the B-2 Spirit stealth bomber. Built for the targets buried too deep for ordinary bombs. 💣

War Radar

37,586 Aufrufe • vor 1 Monat

Dear InterLink Labs 👤 + 🌐 & InterLink Foundation, Your KYC process is TOO SLOW! 😤 400k - 1.4M people stuck in queue for ONE curator is insane! Pls add more curators & speed up approvals. Early miners deserve better! 🙏 👉HOW TO APPLY FOR KYC (Step by Step): Step 1: Go to Verified $ITLG • Open Interlink App > Click on "Verified $ITLG" (the gold coin) • Click "Verify now" • You will see Waiting Queue Step 2: Choose a Curator with LESS Queue Don't just apply to Nigerian curator (419k queue). Check all: • ID #0002 Chinese - 104k queue (FASTEST) ✅ • ID #0010 Apollo - 111k queue ✅ • ID #0009 Global - 508k queue • ID #0001 Indian - 1.4M queue (SLOWEST) ❌ PRO TIP: Pick the one with lowest "In queue" number to get verified faster! Step 3: How to Change Curator • If you already applied to slow one, click "Cancel" then "Apply" to faster one • It will show "Change curator" > Click Confirm • You can only apply to ONE at a time. Step 4: Complete Step 1-3 After KYC? Once your slot is selected (within 24h), you still need to: STEP 1: Migrate your $ITLG - Connect ITLX wallet to transfer your verified $ITLG on-chain > Click "Migrate now" STEP 2: Connect Social - Connect X, Discord etc to improve your HCS > Click "Connect here" STEP 3: Human Credit System - Check your trust score > Click "Check here" - Higher score = More verified $ITLG unlock! QUESTION FOR MY COMMUNITY 👇 Have you completed your Interlink KYC steps? Comment: A) Still in Waiting Queue 😭 B) Done Step 1 - Verified ✅ C) Done Step 2 - Migrated ✅ D) Done All Steps 1-3 ✅ #Interlink #ITLG #ITL #InterlinkLabs

Dear InterLink Labs 👤 + 🌐 & InterLink Foundation, Your KYC process is TOO SLOW! 😤 400k - 1.4M people stuck in queue for ONE curator is insane! Pls add more curators & speed up approvals. Early miners deserve better! 🙏 👉HOW TO APPLY FOR KYC (Step by Step): Step 1: Go to Verified $ITLG • Open Interlink App > Click on "Verified $ITLG" (the gold coin) • Click "Verify now" • You will see Waiting Queue Step 2: Choose a Curator with LESS Queue Don't just apply to Nigerian curator (419k queue). Check all: • ID #0002 Chinese - 104k queue (FASTEST) ✅ • ID #0010 Apollo - 111k queue ✅ • ID #0009 Global - 508k queue • ID #0001 Indian - 1.4M queue (SLOWEST) ❌ PRO TIP: Pick the one with lowest "In queue" number to get verified faster! Step 3: How to Change Curator • If you already applied to slow one, click "Cancel" then "Apply" to faster one • It will show "Change curator" > Click Confirm • You can only apply to ONE at a time. Step 4: Complete Step 1-3 After KYC? Once your slot is selected (within 24h), you still need to: STEP 1: Migrate your $ITLG - Connect ITLX wallet to transfer your verified $ITLG on-chain > Click "Migrate now" STEP 2: Connect Social - Connect X, Discord etc to improve your HCS > Click "Connect here" STEP 3: Human Credit System - Check your trust score > Click "Check here" - Higher score = More verified $ITLG unlock! QUESTION FOR MY COMMUNITY 👇 Have you completed your Interlink KYC steps? Comment: A) Still in Waiting Queue 😭 B) Done Step 1 - Verified ✅ C) Done Step 2 - Migrated ✅ D) Done All Steps 1-3 ✅ #Interlink #ITLG #ITL #InterlinkLabs

open mainnet

28,363 Aufrufe • vor 12 Tagen

🚨🚨🚨 THIS $2 BILLION AIRCRAFT IS MORE DANGEROUS THAN A NUCLEAR WEAPON → AND THE US JUST ACTIVATED IT 🚨🚨🚨 The B-2 Spirit. Inspired by nature's most efficient flyer → the Peregrine Falcon. Millions of years of evolution copied into $2 billion of engineering. Here's what nobody is telling you about what this plane is doing RIGHT NOW in Iran: → Step 1: It took off from Whiteman Air Force Base, Missouri → Step 2: It flew 6,000+ miles WITHOUT being detected by a single radar → Step 3: It entered Iranian airspace completely invisible → Step 4: It dropped the GBU-57 — a 30,000 POUND bunker buster bomb → Step 5: That bomb penetrated 200 FEET underground → Step 6: It destroyed Iran's Fordow nuclear facility — buried inside a MOUNTAIN → Step 7: No missile defense system on Earth saw it coming → Step 8: It flew home. Iran didn't know it was there until the explosions started Why this is MORE important than a nuclear weapon: A nuke destroys everything — cities, civilians, the country that launched it politically. It can NEVER be used without ending the world. The B-2 does the same job SURGICALLY. → It reaches ANY target on the planet within hours → It carries 40,000 lbs of bombs — enough to level a military base → It penetrates 200 feet underground — no bunker is safe → It's INVISIBLE — $400 billion worth of Russian and Chinese air defense systems can't find it → It can strike and return WITHOUT a single soldier touching the ground → Iran spent decades burying its nuclear facilities underground thinking they were safe They weren't. The B-2 Spirit is why Iran will never have a nuclear weapon. Not sanctions. Not diplomacy. Not treaties. A plane they can't see, carrying a bomb they can't stop, hitting a bunker they thought was untouchable. Nature perfected flight over millions of years. The US copied it, strapped 40,000 lbs of explosives to it, and made it invisible. That's not a plane. That's a message. And Iran just received it. 🚨🚨🚨

🚨🚨🚨 THIS $2 BILLION AIRCRAFT IS MORE DANGEROUS THAN A NUCLEAR WEAPON → AND THE US JUST ACTIVATED IT 🚨🚨🚨 The B-2 Spirit. Inspired by nature's most efficient flyer → the Peregrine Falcon. Millions of years of evolution copied into $2 billion of engineering. Here's what nobody is telling you about what this plane is doing RIGHT NOW in Iran: → Step 1: It took off from Whiteman Air Force Base, Missouri → Step 2: It flew 6,000+ miles WITHOUT being detected by a single radar → Step 3: It entered Iranian airspace completely invisible → Step 4: It dropped the GBU-57 — a 30,000 POUND bunker buster bomb → Step 5: That bomb penetrated 200 FEET underground → Step 6: It destroyed Iran's Fordow nuclear facility — buried inside a MOUNTAIN → Step 7: No missile defense system on Earth saw it coming → Step 8: It flew home. Iran didn't know it was there until the explosions started Why this is MORE important than a nuclear weapon: A nuke destroys everything — cities, civilians, the country that launched it politically. It can NEVER be used without ending the world. The B-2 does the same job SURGICALLY. → It reaches ANY target on the planet within hours → It carries 40,000 lbs of bombs — enough to level a military base → It penetrates 200 feet underground — no bunker is safe → It's INVISIBLE — $400 billion worth of Russian and Chinese air defense systems can't find it → It can strike and return WITHOUT a single soldier touching the ground → Iran spent decades burying its nuclear facilities underground thinking they were safe They weren't. The B-2 Spirit is why Iran will never have a nuclear weapon. Not sanctions. Not diplomacy. Not treaties. A plane they can't see, carrying a bomb they can't stop, hitting a bunker they thought was untouchable. Nature perfected flight over millions of years. The US copied it, strapped 40,000 lbs of explosives to it, and made it invisible. That's not a plane. That's a message. And Iran just received it. 🚨🚨🚨

JinWoo Kim, IQ 289

805,723 Aufrufe • vor 6 Monaten

It is highly probable that the #SupremeCourt will decide to dismantle the remaining parts of Section 2 of the #VotingRightsAct, which would eliminate almost half of the black representation in the southern states. Everything occurring at this moment was detailed by rolandsmartin in his book "White Fear." If there has ever been a crucial time for us to unite and combat #whitesupremacy, it is NOW! #BlackAmerica, we must exercise our voting power in high numbers to overrule the #rightwing #Republican Party. I B-Mack also wrote an article how the gutting of Section 2 of the Voting Rights Act will affect Black America, check that out as well! Impact of Eliminating Section 2 of the Voting Rights Act on Black Voter Rights Order White Fear:

It is highly probable that the #SupremeCourt will decide to dismantle the remaining parts of Section 2 of the #VotingRightsAct, which would eliminate almost half of the black representation in the southern states. Everything occurring at this moment was detailed by rolandsmartin in his book "White Fear." If there has ever been a crucial time for us to unite and combat #whitesupremacy, it is NOW! #BlackAmerica, we must exercise our voting power in high numbers to overrule the #rightwing #Republican Party. I B-Mack also wrote an article how the gutting of Section 2 of the Voting Rights Act will affect Black America, check that out as well! Impact of Eliminating Section 2 of the Voting Rights Act on Black Voter Rights Order White Fear:

BMB Empower Network

33,518 Aufrufe • vor 9 Monaten

how to make AI tiktok shop videos (like these): p.s. brands are doing $500k+/m with this strategy ### Step by step: 1. export the top 50 scripts for your product niche via fastmoss or kalodata 2. plug it into Claude ai, have it analyze patterns and trends 3. ask it to write you 3 scripts with this prompt "Analyze the following batch of high-performing TikTok ad scripts for my niche and identify patterns in hooks, structure, tone, pacing, and psychological triggers that likely drove virality, extract the top hook formats and content angles, then write 3 original 15–25 second TikTok scripts for [INSERT PRODUCT NAME HERE] targeting [INSERT DETAILED IDEAL CUSTOMER AVATAR HERE] that naturally highlight these benefits [INSERT 5 MAIN BENEFITS HERE] while speaking to their core pain points, make sure the scripts follow the winning patterns you identified, open with adapted high-performing hooks, and are written in a conversational TikTok-native tone, then paste your analysis and scripts below after reviewing these scripts: [INSERT SCRIPTS HERE]" ### plug ai software You can either manually do this process via eleven labs, capcut, and VEO3 with some inserted photos and b-roll. But I prefer my pre-built tiktok ai automation that produces 120+ videos/day. Once you’ve gathered your scripts, insert them into the software and ask it to create a TikTok-native green screen video featuring an African American male with a deep voice and a tall, muscular build. He should be seated while reading the script, with relevant photos and b-roll footage of [INSERT PRODUCT] playing in the background. You can tweak your avatar and insert clips of your product to customize the automation even more. You can also create podcast style vids, product demo vids, and more. Ai is getting better every day, if you're not learning how to utilize it for your content, you are going to be outcompeted. I made a free step by step playbook thta shos you exactly how to get started and produce 120+ vids/day. Comment "GUIDE" + RT, and I'll send you the playbook (must be following)

Calix

15,325 Aufrufe • vor 1 Jahr

Prompt-to-DeFi must be the 'vibecoding' version for DeFi lovers, powered by INFINIT. As mentioned earlier in my recent posts, this brand-new gateway allows you to be a true creator. Here's how to enter: 1. Head to the Prompt to DeFi gateway on top of INFINIT tech 2. Connect your wallet 3. Use your invite code Here you can get used to the UI with some initial templates. Each line comes up with a step; your job is to guide the agent to perform the exact route from A to B. Key takeaways: + try to explain in the simplest natural language + split the strategy into multiple steps; each step performs an end-to-end route from input to output + modify the steps with the suggestions on INFINIT Head over to the comments for access codes. Enjoy prompting!

andrew.moh

32,125 Aufrufe • vor 9 Monaten

Here's my workflow for reverse-engineering any creator's style using AI: I feed a 65-second video to an AI agent with one prompt "Break down the style, transcribe the script, and give me a replication plan." It came back with a full creative playbook: Step 1: Define your concept hook — One sentence, explainable in under 10 seconds — Must have a visual payoff you can show immediately Step 2: Film your talking head — Dark room, RGB lighting, shoot vertical — Hold your project in your hands during every shot Step 3: Capture 3 types of B-roll — Macro shots of hardware and components — Screen recordings of your creative process — Product reveal in natural light near a window Step 4: Build your VFX explainers — Simple line art with glow effects on pure black — After Effects for control, CapCut for speed Step 5: Design a typography system — Main title card, section headers, word-by-word captions — Define all three before you start editing Step 6: Edit to a 5-act structure — Hook → Conflict → Build → Resolution → CTA It nailed the typography, lighting, music tempo, even the emotional arc. One video in. Full style guide out. Do this for 10 creators you admire. Mix what you like from each. That's how you build your own style.

Here's my workflow for reverse-engineering any creator's style using AI: I feed a 65-second video to an AI agent with one prompt "Break down the style, transcribe the script, and give me a replication plan." It came back with a full creative playbook: Step 1: Define your concept hook — One sentence, explainable in under 10 seconds — Must have a visual payoff you can show immediately Step 2: Film your talking head — Dark room, RGB lighting, shoot vertical — Hold your project in your hands during every shot Step 3: Capture 3 types of B-roll — Macro shots of hardware and components — Screen recordings of your creative process — Product reveal in natural light near a window Step 4: Build your VFX explainers — Simple line art with glow effects on pure black — After Effects for control, CapCut for speed Step 5: Design a typography system — Main title card, section headers, word-by-word captions — Define all three before you start editing Step 6: Edit to a 5-act structure — Hook → Conflict → Build → Resolution → CTA It nailed the typography, lighting, music tempo, even the emotional arc. One video in. Full style guide out. Do this for 10 creators you admire. Mix what you like from each. That's how you build your own style.

The Startup Ideas Podcast (SIP) 🧃

26,560 Aufrufe • vor 6 Monaten

I just cracked the code on AI UGC ads 🤯 51 seconds long. One creator across the entire ad: same face, same voice, same product packaging in every single shot. Talking-head delivery, b-roll cutaways, voiceover running over the b-roll, and the creator moving through different settings. A fully edited, multi-shot ad that actually holds. If you've been stuck generating 8-second clips that look fine but fall apart the second you try to string them into a real ad, no character that holds, no voice that stays the same clip to clip, no product that survives more than one shot, this is the workflow you're missing. Here's exactly how I did it: → Pull a winning authority ad and rewrite the script for your product → Build your creator in ChatGPT Images 2.0 to match the exact buyer demographic → Lock the creator + product as reusable Elements so they stay identical across every shot → Break the script into single-scene shots so each one is a clean generation → Lock the voice on the first clip, then carry that same voice through every shot after → Tag your b-roll shots as voiceover so the line plays over the cutaway instead of lip-syncing → Edit in CapCut: cut the dead air, nudge to 1.1x → Drop the final cut in Captions, add subtitles, export Consistent character, consistent voice, consistent packaging — across talking head, b-roll, voiceover, and multiple settings. No character drift. No robotic voice swap. No disconnected 8-second clips. I put together the full step-by-step playbook with the exact ChatGPT Images 2.0 and Seedance prompts, the element-tagging setup, and the voice trick most people miss. Want it? > Like this post > Comment "UGC" And I'll send it over (must be following so I can DM)

I just cracked the code on AI UGC ads 🤯 51 seconds long. One creator across the entire ad: same face, same voice, same product packaging in every single shot. Talking-head delivery, b-roll cutaways, voiceover running over the b-roll, and the creator moving through different settings. A fully edited, multi-shot ad that actually holds. If you've been stuck generating 8-second clips that look fine but fall apart the second you try to string them into a real ad, no character that holds, no voice that stays the same clip to clip, no product that survives more than one shot, this is the workflow you're missing. Here's exactly how I did it: → Pull a winning authority ad and rewrite the script for your product → Build your creator in ChatGPT Images 2.0 to match the exact buyer demographic → Lock the creator + product as reusable Elements so they stay identical across every shot → Break the script into single-scene shots so each one is a clean generation → Lock the voice on the first clip, then carry that same voice through every shot after → Tag your b-roll shots as voiceover so the line plays over the cutaway instead of lip-syncing → Edit in CapCut: cut the dead air, nudge to 1.1x → Drop the final cut in Captions, add subtitles, export Consistent character, consistent voice, consistent packaging — across talking head, b-roll, voiceover, and multiple settings. No character drift. No robotic voice swap. No disconnected 8-second clips. I put together the full step-by-step playbook with the exact ChatGPT Images 2.0 and Seedance prompts, the element-tagging setup, and the voice trick most people miss. Want it? > Like this post > Comment "UGC" And I'll send it over (must be following so I can DM)

Mike Futia

38,087 Aufrufe • vor 4 Monaten

$Switch Transformer by hand ✍️ ~ 13 steps walkthrough below The Switch Transformer, by Fedus, Zoph, and Shazeer in 2022, is one of the papers that made sparse Mixture of Experts practical at scale. Today, frontier models use MoE to pack enormous parameter counts while activating only a small slice per token: GPT-4, Claude, DeepSeek-V3, and Kimi all follow this pattern. If you want to understand how those models can be huge to store yet still cheap to run, this paper is a good place to start. How does it work? Goal: run five input features through attention, route each one to a single best expert, and read the output off the page. = 1. Given = Input features X1-X5 arrive from the previous block. = 2. Attention matrix = Feed all five features to a query-key attention module to get an attention weight matrix A. = 3. Pooling = Multiply the input features by A to get attention-weighted features Z1-Z5. The effect is to combine features across positions. = 4. Visualize pooling = Z4 is X4 + X5 because the fourth column of A is [0,0,0,1,1]. = 5. Gate values = Multiply the weighted features by the switch matrix. Each gate value says how well expert A, B, or C can probably handle the feature. = 6. Top expert = Pick the row with the highest gate value. Sparse means only the top expert is selected, not all of them. = 7. Routing = Route each Z to its best expert. Every expert has a fixed capacity of 2, so one feature may overflow. = 8. Expert A, linear = Apply the linear layer to the features routed to Expert A. The effect is to combine features across feature dimensions. = 9. Expert A, aggregate = Send the combined feature to the corresponding output column. = 10. Expert B, linear = Apply the linear layer, as in step 8. = 11. Expert B, aggregate = Send the result to the corresponding output column, as in step 9. = 12. Expert C, linear = Apply the linear layer, as in step 8. = 13. Expert C, aggregate = Send the result to the output column. Since one feature exceeded Expert C's capacity, it passes through as-is. Takeaway: a Switch Transformer keeps attention unchanged, then replaces the dense feed-forward network with a sparse set of experts. Most of the parameters sit in the experts, but only a small fraction are used for any one input. That is how GPT-4, Claude, DeepSeek-V3, and Kimi can be enormous to store and still cheap to run. 💾 Save this post!$

Switch Transformer by hand ✍️ ~ 13 steps walkthrough below The Switch Transformer, by Fedus, Zoph, and Shazeer in 2022, is one of the papers that made sparse Mixture of Experts practical at scale. Today, frontier models use MoE to pack enormous parameter counts while activating only a small slice per token: GPT-4, Claude, DeepSeek-V3, and Kimi all follow this pattern. If you want to understand how those models can be huge to store yet still cheap to run, this paper is a good place to start. How does it work? Goal: run five input features through attention, route each one to a single best expert, and read the output off the page. = 1. Given = Input features X1-X5 arrive from the previous block. = 2. Attention matrix = Feed all five features to a query-key attention module to get an attention weight matrix A. = 3. Pooling = Multiply the input features by A to get attention-weighted features Z1-Z5. The effect is to combine features across positions. = 4. Visualize pooling = Z4 is X4 + X5 because the fourth column of A is [0,0,0,1,1]. = 5. Gate values = Multiply the weighted features by the switch matrix. Each gate value says how well expert A, B, or C can probably handle the feature. = 6. Top expert = Pick the row with the highest gate value. Sparse means only the top expert is selected, not all of them. = 7. Routing = Route each Z to its best expert. Every expert has a fixed capacity of 2, so one feature may overflow. = 8. Expert A, linear = Apply the linear layer to the features routed to Expert A. The effect is to combine features across feature dimensions. = 9. Expert A, aggregate = Send the combined feature to the corresponding output column. = 10. Expert B, linear = Apply the linear layer, as in step 8. = 11. Expert B, aggregate = Send the result to the corresponding output column, as in step 9. = 12. Expert C, linear = Apply the linear layer, as in step 8. = 13. Expert C, aggregate = Send the result to the output column. Since one feature exceeded Expert C's capacity, it passes through as-is. Takeaway: a Switch Transformer keeps attention unchanged, then replaces the dense feed-forward network with a sparse set of experts. Most of the parameters sit in the experts, but only a small fraction are used for any one input. That is how GPT-4, Claude, DeepSeek-V3, and Kimi can be enormous to store and still cheap to run. 💾 Save this post!

Tom Yeh

37,410 Aufrufe • vor 2 Monaten

😭🇬🇧STARMER in PANIC after LAWSUIT againt HIM is LAUNCHED by PATRIOTS🚨 Ezra Levant banned from the U.K. — so he’s suing Prime Minister Keir Starmer! Keir Starmer has taken aim at so-called “far-right agitators” ahead of Tommy Robinson's rally on May 16 in London. But Rebel News has a plan to fight back against the British government. Here is how you can help him WIN back OUR freedom. NO.1: Step 1: Sue Keir Starmer He has sued Starmer and you can help him with lawyers cost to take down starmer NO.2: Keep reporting from London Help him get rebel news report in the uk to report the UTK rally NO.3 Send an email to Secretary of State Sarah B. Rogers Write an email to the U.S. diplomat in charge of international free speech. Tell her that U.S. citizens (like Tommy’s friend Don Keith) have been blacklisted. GET STARMER IN COURT Rebel News | Ezra Levant 🍁🚛 | Tommy Robinson 🇬🇧

😭🇬🇧STARMER in PANIC after LAWSUIT againt HIM is LAUNCHED by PATRIOTS🚨 Ezra Levant banned from the U.K. — so he’s suing Prime Minister Keir Starmer! Keir Starmer has taken aim at so-called “far-right agitators” ahead of Tommy Robinson's rally on May 16 in London. But Rebel News has a plan to fight back against the British government. Here is how you can help him WIN back OUR freedom. NO.1: Step 1: Sue Keir Starmer He has sued Starmer and you can help him with lawyers cost to take down starmer NO.2: Keep reporting from London Help him get rebel news report in the uk to report the UTK rally NO.3 Send an email to Secretary of State Sarah B. Rogers Write an email to the U.S. diplomat in charge of international free speech. Tell her that U.S. citizens (like Tommy’s friend Don Keith) have been blacklisted. GET STARMER IN COURT Rebel News | Ezra Levant 🍁🚛 | Tommy Robinson 🇬🇧

Baraka Amuri

36,713 Aufrufe • vor 4 Monaten

Connor Young - Rise, Fall & Rise Again 👏 The 20yr Old CF is the top-scorer in League 2 with Edinburgh City after his hat-trick yesterday. That makes it 13 ⚽️in 13 Apps - Hibs Youth Team player - Joined Rangers at 18 - Scored 7⚽️for the Rangers B Team in Lowland - Dropped to Clyde after Rangers withdrew from Lowland - Only started , 13 times for a struggling Clyde, scoring 3 ⚽️, including one in the Scottish Cup Now signed for Edinburgh City, he has really been able to see his goalscoring instincts and with his current form will be on a lot of Championship & SPFL Premier League sides radar with January coming up. Nothing better than seeing a kid being let go by a Premier club and being able to find somewhere to hit form 👏

Connor Young - Rise, Fall & Rise Again 👏 The 20yr Old CF is the top-scorer in League 2 with Edinburgh City after his hat-trick yesterday. That makes it 13 ⚽️in 13 Apps - Hibs Youth Team player - Joined Rangers at 18 - Scored 7⚽️for the Rangers B Team in Lowland - Dropped to Clyde after Rangers withdrew from Lowland - Only started , 13 times for a struggling Clyde, scoring 3 ⚽️, including one in the Scottish Cup Now signed for Edinburgh City, he has really been able to see his goalscoring instincts and with his current form will be on a lot of Championship & SPFL Premier League sides radar with January coming up. Nothing better than seeing a kid being let go by a Premier club and being able to find somewhere to hit form 👏

John Walker

176,641 Aufrufe • vor 1 Jahr

U-Net by hand ✍️ ~ 17 steps walkthrough below I consider U-Net as a key milestone in deep learning, the first image-to-image model that really worked! It came out of medical imaging, an unusual place, not from NeurIPS or CVPR or ACL. Now it is the backbone of diffusion models, which you see in almost all modern image generation models. I drew the network as a C so the matrix multiplication flows naturally down. Tilt your head to the right and it is a U again. 🤣 Goal: push a 3 x 16 image down to a 2 x 4 bottleneck and back out again, filling in every cell yourself. = 1. Given = An image of three channels, R, G and B, sixteen pixels wide, and every kernel the network will use. = 2. Convolution 1 = Let us slide the first kernel over the image. Each output is one multiply-and-add over a 2 x 3 window, and the result is the green feature map. = 3. Find the maxima = We circle the largest value in each 1 x 2 window. Circling first is worth the extra step: it is the pooling decision, made before anything is written down. = 4. Max pool 1 = Let us copy those maxima down. Sixteen columns become eight, and half the detail is gone for good. = 5. Convolution 2 = We convolve again with the second kernel, deeper into the contracting path. The feature map is blue now. = 6. Find the maxima again = Same move as step 3, on the blue map. = 7. Max pool 2 = Eight columns become four. = 8. The bottleneck = Let us convolve once more. This is the bottom of the U, a 2 x 4 block that is everything the network kept. = 9. Spread it out = We start back up. The transposed convolution writes each bottleneck value into a wider grid, leaving gaps between them. = 10. Transposed convolution 1 = Let us fill those gaps by convolving over the spread-out grid. Four columns become eight. = 11. The first skip = We copy the encoder's matching row straight across. This is the skip connection, and it is the whole reason a U-Net can recover detail that pooling threw away. = 12. Convolution with the skip = Let us convolve the upsampled features together with the copied ones. = 13. Spread it out again = Same as step 9, one level up. = 14. Transposed convolution 2 = Eight columns become sixteen, back to the width we started at. = 15. The second skip = The encoder's first feature map comes across, the one made before any pooling happened. = 16. Convolution and ReLU = We convolve, then cross out every negative and set it to zero. = 17. Output convolution = Let us apply the last kernel. Out comes R', G' and B', an image the same size as the one we started with. The outputs: R' = [3, 0, 7, 0, 7, 0, 17, 0, 3, 0, 9, 0, 2, 0, 6, 0] G' = [1, 20, 1, 10, 1, 12, 1, 19, 2, 5, 1, 11, 1, 3, 1, 7] B' = [4, 20, 8, 10, 8, 12, 18, 19, 5, 5, 10, 11, 3, 3, 7, 7] Congrats! You just calculated a U-Net by hand. 💾 Save this post!

U-Net by hand ✍️ ~ 17 steps walkthrough below I consider U-Net as a key milestone in deep learning, the first image-to-image model that really worked! It came out of medical imaging, an unusual place, not from NeurIPS or CVPR or ACL. Now it is the backbone of diffusion models, which you see in almost all modern image generation models. I drew the network as a C so the matrix multiplication flows naturally down. Tilt your head to the right and it is a U again. 🤣 Goal: push a 3 x 16 image down to a 2 x 4 bottleneck and back out again, filling in every cell yourself. = 1. Given = An image of three channels, R, G and B, sixteen pixels wide, and every kernel the network will use. = 2. Convolution 1 = Let us slide the first kernel over the image. Each output is one multiply-and-add over a 2 x 3 window, and the result is the green feature map. = 3. Find the maxima = We circle the largest value in each 1 x 2 window. Circling first is worth the extra step: it is the pooling decision, made before anything is written down. = 4. Max pool 1 = Let us copy those maxima down. Sixteen columns become eight, and half the detail is gone for good. = 5. Convolution 2 = We convolve again with the second kernel, deeper into the contracting path. The feature map is blue now. = 6. Find the maxima again = Same move as step 3, on the blue map. = 7. Max pool 2 = Eight columns become four. = 8. The bottleneck = Let us convolve once more. This is the bottom of the U, a 2 x 4 block that is everything the network kept. = 9. Spread it out = We start back up. The transposed convolution writes each bottleneck value into a wider grid, leaving gaps between them. = 10. Transposed convolution 1 = Let us fill those gaps by convolving over the spread-out grid. Four columns become eight. = 11. The first skip = We copy the encoder's matching row straight across. This is the skip connection, and it is the whole reason a U-Net can recover detail that pooling threw away. = 12. Convolution with the skip = Let us convolve the upsampled features together with the copied ones. = 13. Spread it out again = Same as step 9, one level up. = 14. Transposed convolution 2 = Eight columns become sixteen, back to the width we started at. = 15. The second skip = The encoder's first feature map comes across, the one made before any pooling happened. = 16. Convolution and ReLU = We convolve, then cross out every negative and set it to zero. = 17. Output convolution = Let us apply the last kernel. Out comes R', G' and B', an image the same size as the one we started with. The outputs: R' = [3, 0, 7, 0, 7, 0, 17, 0, 3, 0, 9, 0, 2, 0, 6, 0] G' = [1, 20, 1, 10, 1, 12, 1, 19, 2, 5, 1, 11, 1, 3, 1, 7] B' = [4, 20, 8, 10, 8, 12, 18, 19, 5, 5, 10, 11, 3, 3, 7, 7] Congrats! You just calculated a U-Net by hand. 💾 Save this post!

Tom Yeh

17,722 Aufrufe • vor 2 Monaten

On June 21, 2025, seven B-2 Spirit stealth bombers departed Whiteman Air Force Base in Missouri, flew east across the Atlantic, and dropped twelve 30,000-pound GBU-57 Massive Ordnance Penetrators on Iran’s nuclear enrichment facilities at Fordow, Isfahan, and Natanz. That operation was called Midnight Hammer. It lasted 36 hours. The chairman of the Joint Chiefs quoted a trailing pilot as saying it was “the brightest explosion I’ve ever seen.” Each bomb weighed fifteen tons. It was the first time the GBU-57, the largest conventional weapon in the American arsenal, had been used in combat. Eight months later, the B-2s went back. CENTCOM confirmed on March 1 that B-2 stealth bombers armed with 2,000-pound bombs struck Iran’s “hardened ballistic missile facilities” during overnight operations as part of Operation Epic Fury. Fox News reported, citing a US defense official, that four B-2s dropped “dozens of 2,000 lb bombs on underground ballistic missile sites in Iran.” Open-source flight analysts tracked four aircraft under the callsign PETRO41 passing over the Strait of Gibraltar on the return leg, the same routing used during Midnight Hammer. The bombers refueled over the central Atlantic from KC-46 tankers staged at Lajes Air Base in the Azores. On return, they diverted to Dyess Air Force Base in Texas due to weather at Whiteman. Planet satellite imagery from March 1 confirmed the damage: collapsed tunnel entrances at Tabriz North Missile Base, compared against February 23 baseline images published by Newsweek. The weapon choice is the story nobody is telling. Midnight Hammer used 30,000-pound bunker busters because the target was a mountain. Fordow is an enrichment facility buried deep inside solid rock. You need the heaviest bomb ever built to reach it. Operation Epic Fury used 2,000-pound GBU-31 JDAMs with BLU-109 penetrating warheads because the targets are structurally different. Iran’s ballistic missile facilities are not single deep bunkers. They are what defense analysts call “missile cities,” networks of interconnected tunnels carved into mountainsides, with dozens of entrance points, storage chambers, and in some cases, launch fissures in the rock ceilings through which missiles can be fired from inside the mountain itself. You do not kill a missile city with one enormous bomb. You collapse it entrance by entrance. Dozens of 2,000-pound precision strikes across multiple tunnel openings. That is what “dozens of bombs” means. That is what the satellite imagery at Tabriz confirms: not a single crater but a systematic closure of the tunnel network. The mission duration also tells a story. The War Zone reported that the B-2s flew from CONUS because the UK denied access to RAF Fairford and Diego Garcia, the two forward bases equipped for sustained bomber operations. That forced 36-plus-hour round trips from Missouri. Starmer reversed that decision on March 1, granting base access. Future B-2 sorties can now stage from Fairford, cutting transit time dramatically. The four-week war just became an accelerating air campaign. Two B-2 missions against Iran in eight months. The first destroyed the nuclear program with the largest bombs ever dropped. The second is methodically collapsing every mountain that houses the missiles built to avenge the first. The bomber that ended the nuclear threat is now ending the retaliatory capability. Same aircraft. Same air base. Different target architecture. Same result: the mountains of Iran are being opened from the outside.

On June 21, 2025, seven B-2 Spirit stealth bombers departed Whiteman Air Force Base in Missouri, flew east across the Atlantic, and dropped twelve 30,000-pound GBU-57 Massive Ordnance Penetrators on Iran’s nuclear enrichment facilities at Fordow, Isfahan, and Natanz. That operation was called Midnight Hammer. It lasted 36 hours. The chairman of the Joint Chiefs quoted a trailing pilot as saying it was “the brightest explosion I’ve ever seen.” Each bomb weighed fifteen tons. It was the first time the GBU-57, the largest conventional weapon in the American arsenal, had been used in combat. Eight months later, the B-2s went back. CENTCOM confirmed on March 1 that B-2 stealth bombers armed with 2,000-pound bombs struck Iran’s “hardened ballistic missile facilities” during overnight operations as part of Operation Epic Fury. Fox News reported, citing a US defense official, that four B-2s dropped “dozens of 2,000 lb bombs on underground ballistic missile sites in Iran.” Open-source flight analysts tracked four aircraft under the callsign PETRO41 passing over the Strait of Gibraltar on the return leg, the same routing used during Midnight Hammer. The bombers refueled over the central Atlantic from KC-46 tankers staged at Lajes Air Base in the Azores. On return, they diverted to Dyess Air Force Base in Texas due to weather at Whiteman. Planet satellite imagery from March 1 confirmed the damage: collapsed tunnel entrances at Tabriz North Missile Base, compared against February 23 baseline images published by Newsweek. The weapon choice is the story nobody is telling. Midnight Hammer used 30,000-pound bunker busters because the target was a mountain. Fordow is an enrichment facility buried deep inside solid rock. You need the heaviest bomb ever built to reach it. Operation Epic Fury used 2,000-pound GBU-31 JDAMs with BLU-109 penetrating warheads because the targets are structurally different. Iran’s ballistic missile facilities are not single deep bunkers. They are what defense analysts call “missile cities,” networks of interconnected tunnels carved into mountainsides, with dozens of entrance points, storage chambers, and in some cases, launch fissures in the rock ceilings through which missiles can be fired from inside the mountain itself. You do not kill a missile city with one enormous bomb. You collapse it entrance by entrance. Dozens of 2,000-pound precision strikes across multiple tunnel openings. That is what “dozens of bombs” means. That is what the satellite imagery at Tabriz confirms: not a single crater but a systematic closure of the tunnel network. The mission duration also tells a story. The War Zone reported that the B-2s flew from CONUS because the UK denied access to RAF Fairford and Diego Garcia, the two forward bases equipped for sustained bomber operations. That forced 36-plus-hour round trips from Missouri. Starmer reversed that decision on March 1, granting base access. Future B-2 sorties can now stage from Fairford, cutting transit time dramatically. The four-week war just became an accelerating air campaign. Two B-2 missions against Iran in eight months. The first destroyed the nuclear program with the largest bombs ever dropped. The second is methodically collapsing every mountain that houses the missiles built to avenge the first. The bomber that ended the nuclear threat is now ending the retaliatory capability. Same aircraft. Same air base. Different target architecture. Same result: the mountains of Iran are being opened from the outside.

Shanaka Anslem Perera ⚡

52,963 Aufrufe • vor 7 Monaten

Poetry in motion, even in the middle of a fight. 🌸⚔️ Prompt Below ⤵️ Generate a 30-second, 16:9, 24fps high-budget mythological Wuxia action short film. AAA cinematic quality, photorealistic live-action. No subtitles, no UI, no logos, no text, no voiceover, no background music—retain only environmental and combat audio: mountain gusts, silk robes snapping, foot-taps on roof tiles, flying stone tiles shattering, blade hums, heavy air-displacements, and focused breathing. 【Visual Style & Characters】 Protagonist: A female martial arts master in flowing crimson-and-gold silk robes, hair held by a jade pin, wielding a flexible steel double-edged Jian (straight sword). Ethereal, lethal, calm aura. Antagonists: Two distinct shadow monks in dark silk wraps. Monk A wields a heavy iron three-section staff for variable-range whipping and trapping strikes. Monk B wields a pair of heavy iron ring-daggers for high-speed spinning slices. Must remain two distinct fighters throughout. 【Combat Dynamics & Actions】 Super-speed Wuxia agility and gravity-defying aerial swordplay. Continuous motion—no pauses, no eye-contact pauses, no static posing. Every jump, aerial flip, wall-run, and slide binds directly into a sword strike or parry. 【Camera & Environment】 Environment: Ancient mountain temple courtyard surrounded by misty cliffs, terracotta roof tiles, burning incense burners, and floating cherry blossom petals. Camera: Feral, ultra-dynamic camera tracking aerial movements. Uses low-angle tilts, quick whip-pans, camera lagging slightly behind hyper-speed leaps before snapping back, and lens-grazing debris. 【Timeline】 0–6s: Immediate combat. Monk A’s iron staff whips into frame; Monk B dives from a temple roof. Protagonist low-slides across tile floors under the staff whip, uses a one-hand ground spin to parry B’s daggers, and leaps upward to dodge a low staff sweep. 6–12s: Protagonist step-runs along the vertical surface of a stone pillar, executes a reverse aerial twist-slash at Monk B, lands onto a roof beam, and side-flips through Monk A’s staff arc. 12–18s: Midpoint close-quarters tangle. Trapped between both attackers on a narrow stone bridge. Camera rapid orbital whip. Protagonist drops flat to let their weapons collide, then performs a 360° spinning leg sweep clearing both. 18–24s: Peak aerial agility. Z-pattern wall-rebounds → hilt strikes to weapon joints → mid-air sword deflections → cartwheel dodges through incense smoke. 24–30s: Climax finish. Protagonist slices straight through their combined defense. A rapid triple-strike disarms Monk B, sending him flying into a stone incense burner that shatters on impact. She instantly pivots, leaping into a 360° aerial full-body rotation sword cleave against Monk A, blowing back mist and petals. Final frame: protagonist lands lightly on one foot, sword held diagonally, robes settling in the wind, fast dolly-in to face, hard cut. 【Negative Prompt / Exclusions】 Stance-holding, slow motion, turn-based attacks, idle stances, prolonged eye-locks, magic glowing spell circles, excessive particle effects obscuring action, UI, text, watermarks, subtitles.

Poetry in motion, even in the middle of a fight. 🌸⚔️ Prompt Below ⤵️ Generate a 30-second, 16:9, 24fps high-budget mythological Wuxia action short film. AAA cinematic quality, photorealistic live-action. No subtitles, no UI, no logos, no text, no voiceover, no background music—retain only environmental and combat audio: mountain gusts, silk robes snapping, foot-taps on roof tiles, flying stone tiles shattering, blade hums, heavy air-displacements, and focused breathing. 【Visual Style & Characters】 Protagonist: A female martial arts master in flowing crimson-and-gold silk robes, hair held by a jade pin, wielding a flexible steel double-edged Jian (straight sword). Ethereal, lethal, calm aura. Antagonists: Two distinct shadow monks in dark silk wraps. Monk A wields a heavy iron three-section staff for variable-range whipping and trapping strikes. Monk B wields a pair of heavy iron ring-daggers for high-speed spinning slices. Must remain two distinct fighters throughout. 【Combat Dynamics & Actions】 Super-speed Wuxia agility and gravity-defying aerial swordplay. Continuous motion—no pauses, no eye-contact pauses, no static posing. Every jump, aerial flip, wall-run, and slide binds directly into a sword strike or parry. 【Camera & Environment】 Environment: Ancient mountain temple courtyard surrounded by misty cliffs, terracotta roof tiles, burning incense burners, and floating cherry blossom petals. Camera: Feral, ultra-dynamic camera tracking aerial movements. Uses low-angle tilts, quick whip-pans, camera lagging slightly behind hyper-speed leaps before snapping back, and lens-grazing debris. 【Timeline】 0–6s: Immediate combat. Monk A’s iron staff whips into frame; Monk B dives from a temple roof. Protagonist low-slides across tile floors under the staff whip, uses a one-hand ground spin to parry B’s daggers, and leaps upward to dodge a low staff sweep. 6–12s: Protagonist step-runs along the vertical surface of a stone pillar, executes a reverse aerial twist-slash at Monk B, lands onto a roof beam, and side-flips through Monk A’s staff arc. 12–18s: Midpoint close-quarters tangle. Trapped between both attackers on a narrow stone bridge. Camera rapid orbital whip. Protagonist drops flat to let their weapons collide, then performs a 360° spinning leg sweep clearing both. 18–24s: Peak aerial agility. Z-pattern wall-rebounds → hilt strikes to weapon joints → mid-air sword deflections → cartwheel dodges through incense smoke. 24–30s: Climax finish. Protagonist slices straight through their combined defense. A rapid triple-strike disarms Monk B, sending him flying into a stone incense burner that shatters on impact. She instantly pivots, leaping into a 360° aerial full-body rotation sword cleave against Monk A, blowing back mist and petals. Final frame: protagonist lands lightly on one foot, sword held diagonally, robes settling in the wind, fast dolly-in to face, hard cut. 【Negative Prompt / Exclusions】 Stance-holding, slow motion, turn-based attacks, idle stances, prolonged eye-locks, magic glowing spell circles, excessive particle effects obscuring action, UI, text, watermarks, subtitles.

Iqra Saifi

12,360 Aufrufe • vor 8 Tagen

I just swapped a UGC creator with an AI character in under 10 minutes 🤯 Same video. Same expressions. Same lip sync. Completely different face. This workflow is perfect for DTC brands & agencies who want to test multiple creator variations without filming anyone new. Here's the use case: You've got a UGC video that's printing money. The script converts. The pacing is dialed. The energy is right. But you want to test different faces—different ages, demographics, vibes—to see what actually drives the highest CTR. The old way: → Hire 5 new creators → Brief each one → Pray they match the original energy → Pay $500+ per video → Wait 2-3 weeks The AI way: → Take your original UGC video → Generate an AI character with Nano Banana → Run both through FAL AI's Wan Animate → AI swaps the face while keeping every expression & movement → Done in minutes The motion tracking is ridiculous. Every smile. Every head tilt. Every hand movement. Perfectly mapped onto the new face. What this unlocks: → One winning script, 5 different creator looks → A/B test demographics without reshooting → Localize ads for different markets instantly → Refresh fatigued creative with a new face I recorded a full step-by-step Loom showing exactly how to do this. Want the tutorial? -> Comment "FACE" -> Like this post And I'll send it over (must be following so I can DM) (Please use responsibly 😂)

I just swapped a UGC creator with an AI character in under 10 minutes 🤯 Same video. Same expressions. Same lip sync. Completely different face. This workflow is perfect for DTC brands & agencies who want to test multiple creator variations without filming anyone new. Here's the use case: You've got a UGC video that's printing money. The script converts. The pacing is dialed. The energy is right. But you want to test different faces—different ages, demographics, vibes—to see what actually drives the highest CTR. The old way: → Hire 5 new creators → Brief each one → Pray they match the original energy → Pay $500+ per video → Wait 2-3 weeks The AI way: → Take your original UGC video → Generate an AI character with Nano Banana → Run both through FAL AI's Wan Animate → AI swaps the face while keeping every expression & movement → Done in minutes The motion tracking is ridiculous. Every smile. Every head tilt. Every hand movement. Perfectly mapped onto the new face. What this unlocks: → One winning script, 5 different creator looks → A/B test demographics without reshooting → Localize ads for different markets instantly → Refresh fatigued creative with a new face I recorded a full step-by-step Loom showing exactly how to do this. Want the tutorial? -> Comment "FACE" -> Like this post And I'll send it over (must be following so I can DM) (Please use responsibly 😂)

Mike Futia

38,862 Aufrufe • vor 9 Monaten

No studio. No mic. No camera. No host. No guest. And yet — I had my podcast in under 5 hours 🎙️ For a while now, I’ve been exploring ways to convert my English videos into Hindi. Thought about re-recording, dubbing, even using an AI avatar — but all of it felt expensive and effort-heavy So I tried this instead: Step 1: NotebookLM has this feature called “audio overview”. It takes any written text and turns it into a podcast-styled voiceover. I simply dropped my Youtube video URL — and within 10 minutes, there was a clean, translated Hindi podcast with dual-voice narration that actually sounds like two people chatting Step 2: Spent 3 hours stitching everything together: a) Recorded a short intro using Tella b) Generated AI visuals of my podcast co-hosts (Pixelcut + ChatGPT) c) Edited & sequenced everything using DaVinci Resolve — manual cuts, trims, timing etc. Step 3: The usual upload drill — thumbnail + title + description + tags etc. Took me another 1.5 hours And in under 5 hours, I had my own PODCAST EPISODE 👏👏👏 I’ll get faster (and better) with time but imagine the possibilities. Now, I can easily convert my earlier videos — not just in Hindi, but even Tamil, Marathi & other regional languages Likewise, if you want to learn about an industry. Just grab 6–7 relevant video URLs or a couple of research reports, drop them into NotebookLM — and boom, you’ve got yourself a podcast in less than 10 minutes. Turn your commute into a mini MBA 👨‍🎓 It’s early days — but this feels like a whole new format PS: I do wish I had that deep RJ-style voice like my co-hosts 😄 .

Shankar Nath

15,005 Aufrufe • vor 1 Jahr

Google just announced the biggest upgrade to Search in over 25 years. For brands the opportunity here is pretty enormous. Here is what the new Search actually looks like and how you should take advantage: The search box now accepts text, images, files, videos, and open Chrome tabs. It expands dynamically as you type. It also anticipates your intent before you finish asking. This is the version of Search that SEO Stuff has been helping customers build for. The biggest opportunity here is what happens after the search. Google's new information agents run 24/7 in the background on behalf of your buyer. And that's why it has never been more important to understand how Google, ChatGPT, Claude and every other AI platform sees your brand. (If you want to see where your site stands across Google and AI search, start here: Here is exactly how Google's new information agents work: Step 1: The buyer does a total brain dump of what they want to stay updated on. Essentially a full description of their problem, their category, their needs. Step 2: The agent breaks down that question and maps out a plan across every relevant sub-topic. Step 3: It determines urgency and what kind of intel the buyer needs right now versus later. Step 4: It sets triggers and monitors the web continuously, scanning blogs, news sites, and social posts for relevant information as it changes. Step 5: It sends the buyer an intelligent synthesized update with links and the ability to take action. Here is why this is a massive opportunity: AI Mode already has 1 billion monthly users. Queries are more than doubling every quarter. And multiple studies have shown that users arriving via AI search are more likely to convert. With information agents running continuously for over a billion users, the brands in that cited source pool are being recommended around the clock, automatically, to buyers who are actively monitoring their category. The brands that build content depth and editorial authority now are building a presence that buildings on itself 24 hours a day. This is what SEO Stuff builds for every customer. Content that covers every sub-question a buyer in your category asks, so the agent finds you at every step of its plan. Authority building from trusted websites that signal credibility to every retrieval system Google has ever built. One investment. Continuous recommendations. Around the clock. Check it out: Our most popular done-for-you package: Our done-for-you "content only" package: There is a reason more than 80 percent of SEO Stuff customers reorder. The results continue long after the work is done. #GoogleIO📷📷📷 #Google📷📷📷

Alex Groberman

157,952 Aufrufe • vor 4 Monaten

"I don't want to look at the bricks around me and ask how high I can stack them. I want to design the most magical, wonderful, mythical castle on a hill and then figure out how to build it." Melanie Perkins (Melanie Perkins) is co-founder and CEO of Canva, one of the hottest private companies in the world—last valued at $42B, generating $3.3B in ARR, profitable for 8 years straight—and is on track to become the most successful female founder in history. But it wasn't always so rosy. She was rejected by over 100 investors, her team had to spend 2 years rewriting their product from scratch (unable to ship anything during that time, something they expected to take 6 months), and they went through a major pivot a few years in. In a rare interview, Melanie and I discuss: 🔸 How “Column B” has been the key to Canva's success 🔸 How Canva survived a painful two-year period without shipping any new features 🔸 How to build a ladder to the moon with small rungs 🔸 Canva’s “two-step plan”: build one of the world’s most valuable companies, then do the most good possible 🔸 So much more Listen now 👇 • YouTube: • Spotify: • Apple: Thank you to our wonderful sponsors for supporting the podcast: 🏆 Vanta — Automate compliance. Simplify security: 🏆 Stripe — Helping companies of all sizes grow revenue: 🏆 Justworks — The all-in-one HR solution for managing your small business with confidence:

"I don't want to look at the bricks around me and ask how high I can stack them. I want to design the most magical, wonderful, mythical castle on a hill and then figure out how to build it." Melanie Perkins (Melanie Perkins) is co-founder and CEO of Canva, one of the hottest private companies in the world—last valued at $42B, generating $3.3B in ARR, profitable for 8 years straight—and is on track to become the most successful female founder in history. But it wasn't always so rosy. She was rejected by over 100 investors, her team had to spend 2 years rewriting their product from scratch (unable to ship anything during that time, something they expected to take 6 months), and they went through a major pivot a few years in. In a rare interview, Melanie and I discuss: 🔸 How “Column B” has been the key to Canva's success 🔸 How Canva survived a painful two-year period without shipping any new features 🔸 How to build a ladder to the moon with small rungs 🔸 Canva’s “two-step plan”: build one of the world’s most valuable companies, then do the most good possible 🔸 So much more Listen now 👇 • YouTube: • Spotify: • Apple: Thank you to our wonderful sponsors for supporting the podcast: 🏆 Vanta — Automate compliance. Simplify security: 🏆 Stripe — Helping companies of all sizes grow revenue: 🏆 Justworks — The all-in-one HR solution for managing your small business with confidence:

Lenny Rachitsky

128,791 Aufrufe • vor 11 Monaten