Neste
But how do AI images and videos actually work? | Guest video by Welch Labs 38 min 
BYE 3 min 
Glass is glass 5 min 
Inside the Secret Country Erased from Maps (what are they hiding?) 29 min 
Digging A SECRET GARAGE Part 1 25 min 
I Escaped the Apple Ecosystem - It Backfired 17 min 
Mesa 4 min 
The Los Angeles Aqueduct is Wild 23 min 
The Insane Engineering of the V-22 Osprey 24 min 
24 Hours in 2 Countries that Hate Each Other 49 min 
Plasma Globe is a ⚡Shocking⚡🔥Fire Hazard🔥 12 min 
Fastest Scam Ever 10 min 
Fixing the Most Dangerous Dam in the World 23 min 
How Disney Built A Realistic Lightsaber 17 min 
Why Fast Food Got So Expensive 20 min 
$100 vs $100,000 Holiday: Asia Edition 63 min 
Actually a Wild Situation 24 min 
How 60 Minutes got inside the world of pardon brokers #shorts 2 min 
The Story of Bill Stephens | Fascinating Horror Shorts 1 min 
How to undo the damage of sitting all day | The Gray Area 42 min 
Name That Infrastructure (Ep. 14) #shorts 1 min 
Man Builds Snow Log Cabin In The Forest Of Finland | Start To Finish By @sharingsurvival 30 min 
Are Beavers Actually Good at Dams? 20 min 
California’s Tallest Bridge Has Nothing Underneath 18 min 
Man Builds DIY Off-Grid WOODEN HOUSE In The Mountains | Start To Finish By @CountryBoy-82 
Name That Infrastructure (Ep. 13) #shorts 1 min 
Practical Magic (1998) - Sisterly Witchy Mischief (4/11) | Movieclips 2 min 
Abandoned - Fort Rapids Indoor Water Park 19 min 
The missing ingredient in how we learn 6 min 
Lindsey Clancy and the surprisingly simple solution to helping moms with postpartum depression 2 min 
Why even the best aquariums can’t give animals what they need 2 min 
Jony Ive Shows The Real Process Behind the Ferrari Luce 46 min 
The New Hunt for the Colossal Squid 19 min 
How to remove AI from your life 3 min 
Edgar Allan Poe's story of the Red Death - Iseult Gillespie 8 min 
Peer Review and the Quest for Truth: Crash Course Scientific Thinking #4 11 min 
How Car Dealerships Scam America 18 min 
What’s at the center of the earth?: Crash Course Geology #4 10 min 
Janus Is Two-Faced #scishow #science #space 1 min 
The Weird Reason Tornadoes Are Clumping Together #science #scishow #stem #shorts 1 min 
Those Ancient Giant Fungi Weren’t Fungi After All 15 min 
They’re lying to you 2 min 
Spain’s temperatures are HIGH rn 🔥 1 min 
THE WAREHOUSE #PeppaPig #Shorts 1 min 
SciShow is heading into the field TOMORROW with the all new Field Trips mini series. 1 min 
Peppa Pig Learns about the TOOTH FAIRY! 🧚🦷 | Losing a Tooth! | Peppa Pig Official Full Episodes 45 min 
Peppa Pig Tales 2026 🫣 Peppa Plays CAMOUFLAGE HIDE AND SEEK! 👀 BRAND NEW Peppa Pig Episodes 31 min 
What happens when we coat a pingpong ball with metal? #tbt #hacksmith #makeitreal 2 min 
Hybrid Power Systems! | Real Life Power Armor (Part 4/6) 23 min 
Take a Bath, Dirty Monster! | Good Habits Song | Bath Song | Nursery Rhymes | BabyBus 24 min 
Should you rinse your dishes before placing them in the dishwasher? - Rachel Yang 6 min Diffusion models, CLIP, and the math of turning text into images Welch Labs Book: https://www.welchlabs.com/resources/imaginary-numbers-book Sections 0:00 - Intro 3:37 - CLIP 6:25 - Shared Embedding Space 8:16 - Diffusion Models & DDPM 11:44 - Learning Vector Fields 22:00 - DDIM 25:25 - Dall E 2 26:37 - Conditioning 30:02 - Guidance 33:39 - Negative Prompts 34:27 - Outro 35:32 - About guest videos Special Thanks to: Jonathan Ho - Jonathan is the Author of the DDPM paper and the Classifier Free Guidance Paper. https://arxiv.org/pdf/2006.11239 https://arxiv.org/pdf/2207.12598 Preetum Nakkiran - Preetum has an excellent introductory diffusion tutorial: https://arxiv.org/pdf/2406.08929 Chenyang Yuan - Many of the animations in this video were implemented using manim and Chenyang’s smalldiffusion library: https://github.com/yuanchenyang/smalldiffusion Cheyang also has a terrific tutorial and MIT course on diffusion models https://www.chenyang.co/diffusion.html https://www.practical-diffusion.org/ Other References All of Sander Dieleman’s diffusion blog posts are fantastic: https://sander.ai/ CLIP Paper: https://arxiv.org/pdf/2103.00020 DDIM Paper: https://arxiv.org/pdf/2010.02502 Score-Based Generative Modeling: https://arxiv.org/pdf/2011.13456 Wan2.1: https://github.com/Wan-Video/Wan2.1 Stable Diffusion: https://huggingface.co/stabilityai/stable-diffusion-2 Midjourney: https://www.midjourney.com/ Veo: https://deepmind.google/models/veo/ DallE 2 paper: https://cdn.openai.com/papers/dall-e-2.pdf Code for this video: https://github.com/stephencwelch/manim_videos/tree/master/_2025/sora Written by: Stephen Welch, with very helpful feedback from Grant Sanderson Produced by: Stephen Welch, Sam Baskin, and Pranav Gundu Technical Notes The noise videos in the opening have been passed through a VAE (actually, diffusion process happens in a compressed “latent” space), which acts very much like a video compressor - this is why the noise videos don’t look like pure salt and pepper. 6:15 CLIP: Although directly minimizing cosine similarity would push our vectors 180 degrees apart on a single batch, overall in practice, we need CLIP to maximize the uniformity of concepts over the hypersphere it's operating on. For this reason, we animated these vectors as orthogonal-ish. See: https://proceedings.mlr.press/v119/wang20k/wang20k.pdf Per Chenyang Yuan: at 10:15, the blurry image that results when removing random noise in DDPM is probably due to a mismatch in noise levels when calling the denoiser. When the denoiser is called on x_{t-1} during DDPM sampling, it is expected to have a certain noise level (let's call it sigma_{t-1}). If you generate x_{t-1} from x_t without adding noise, then the noise present in x_{t-1} is always smaller than sigma_{t-1}. This causes the denoiser to remove too much noise, thus pointing towards the mean of the dataset. The text conditioning input to stable diffusion is not the 512-dim text embedding vector, but the output of the layer before that, [with dimension 77x512](https://stackoverflow.com/a/79243065) For the vectors at 31:40 - Some implementations use f(x, t, cat) + alpha(f(x, t, cat) - f(x, t)), and some that do f(x, t) + alpha(f(x, t, cat) - f(x, t)), where an alpha value of 1 corresponds to no guidance. I chose the second format here to keep things simpler. At 30:30, the unconditional t=1 vector field looks a bit different from what it did at the 17:15 mark. This is the result of different models trained for different parts of the video, and likely a result of different random initializations. Premium Beat Music ID: EEDYZ3FP44YX8OWT
Pakker publisert av skaperen bak denne videoen.
Denne skaperen har ikke satt prisene sine ennå. Send en forespørsel, og vi videreformidler den — prisen avtales før noe belastes.
En tilbakevendende plass i innholdet mitt. Abonnenter går inn i en kø — eller en tilfeldig trekning når det er flere abonnenter enn plasser — og vises i selve videoen, i krediteringene eller i beskrivelsen.
per måned
Pris på forespørsel
En garantert opptreden på minst 10 sekunder i min neste video. Prisen avhenger av hvor i spilletiden innslaget havner.
Pris på forespørsel
En garantert opptreden på minst 10 sekunder i en Short klippet fra en av mine eksisterende videoer.
engangsbeløp
Pris på forespørsel