הבא בתור
But how do AI images and videos actually work? | Guest video by Welch Labs 38 דק׳ 
Digging A SECRET GARAGE Part 2 22 דק׳ 
Humanity’s Real Plan to Stop This Asteroid (Feat. Mark Rober) 22 דק׳ 
NOT TODAY 3 דק׳ 
Something’s Happening to the Ozone Hole 17 דק׳ 
Digging A SECRET GARAGE Part 3 21 דק׳ 
Скользкий Пол | Counter-Strike 2 13 דק׳ 
Are you one of the lucky 50% that experience it? 2 דק׳ 
The $1 Trillion Problem Diamonds Can Solve 17 דק׳ 
WILL THIS DRIFT TRIKE WORK ON GRASS?? 19 דק׳ 
Killed By A View Of Mount Fuji: BOAC Flight 911 | Fascinating Horror 11 דק׳ 
The Grid That Doubles the Strength of the Ground 19 דק׳ 
Building the BATCAVE 1 דק׳ 
About 70% of Trump’s clemency actions did not go through DOJ #shorts 3 דק׳ 
My Thoughts on Wolverine 22 דק׳ 
What a Wild Journey 23 דק׳ 
How the Iran War Spiked Oil Prices 18 דק׳ 
The Largest Biological Attack in US History | Fascinating Horror 11 דק׳ 
The Childhood Game Collection - Scott The Woz 138 דק׳ 
Going In Deep On Data | YC Paper Club 53 דק׳ 
Name That Infrastructure (Ep. 2) #shorts 1 דק׳ 
How Liquid Dampers in Skyscrapers Work 20 דק׳ 
Do Retention Ponds Actually Work? 18 דק׳ 
The Hidden Engineering of Floating Bridges 18 דק׳ 
Man Builds Snow Log Cabin In The Forest Of Finland | Start To Finish By @sharingsurvival 30 דק׳ 
Men Transform Abandoned French House Under $80,000 | Start to Finish by @petitcopeau 
How are holograms possible? 47 דק׳ 
How a crucial battleground candidate is rethinking AI safety 2 דק׳ 
Psychedelics, mind control, radiation-eating: The secret superpowers of fungi - Max G. Levy 7 דק׳ 
Germ Theory Is Older Than You Think 1 דק׳ 
Evermore: The Theme Park That Wasn't 229 דק׳ 
72 Hours Lost in the World’s Richest Dictatorship (no hotel, no food, no transportation...) 38 דק׳ 
Five things you need to know for the 2026 midterms | America, Actually 31 דק׳ 
How do Maglev trains work? - Ning Zhao 7 דק׳ 
Ads in Video Games 20 דק׳ 
For all the hopeless romantics 2 דק׳ 
What Went Wrong With California High Speed Rail 22 דק׳ 
Name That Infrastructure (Ep. 3) #shorts 1 דק׳ 
SciShow is in the field NOW! 1 דק׳ 
Geothermal Energy is Changing | Quaise Energy Documentary 22 דק׳ 
How We Could Reach Proxima Centauri B in 20 Years 18 דק׳ 
It took me 30+ years to realize what I’ll teach you in 10 minutes 13 דק׳ 
Trust the process #shorts 1 דק׳ 
I Tried To Make Something In America (The Smarter Scrubber Experiment) - Smarter Every Day 308 49 דק׳ 
What toys have kids played with throughout history? 6 דק׳ 
Jumpstart by Phone 14 דק׳ 
The Stolen Recipe | Detective Cartoon | Educational | Sheriff Labrador | Kids Cartoon | BabyBus 32 דק׳ 
DEFCON is Not What I Expected... 16 דק׳ 
Peppa Pig Tales 2026 🤫 Peppa Finds a SECRET Shop Passageway 🔒 BRAND NEW Peppa Pig Episodes 31 דק׳ 
Peppa Pig WINS the DUCK RACE for Baby EVIE! 🦆🏆 | BRAND NEW Peppa Pig Full Episodes | 4 Hours 243 דק׳ 
These Rocks Are Older Than the Sun 13 דק׳ 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
Packages published by the creator behind this video.
This creator has not set their rates yet. Send a request and we will pass it on — the price is agreed before anything is charged.
A recurring seat in my content. Subscribers enter a queue — or a random draw when there are more subscribers than slots — and appear in the video itself, in the credits, or in the description.
per month
Price on request
A guaranteed appearance of at least 10 seconds in my next video. Price depends on where in the runtime the insert lands.
Price on request
A guaranteed appearance of at least 10 seconds inside a Short cut from one of my existing videos.
one-off
Price on request