월간 참여
월간내 콘텐츠의 정기 좌석. 구독자는 대기열에 들어가거나 — 슬롯보다 구독자가 많을 때는 무작위 추첨으로 — 비디오 자체, 크레딧 또는 설명에 나타납니다.
월간
가격 문의
다음
But how do AI images and videos actually work? | Guest video by Welch Labs 38 분 
Hurricane vs. Tiny Houses 23 분 
None an Island 4 분 
LOSST 4 분 
Meet the slow loris, the only venomous primate in existence - Stephanie A. Poindexter 6 분 
iPhone 18 Pro Review - My Last iPhone? 14 분 
Why NASA Is Sending This Truck To The Moon 24 분 
I Thought this would be Safe! 17 분 
This 1950's ATV Concept ? 15 분 
Solar Farms in Space - Reflect Orbital 32 분 
Eating Cheap VS Expensive Food 16 분 
The vulnpocalypse might not be so bad after all 34 분 
Is there a Priest Hole in Your House? | Fascinating Horror Shorts 1 분 
The New Moist Cr1TiKaL 2 분 
How ICE's Surveillance System Works 21 분 
Abandoned - Nara Dreamland 17 분 
I Opened A FAKE Apple Store 18 분 
Anthropic reveals hardware specs and Claude updates, OpenAI talks security, and Runway's new model 35 분 
Mass Magnetics: USA-made magnetics for robotics and defense 3 분 
Man Turns an Empty Space into a Dream Garage | @SmartEasyDIYer 
Practical Magic (1998) - The Exorcism of Nicole Kidman! (10/11) | Movieclips 6 분 
Name That Infrastructure (Ep. 10) #shorts 1 분 
They Call It A "Drowning Machine" | Fascinating Horror Shorts 1 분 
Recreating an Ancient Pump (with no moving parts) 14 분 
Dad Builds 2 DIY Dream MOTORBIKES For His Daughters | Start To Finish 
Why Are There No Short Arch Dams? 17 분 
Do Retention Ponds Actually Work? 18 분 
Why straight dating is harder than ever | The Gray Area 56 분 
Toys R Us 2026 Update 12 분 
What does a parent owe their child? | The Gray Area 27 분 
Neutron stars gave you jewelry. #shorts #science #SciShow 1 분 
Flamethrower Underware! 12 분 ![Home Budgeting with Bobby Berk [Partner content from @progressive]](https://adm.6-9.space/storage/films/posters/yt_home-budgeting-with-bobby-berk-partner-content-from-at-progressive_1789913623.jpg)
Home Budgeting with Bobby Berk [Partner content from @progressive] 5 분 
An Attempt Was Made! 17 분 
We Took Our Food Delivery Man on His First Vacation 46 분 
We’ve just launched *five* new channels! #shorts 1 분 
It’s ok if you zone out during this video 1 분 
What your nipple shape says about you 5 분 
Why Fast Food Got So Expensive 20 분 
The Terrible Design of Newark Airport 17 분 
Is it Possible to Prepare for an Earthquake?: Crash Course Geology #15 13 분 
Name That Infrastructure (Ep. 7) #shorts 1 분 
Name That Infrastructure (Ep. 3) #shorts 1 분 
The magical forest realm hiding at the bottom of the ocean - Luka Seamus Wright and Salomé Buglass 6 분 
AI News: The AI World is REALLY Scared Right Now 36 분 
How To Become Dangerously Self-Educated (with AI) 21 분 
Peppa Pig Tales 2026⏳Peppa Sees BABY Mummy & Daddy Pig! Secret Video❓BRAND NEW Peppa Pig Episodes 31 분 
AI News: The New Model That's As Good As Fable 20 분 
Your Brain Peaked at Birth 2 분 
This AI Found Money I Was Wasting 2 분 
Why avalanches are most deadly when they stop - Simon Trautman 6 분 But how do AI images and videos actually work? | Guest video by Welch Labs
2025 38 분
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
이 크리에이터와 작업하기
이 비디오의 크리에이터가 게시한 패키지.
이 크리에이터는 아직 요금을 설정하지 않았습니다. 요청을 보내주시면 전달하겠습니다 — 무엇이든 청구되기 전에 가격이 합의됩니다.