Загрузка видео...

Не удалось загрузить видео

На главную

RIFLEx A Free Lunch for Length Extrapolation in Video Diffusion Transformers TL;DR: Effortlessly extend your video with just one line of code: freq[k-1]=(2*np.pi)/(L*s).

_akhaliq's profile picture

AK

504,953 subscribers

18,283 просмотров • 1 год назад •via X (Twitter)

ОбразованиеНаука и технологии
LIVE

Anya Rossi• Live Now

Private livecam show

Комментарии: 0

Нет доступных комментариев

Здесь появятся комментарии из оригинального поста

Похожие видео

DimensionX: Create Any 3D and 4D Scenes from a Single Image with Controllable Video Diffusion TL;DR: Create 3/4DGS from Video Diffusion Note: Some first inference code released (not all yet). Contributions (cited): • We present DimensionX, a novel framework for generating photorealistic 3D and 4D scenes from only a single image using controllable video diffusion. • We propose ST-Director, which decouples the spatial and temporal priors in video diffusion models by learning (spatial and temporal) dimension-aware modules with our curated datasets. We further enhance the hybriddimension control with a training-free composition approach according to the essence of video diffusion denoising process. • To bridge the gap between video diffusion and real-world scenes, we design a trajectory-aware mechanism for 3D generation and an identity-preserving denoising approach for 4D generation, enabling more realistic and controllable scene synthesis. • Extensive experiments manifest that our DimensionX delivers superior performance in video, 3D, and 4D generation compared with baseline methods.
0:18

DimensionX: Create Any 3D and 4D Scenes from a Single Image with Controllable Video Diffusion TL;DR: Create 3/4DGS from Video Diffusion Note: Some first inference code released (not all yet). Contributions (cited): • We present DimensionX, a novel framework for generating photorealistic 3D and 4D scenes from only a single image using controllable video diffusion. • We propose ST-Director, which decouples the spatial and temporal priors in video diffusion models by learning (spatial and temporal) dimension-aware modules with our curated datasets. We further enhance the hybriddimension control with a training-free composition approach according to the essence of video diffusion denoising process. • To bridge the gap between video diffusion and real-world scenes, we design a trajectory-aware mechanism for 3D generation and an identity-preserving denoising approach for 4D generation, enabling more realistic and controllable scene synthesis. • Extensive experiments manifest that our DimensionX delivers superior performance in video, 3D, and 4D generation compared with baseline methods.

MrNeRF

17,037 просмотров • 1 год назад

Abracadabra 🪄 #define t (t+fract(1e4*sin(dot(FC.xy,vec2(137,-13))))/6e1) #define S smoothstep float i,d,s,m,l,x=abs(mod(t/4.,2.)-1.),a=x 6.)break;l=length(.2*k.xy-vec2(sin(t)/9.,.6+sin(t+t)/9.));k.y<-5.?k.y=-k.y-10.,m=.5:m=1.;k.xz*=mat2(cos(a*6.28+k.y*.3*S(.2,.5,x)*S(.7,.5,x)+vec4(0,33,11,0)));for(p=k*.5,s=.01;s<1.;s+=s)p.y+=.95+abs(dot(sin(p.x+2.*t+p/s),.2+p-p))*s;l=mix(sin(length(k*k.x)),mix(sin(length(p)),l,.5-l),S(5.5,6.,p.y));p=abs(k);d+=s=.012+.09*abs(max(sin(length(k)+l),max(max(max(p.x,p.y),p.z),dot(p,vec3(.577))*mix(.5,.9,a))-3.)-i/1e2);}o=tanh(o*o/1e6)*m;
0:15

Abracadabra 🪄 #define t (t+fract(1e4*sin(dot(FC.xy,vec2(137,-13))))/6e1) #define S smoothstep float i,d,s,m,l,x=abs(mod(t/4.,2.)-1.),a=x 6.)break;l=length(.2*k.xy-vec2(sin(t)/9.,.6+sin(t+t)/9.));k.y<-5.?k.y=-k.y-10.,m=.5:m=1.;k.xz*=mat2(cos(a*6.28+k.y*.3*S(.2,.5,x)*S(.7,.5,x)+vec4(0,33,11,0)));for(p=k*.5,s=.01;s<1.;s+=s)p.y+=.95+abs(dot(sin(p.x+2.*t+p/s),.2+p-p))*s;l=mix(sin(length(k*k.x)),mix(sin(length(p)),l,.5-l),S(5.5,6.,p.y));p=abs(k);d+=s=.012+.09*abs(max(sin(length(k)+l),max(max(max(p.x,p.y),p.z),dot(p,vec3(.577))*mix(.5,.9,a))-3.)-i/1e2);}o=tanh(o*o/1e6)*m;

Jae

28,112 просмотров • 6 месяцев назад

NVIDIA just released a very impressive text-to-video paper. Video Latent Diffusion Models (Video LDMs) use a diffusion model in a compressed latent space to generate high-resolution videos. Here's a brief overview of how it works: 1. Pre-train image LDM on a dataset of images. 2. Turn the image LDM into a Video LDM by adding temporal layers to model video frames. 3. Fine-tune the Video LDM on encoded video sequences to create a video generator. 4. Temporally align diffusion model upsamplers to generate high-resolution videos. 5. Validate Video LDM on real driving videos of 512x1024 resolution, achieving state-of-the-art performance. 6. Apply the approach in creative content creation with text-to-video modeling. Paper: Project:
0:14

NVIDIA just released a very impressive text-to-video paper. Video Latent Diffusion Models (Video LDMs) use a diffusion model in a compressed latent space to generate high-resolution videos. Here's a brief overview of how it works: 1. Pre-train image LDM on a dataset of images. 2. Turn the image LDM into a Video LDM by adding temporal layers to model video frames. 3. Fine-tune the Video LDM on encoded video sequences to create a video generator. 4. Temporally align diffusion model upsamplers to generate high-resolution videos. 5. Validate Video LDM on real driving videos of 512x1024 resolution, achieving state-of-the-art performance. 6. Apply the approach in creative content creation with text-to-video modeling. Paper: Project:

Lior Alexander

158,545 просмотров • 3 лет назад

Your chance to win a free golf swing analysis from one of our world class players! Click the video to sign up for free and be in with a chance!
0:19

Your chance to win a free golf swing analysis from one of our world class players! Click the video to sign up for free and be in with a chance!

Majesticks Golf Club

41,388 просмотров • 1 год назад

Preview of the new free full length oily soles video I just uploaded 💦
0:10

Preview of the new free full length oily soles video I just uploaded 💦

Greek Goddess Li 24k 🔥

21,003 просмотров • 1 год назад

Blessing your TL with this video of Beyoncé 🙏🏾❤️
0:12

Blessing your TL with this video of Beyoncé 🙏🏾❤️

THUGGA

42,113 просмотров • 1 год назад

🚨Announcing our #ICLR2025 Oral! 🔥Diffusion LMs are on the rise for parallel text generation! But unlike autoregressive LMs, they struggle with quality, fixed-length constraints & lack of KV caching. 🚀Introducing Block Diffusion—combining autoregressive and diffusion models for the best of both worlds! 👇1/7
0:17

🚨Announcing our #ICLR2025 Oral! 🔥Diffusion LMs are on the rise for parallel text generation! But unlike autoregressive LMs, they struggle with quality, fixed-length constraints & lack of KV caching. 🚀Introducing Block Diffusion—combining autoregressive and diffusion models for the best of both worlds! 👇1/7

Marianne Arriola

148,599 просмотров • 1 год назад

Introducing Expect Let agents test your code in a real browser 1. Run Claude Code / Codex to QA your app 2. Watch a video of every bug found 3. Fix and repeat until passing Run as a CLI or agent skill. Fully open source
0:16

Introducing Expect Let agents test your code in a real browser 1. Run Claude Code / Codex to QA your app 2. Watch a video of every bug found 3. Fix and repeat until passing Run as a CLI or agent skill. Fully open source

Aiden Bai

841,176 просмотров • 3 месяцев назад

Stormy Torus [480] float i,d,w,s,n,m=1.;for(mat2 R=mat2(cos(sin(t/2.)*.785+vec4(0,33,11,0)));i++<1e2;){vec3 k,p=vec3((FC.xy*2.-r.xy)/r.y*d,d-10.);p.xz*=R;if(p.y<-6.3){p.y =-p.y-9.;m=.5;}k=p;for(p*=.5,n=.01;n<.2;n+=n)p.yz+=cos(p.xy*.01)-abs(dot(sin(.02*p.z+.03*p.y+2.*t+.3*p/n),p-p+n));s=length(k.xy)-4.;d+=w=.01+.07*abs(max(w=mix(sin(length(ceil(k*4.).z+k)),sin(length(p)-1.),smoothstep(5.,5.5,p.y)),sqrt(s*s+k*k).z-1.5)-i/150.);o+=max(sin(vec4(1,2,3,1)+i*.5)*1.3/w,-length(k*k));}o=tanh(o*o/1e6)*m; #glsl
0:19

Stormy Torus [480] float i,d,w,s,n,m=1.;for(mat2 R=mat2(cos(sin(t/2.)*.785+vec4(0,33,11,0)));i++<1e2;){vec3 k,p=vec3((FC.xy*2.-r.xy)/r.y*d,d-10.);p.xz*=R;if(p.y<-6.3){p.y =-p.y-9.;m=.5;}k=p;for(p*=.5,n=.01;n<.2;n+=n)p.yz+=cos(p.xy*.01)-abs(dot(sin(.02*p.z+.03*p.y+2.*t+.3*p/n),p-p+n));s=length(k.xy)-4.;d+=w=.01+.07*abs(max(w=mix(sin(length(ceil(k*4.).z+k)),sin(length(p)-1.),smoothstep(5.,5.5,p.y)),sqrt(s*s+k*k).z-1.5)-i/150.);o+=max(sin(vec4(1,2,3,1)+i*.5)*1.3/w,-length(k*k));}o=tanh(o*o/1e6)*m; #glsl

Jae

13,287 просмотров • 7 месяцев назад

Next.js 15.1: Easier error debugging ✓ Hide¹ stack frames for dependency code ✓ Surface source of the error in your code Video: Terminal output² of a Zod parsing error
0:10

Next.js 15.1: Easier error debugging ✓ Hide¹ stack frames for dependency code ✓ Surface source of the error in your code Video: Terminal output² of a Zod parsing error

Delba

49,737 просмотров • 1 год назад

MaskINT: Video Editing via Interpolative Non-autoregressive Masked Transformers paper page: Recent advances in generative AI have significantly enhanced image and video editing, particularly in the context of text prompt control. State-of-the-art approaches predominantly rely on diffusion models to accomplish these tasks. However, the computational demands of diffusion-based methods are substantial, often necessitating large-scale paired datasets for training, and therefore challenging the deployment in practical applications. This study addresses this challenge by breaking down the text-based video editing process into two separate stages. In the first stage, we leverage an existing text-to-image diffusion model to simultaneously edit a few keyframes without additional fine-tuning. In the second stage, we introduce an efficient model called MaskINT, which is built on non-autoregressive masked generative transformers and specializes in frame interpolation between the keyframes, benefiting from structural guidance provided by intermediate frames. Our comprehensive set of experiments illustrates the efficacy and efficiency of MaskINT when compared to other diffusion-based methodologies. This research offers a practical solution for text-based video editing and showcases the potential of non-autoregressive masked generative transformers in this domain.
0:19

MaskINT: Video Editing via Interpolative Non-autoregressive Masked Transformers paper page: Recent advances in generative AI have significantly enhanced image and video editing, particularly in the context of text prompt control. State-of-the-art approaches predominantly rely on diffusion models to accomplish these tasks. However, the computational demands of diffusion-based methods are substantial, often necessitating large-scale paired datasets for training, and therefore challenging the deployment in practical applications. This study addresses this challenge by breaking down the text-based video editing process into two separate stages. In the first stage, we leverage an existing text-to-image diffusion model to simultaneously edit a few keyframes without additional fine-tuning. In the second stage, we introduce an efficient model called MaskINT, which is built on non-autoregressive masked generative transformers and specializes in frame interpolation between the keyframes, benefiting from structural guidance provided by intermediate frames. Our comprehensive set of experiments illustrates the efficacy and efficiency of MaskINT when compared to other diffusion-based methodologies. This research offers a practical solution for text-based video editing and showcases the potential of non-autoregressive masked generative transformers in this domain.

AK

25,449 просмотров • 2 лет назад

‘Oh s**t… what are we gonna do now?’ 🤖🔥 For a lot of kids in 1986, that moment in ‘Transformers: The Movie’ was genuinely shocking. Spike Witwicky dropped a line you’d *never* hear on the TV show. It was definitely one of those ‘did he really just say that?!’ 😳 moments
0:18

‘Oh s**t… what are we gonna do now?’ 🤖🔥 For a lot of kids in 1986, that moment in ‘Transformers: The Movie’ was genuinely shocking. Spike Witwicky dropped a line you’d *never* hear on the TV show. It was definitely one of those ‘did he really just say that?!’ 😳 moments

RetroNewsNow

260,221 просмотров • 1 месяц назад

💾 How to Download ANY X Video in Seconds: #1 Follow me X Downloader Bot #2 Tag me in a tweet with the video #3 Get your download link instantly! ⚡ Just reply with any video and I'll hook you up Works with normal videos, gifs & more! No limits, Super fast & Free 👇 Try it now - tag me below!
0:10

💾 How to Download ANY X Video in Seconds: #1 Follow me X Downloader Bot #2 Tag me in a tweet with the video #3 Get your download link instantly! ⚡ Just reply with any video and I'll hook you up Works with normal videos, gifs & more! No limits, Super fast & Free 👇 Try it now - tag me below!

X Downloader Bot

348,006 просмотров • 14 дней назад

I've created a utility for debugging React performance. After spending a few hours trying to connect react-devtools with Electron, I decided to write my own from scratch. You can add it to your project with just one line of code -
0:11

I've created a utility for debugging React performance. After spending a few hours trying to connect react-devtools with Electron, I decided to write my own from scratch. You can add it to your project with just one line of code -

Adam Pietrasiak

24,083 просмотров • 2 месяцев назад

Is this your fav Video of me? Spicy link in the comment section ;) subscribe for a free pic! Followers CANNOT chat with me
0:15

Sensitive content

Is this your fav Video of me? Spicy link in the comment section ;) subscribe for a free pic! Followers CANNOT chat with me

Vina

15,281 просмотров • 1 год назад

6 APIs for your next project 1. Fine-tune LLMs without writing a single line of code 2. Pokemon API 3. Memes API 4. Dogs API 5. Bored API
0:19

6 APIs for your next project 1. Fine-tune LLMs without writing a single line of code 2. Pokemon API 3. Memes API 4. Dogs API 5. Bored API

Csaba Kissi

54,072 просмотров • 2 лет назад

window seat reflection removal Reflection Removal through Efficient Adaptation of Diffusion Transformers
0:15

window seat reflection removal Reflection Removal through Efficient Adaptation of Diffusion Transformers

AK

18,839 просмотров • 6 месяцев назад

We are now offering all new users 100 credits per month free of charge to try Flow (the equivalent of 1 Veo 3 Quality video or 5 Veo 3 Fast videos). Go beyond the single clip — trim and extend your scenes (with consistency!) in today.
0:17

We are now offering all new users 100 credits per month free of charge to try Flow (the equivalent of 1 Veo 3 Quality video or 5 Veo 3 Fast videos). Go beyond the single clip — trim and extend your scenes (with consistency!) in today.

FlowbyGoogle

94,713 просмотров • 10 месяцев назад

Enjoy this video of a golden retriever puppy effortlessly falling asleep in seconds
0:15

Enjoy this video of a golden retriever puppy effortlessly falling asleep in seconds

Puppies 🐶

60,779 просмотров • 9 месяцев назад