正在加载视频...

视频加载失败

加载此视频时出现问题。这可能是由于临时网络问题，或视频可能不可用。

Stable Diffusion 100% in the browser with running at < 1s per image ⚡️ Thanks to SD Turbo by Axel Sauer & al Stability AI and diffusers.js on webGPU by Arthur Islamov Demo: More info in ↓🧵

Cyril Diagne

25,027 subscribers

96,629 次观看 • 2 年前 •via X (Twitter)

科学技术教育

Anya Rossi• Live Now

Private livecam show

0 条评论

暂无评论

原始帖子的评论将显示在这里

相关视频

Meta's Segment Anything Model (SAM) can now run in your browser w/ WebGPU (+ fp16), meaning up to 8x faster image encoding (10s → 1.25s)! 🤯⚡️ Video is not sped up! Everything runs 100% locally thanks to 🤗 Transformers.js and onnxruntime-web! 🔗 Demo:

Meta's Segment Anything Model (SAM) can now run in your browser w/ WebGPU (+ fp16), meaning up to 8x faster image encoding (10s → 1.25s)! 🤯⚡️ Video is not sped up! Everything runs 100% locally thanks to 🤗 Transformers.js and onnxruntime-web! 🔗 Demo:

Xenova

120,352 次观看 • 2 年前

Today, we are adding Stable Video Diffusion, our foundation model for generative video to the Stability AI Developer Platform API. The model can generate 2 seconds of video, comprising of 25 generated frames and 24 frames of FILM interpolation, within an average time of 41 seconds. Developers interested in utilizing Stable Video Diffusion through an API can access it now on the Stability AI Developer Platform. Learn more here:

Today, we are adding Stable Video Diffusion, our foundation model for generative video to the Stability AI Developer Platform API. The model can generate 2 seconds of video, comprising of 25 generated frames and 24 frames of FILM interpolation, within an average time of 41 seconds. Developers interested in utilizing Stable Video Diffusion through an API can access it now on the Stability AI Developer Platform. Learn more here:

Stability AI

175,837 次观看 • 2 年前

ENHANCE! - Infinitely zoom into any image. It uses what I call "creative upscaling" with #nanobanana to shape pixels into something plausible. Now featured in the Google AI Studio showcase: All gifs are rendered by the app and not "real-time" More in 🧵

ENHANCE! - Infinitely zoom into any image. It uses what I call "creative upscaling" with #nanobanana to shape pixels into something plausible. Now featured in the Google AI Studio showcase: All gifs are rendered by the app and not "real-time" More in 🧵

Rikard Lindström

10,860 次观看 • 11 个月前

The Hidden Language of Diffusion Models paper page: tackle the challenge of understanding concept representations in text-to-image models by decomposing an input text prompt into a small set of interpretable elements. This is achieved by learning a pseudo-token that is a sparse weighted combination of tokens from the model's vocabulary, with the objective of reconstructing the images generated for the given concept. Applied over the state-of-the-art Stable Diffusion model, this decomposition reveals non-trivial and surprising structures in the representations of concepts. For example, we find that some concepts such as "a president" or "a composer" are dominated by specific instances (e.g., "Obama", "Biden") and their interpolations. Other concepts, such as "happiness" combine associated terms that can be concrete ("family", "laughter") or abstract ("friendship", "emotion"). In addition to peering into the inner workings of Stable Diffusion, our method also enables applications such as single-image decomposition to tokens, bias detection and mitigation, and semantic image manipulation

The Hidden Language of Diffusion Models paper page: tackle the challenge of understanding concept representations in text-to-image models by decomposing an input text prompt into a small set of interpretable elements. This is achieved by learning a pseudo-token that is a sparse weighted combination of tokens from the model's vocabulary, with the objective of reconstructing the images generated for the given concept. Applied over the state-of-the-art Stable Diffusion model, this decomposition reveals non-trivial and surprising structures in the representations of concepts. For example, we find that some concepts such as "a president" or "a composer" are dominated by specific instances (e.g., "Obama", "Biden") and their interpolations. Other concepts, such as "happiness" combine associated terms that can be concrete ("family", "laughter") or abstract ("friendship", "emotion"). In addition to peering into the inner workings of Stable Diffusion, our method also enables applications such as single-image decomposition to tokens, bias detection and mitigation, and semantic image manipulation

AK

41,830 次观看 • 3 年前

🧵AI scene exploration: Midjourney x Gen-3 Alpha Turbo guide One of my favorite things to do with AI, is to use Advanced Camera Controls in Gen-3 Alpha Turbo by Runway It lets me see my interiors, cars and portraits from different angles, so it feels like a cool prelude before AI 3D, that we'll hopefully see at some point. Let me show you how to do it step-by step 👇

🧵AI scene exploration: Midjourney x Gen-3 Alpha Turbo guide One of my favorite things to do with AI, is to use Advanced Camera Controls in Gen-3 Alpha Turbo by Runway It lets me see my interiors, cars and portraits from different angles, so it feels like a cool prelude before AI 3D, that we'll hopefully see at some point. Let me show you how to do it step-by step 👇

Everett World

68,655 次观看 • 1 年前

HTML enters 3D! Or vice versa? With the new HTML in Canvas by WICG, we can finally put native DOM elements directly into WebGL/WebGPU scenes. It is experimental for now, but the possibilities for 3D interfaces and special effects are huge. This demo was built using Three.js and Omma AI (tool by Spline ) It’s a fun new way to explore what the web can do! Are you interested in seeing the demo?

HTML enters 3D! Or vice versa? With the new HTML in Canvas by WICG, we can finally put native DOM elements directly into WebGL/WebGPU scenes. It is experimental for now, but the possibilities for 3D interfaces and special effects are huge. This demo was built using Three.js and Omma AI (tool by Spline ) It’s a fun new way to explore what the web can do! Are you interested in seeing the demo?

Gábor Pribék

176,349 次观看 • 4 个月前

🧵1/5 I’m fucking around with 6529, let’s find out what it means to seize the memes of production. I will be minting this rememe of meme card #1 from @6529collections by 6529er on melon.ooo 🍉 Price: FREE Supply: 100 editions Allowlist and more info in thread

🧵1/5 I’m fucking around with 6529, let’s find out what it means to seize the memes of production. I will be minting this rememe of meme card #1 from @6529collections by 6529er on melon.ooo 🍉 Price: FREE Supply: 100 editions Allowlist and more info in thread

Roger Skaer

44,203 次观看 • 3 年前

Apple wasn’t kidding, the iPhone 17 Pro is really built for running LLMs Here’s LFM2 8B A1B by running on-device with MLX in Locally AI - Local AI Chat, the iPhone runs the 8B model with zero struggle Thanks Prince Canuma for the port to MLX, it made the MLX Swift port possible

Apple wasn’t kidding, the iPhone 17 Pro is really built for running LLMs Here’s LFM2 8B A1B by running on-device with MLX in Locally AI - Local AI Chat, the iPhone runs the 8B model with zero struggle Thanks Prince Canuma for the port to MLX, it made the MLX Swift port possible

Adrien Grondin

493,561 次观看 • 10 个月前

A GPT-4 level chatbot, available to use completely free, running at over 800 tokens per second on Groq. I'm genuinely mindblown by LlaMA 3. Try it with the link in the next tweet.

A GPT-4 level chatbot, available to use completely free, running at over 800 tokens per second on Groq. I'm genuinely mindblown by LlaMA 3. Try it with the link in the next tweet.

Rowan Cheung

1,242,145 次观看 • 2 年前

Along with @levelsio, we’re looking for the best games you make with AI. More in his tweet below. Cursor 3 has a built-in browser, image generation, and cloud agents that can playtest, while Composer 2 is fast with Three.js. Great for building games! Even developers at Rockstar are using AI now. Games can be built faster, made more beautiful and complex, and created by more people. It's surprisingly easy if you have no experience. The games shown below were already made with Cursor by Startracker 🔺, Danny Limanseta, David Bui, and Chong-U. Show us what you can build! Can't wait to see your submissions.

Along with @levelsio, we’re looking for the best games you make with AI. More in his tweet below. Cursor 3 has a built-in browser, image generation, and cloud agents that can playtest, while Composer 2 is fast with Three.js. Great for building games! Even developers at Rockstar are using AI now. Games can be built faster, made more beautiful and complex, and created by more people. It's surprisingly easy if you have no experience. The games shown below were already made with Cursor by Startracker 🔺, Danny Limanseta, David Bui, and Chong-U. Show us what you can build! Can't wait to see your submissions.

edwin

78,395 次观看 • 4 个月前

🚀Major update to Blender MCP: 🌄 Generate high-quality 3D assets in Blender by just prompting, thanks to Hyper3D by Deemos's Rodin AI. 🎬 Demo: Creating a cozy scene with a cat. Just describe the assets you want, and it will show up right in Blender. No manual modeling needed!

🚀Major update to Blender MCP: 🌄 Generate high-quality 3D assets in Blender by just prompting, thanks to Hyper3D by Deemos's Rodin AI. 🎬 Demo: Creating a cozy scene with a cat. Just describe the assets you want, and it will show up right in Blender. No manual modeling needed!

siddharth ahuja

109,971 次观看 • 1 年前

Seedance 2.0 - Test Demo 04 I used one image I generated back in 2024 by MJ V5 to Seedance S2.0. This image style is mixed with 80s animation and 70s illustration. Wanna check how Seedance 2.0 can handle the mocap for this style? And here is the result. #seedance2 Dreamina AI Midjourney

Seedance 2.0 - Test Demo 04 I used one image I generated back in 2024 by MJ V5 to Seedance S2.0. This image style is mixed with 80s animation and 70s illustration. Wanna check how Seedance 2.0 can handle the mocap for this style? And here is the result. #seedance2 Dreamina AI Midjourney

Jim HuiHui

32,394 次观看 • 6 个月前

This AI Agent generates 1000s of winning static ads automatically. → Auto-generates 100+ unique image variations → AI writes custom prompts for each ad angle → OpenAI Image creates on-brand visuals in seconds → Links each image to specific ad copy automatically → Bulk generation with one checkbox → Editable prompts before image creation → Hosted images ready for immediate upload → Built in n8n + Airtable (zero coding) Perfect for ecomm operators, creative strategists, and Meta ad agencies. The workflow: 1. Drop in product URL 2. AI researches brand & audience 3. Generates 100+ ad copy variations 4. Creates 100+ custom image prompts 5. Produces 100+ static ad images All in under 5 minutes. >> No more creative bottlenecks. >> No more waiting on designers. Just static ads, at scale, on demand. Want access to the n8n template + Airtable database? Like + Comment "ADS" and I'll send it over.

Mike Futia

33,001 次观看 • 1 年前

Introducing Muse Image and Muse Video, the first media generation models developed by Meta Superintelligence Labs. Muse Image is our most advanced image generation model yet. It follows instructions faithfully, edits with precision, composes from multiple references, and draws on Instagram for social context. It also brings agentic tool use capabilities to image generation and integrates with Muse Spark. You can try Muse Image in the Meta AI app and web, as well as in Instagram Stories and WhatsApp – starting in limited countries with more locations on the way. Today we’re also previewing Muse Video, which is built upon the same pretraining base as Muse Image to deliver exceptional visual fidelity with native audio support. Learn more about both models:

Introducing Muse Image and Muse Video, the first media generation models developed by Meta Superintelligence Labs. Muse Image is our most advanced image generation model yet. It follows instructions faithfully, edits with precision, composes from multiple references, and draws on Instagram for social context. It also brings agentic tool use capabilities to image generation and integrates with Muse Spark. You can try Muse Image in the Meta AI app and web, as well as in Instagram Stories and WhatsApp – starting in limited countries with more locations on the way. Today we’re also previewing Muse Video, which is built upon the same pretraining base as Muse Image to deliver exceptional visual fidelity with native audio support. Learn more about both models:

AI at Meta

844,718 次观看 • 1 个月前

Lattice light-sheet #microscopy with 4D image analysis reveals #Bcell receptors #immunoglobulin localise into clusters to aid recognition of their #antigen. Video & research by Deniz Saltukoglu et al BIOSS Freiburg in The EMBO Journal. On

Lattice light-sheet #microscopy with 4D image analysis reveals #Bcell receptors #immunoglobulin localise into clusters to aid recognition of their #antigen. Video & research by Deniz Saltukoglu et al BIOSS Freiburg in The EMBO Journal. On

Biomedical Picture of the Day BPoD

11,265 次观看 • 2 年前

⚡️ Jujutsu Kaisen in Concert is finally touring across Asia & Australia! ⚡️ Ticket on-sale dates across Asia & Australia will vary by city. Sign up now for first access and be the first to know when tickets go live → 🎟️ The tour premieres in just weeks in London — with tickets available across EU and San Antonio, U.S. this June. Enter the domain of music and mayhem at Jujutsu Kaisen in Concert, featuring live music paired with epic visuals from all 3 seasons and a special Season 3 encore. 🔥 More info available at

⚡️ Jujutsu Kaisen in Concert is finally touring across Asia & Australia! ⚡️ Ticket on-sale dates across Asia & Australia will vary by city. Sign up now for first access and be the first to know when tickets go live → 🎟️ The tour premieres in just weeks in London — with tickets available across EU and San Antonio, U.S. this June. Enter the domain of music and mayhem at Jujutsu Kaisen in Concert, featuring live music paired with epic visuals from all 3 seasons and a special Season 3 encore. 🔥 More info available at

JUJUTSU KAISEN EN

53,390 次观看 • 3 个月前

Chop the gradients ✂️! We found that truncating decoder gradients in latent video diffusion to a fixed window allows us to finetune on videos with pixel-wise perceptual losses without running out of memory. Pixel losses have been essential for image generation and reconstruction, but until now, they haven't scaled to long-duration, high-resolution video diffusion due to recursive activation accumulation in causal decoders, leading to OOM during training 💥📉. Project: Video diffusion models can do a lot more 🚀 when you can backprop the decoder! Post-process neural rendered scenes, super-resolve videos, harmonize lighting in controlled synthetic driving scenes, and inpaint videos — all in a single step ⚡ with a quick finetune from a standard diffusion model.

Chop the gradients ✂️! We found that truncating decoder gradients in latent video diffusion to a fixed window allows us to finetune on videos with pixel-wise perceptual losses without running out of memory. Pixel losses have been essential for image generation and reconstruction, but until now, they haven't scaled to long-duration, high-resolution video diffusion due to recursive activation accumulation in causal decoders, leading to OOM during training 💥📉. Project: Video diffusion models can do a lot more 🚀 when you can backprop the decoder! Post-process neural rendered scenes, super-resolve videos, harmonize lighting in controlled synthetic driving scenes, and inpaint videos — all in a single step ⚡ with a quick finetune from a standard diffusion model.

Felix Heide

28,399 次观看 • 4 个月前

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.

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

17,062 次观看 • 1 年前

TGE vestings are done, all presale has exited and price discovery is here. What's next for $SUIAI? 🤖 Monday will be product Demo to the world and the week will be packed with news and announcements. Just a small sneak peak of what's to come: - Product Demo - First AI Agent on Sui - Coverage on/by big Crypto shows and infuental people in the space - Staking goes live Things are about to go wild in #AI on $SUI.

TGE vestings are done, all presale has exited and price discovery is here. What's next for $SUIAI? 🤖 Monday will be product Demo to the world and the week will be packed with news and announcements. Just a small sneak peak of what's to come: - Product Demo - First AI Agent on Sui - Coverage on/by big Crypto shows and infuental people in the space - Staking goes live Things are about to go wild in #AI on $SUI.

SUI Agents

83,957 次观看 • 1 年前

Decentralized infrastructure is set to push the boundaries of gaming and many more industries for the foreseeable future! 🕹️ We're proud to play a big role in #DePIN revolution, by combining it with AI and gaming 💪 Thanks for the spotlight Coin Bureau ❤️

Decentralized infrastructure is set to push the boundaries of gaming and many more industries for the foreseeable future! 🕹️ We're proud to play a big role in #DePIN revolution, by combining it with AI and gaming 💪 Thanks for the spotlight Coin Bureau ❤️

NATIX Network

83,861 次观看 • 2 年前