Image created with Flux Pro v1.1 Ultra. Image prompt: Ornate showgirl glamour in orange-and-teal tones, dazzling headset props glittering with rhinestones featuring holographic dancers, stylized text “ARVR” projected in teal laser-light on a feathered backdrop; spotlit, dramatic contrast, vintage grain, cinematic, high-detail
Genie 2 vs Genie 3. Just 7 months between them. The bitter lesson continues to be bitter. https://x.com/bilawalsidhu/status/1952792880285896710
Genie 3 feels like playing a dream – a controlled hallucination of reality. Really makes you wonder if reality is just the same – except instead of just a few minutes, we can recall a whole lifetime. https://x.com/bilawalsidhu/status/1952895900390404231
Genie 3 generates interactive video in real-time. Just need to generate offset left/right eye views and you’ve got stereo VR worlds. No 3D models, no game engine – just generated dreams you can walk through. The holodeck is closer than you think.”” / X https://x.com/bilawalsidhu/status/1953094066993803454
genie 3 is wild. imagine looking over at your reflection in the tv screen and it’s just you standing there with a gopro strapped to your head… 🤯”” / X https://x.com/bilawalsidhu/status/1953158780835012881
Genie 3: A new frontier for world models – Google DeepMind https://deepmind.google/discover/blog/genie-3-a-new-frontier-for-world-models/
Genie-3 just achieved what AAA game engines do – but WITHOUT any 3D models. Interactive REAL-TIME video generation @ 24 fps Wild how this model figured out complex effects like exposure shifts, volumetric god rays, and phenomena we need to code explicitly in 3D engines TL;DR 🧵 https://x.com/bilawalsidhu/status/1952742891295764620
Google is flexing their AI muscles again. DeepMind has unveiled Genie 3, a real-time interactive, general-purpose world model that generates environments from text prompts, with visual memory extending as far back as one minute to keep scenes consistent. It could help advance https://x.com/TheHumanoidHub/status/1952801280059183210
hollup! so can you pre-load a chunk of a real video into genie’s world memory, so what you’ve seen IRL is actually what you see when you look around? can genie 3 basically do neural scene reconstruction in real time?! even if it’s not a “”factual”” rendition of the world like”” / X https://x.com/bilawalsidhu/status/1953187618700574851
Introducing Genie 3, the most advanced world simulator ever created, enabled by numerous research breakthroughs. 🤯 Featuring high fidelity visuals, 20-24 fps, prompting on the go, world memory, and more. https://x.com/OfficialLoganK/status/1952732206176112915
One word: relentless. just in the past two weeks, we’ve shipped: 🌐 Genie 3 – the most advanced world simulator ever 🤔 Gemini 2.5 Pro Deep Think available to Ultra subs 🎓 Gemini Pro free for uni students & $1B for US ed 🌍 AlphaEarth – a geospatial model of the entire planet”” / X https://x.com/demishassabis/status/1953887339094143156
RT @OriolVinyalsML: Incredible evolution of “”Neural Video Games””: from GQN (2018) to Genie3 (2025). The future is exciting! https://x.com/demishassabis/status/1952890039643353219
Sparks of in-context learning in Genie 3. You can prompt Genie 3 with a video (e.g. Veo 3) then control from there. Genie 3 will mimic the dynamics. I think we have only scratched the surface of what can be done with prompting and post-training of foundational world models.”” / X https://x.com/_rockt/status/1953117236975030653
We need to go deeper. Genie 3 is having its inception moment.”” / X https://x.com/shlomifruchter/status/1953155882902274126
We’re entering the era of infinite AI training environments. Google DeepMind just announced Genie 3, the first real-time interactive world model that creates worlds from text prompts. The video below shows a controllable environment generated by Genie 3 in real time. Insane. https://x.com/rowancheung/status/1952732216959623583
World modeling for robotics is incredibly hard because (1) control of humanoid robots & 5-finger hands is wayyy harder than ⬆️⬅️⬇️➡️ in games (Genie 3); and (2) object interaction is much more diverse than FSD, which needs to *avoid* coming into contact. Our GR00T Dreams work was”” / X https://x.com/DrJimFan/status/1952760780706984051
🌋 Volcano rock 🌊 Ocean wave ⚡ Storm cloud ASMR Made With #Veo3 Automated With My AI Autopilot! https://x.com/Mentor/status/1942016976827863103
🚨Veo3 Update is Here🚨 Wow, this will change how I make film with AI! With Google Veo3, you can now make yourself talk anything in any language, anywhere. What would you create?? https://x.com/herokominato/status/1942729320948256828
An alien vlogs his first Ahmedabad trip, discovering ‘Khalasi’. 👽 Google Veo 3 vividly realizes imagination, speaking Gujarati effortlessly with 100% auto-generated audio. #AI #Khalasi #GenerativeAI #Veo3 https://x.com/drashyakuruwa/status/1942647461522333777
Comparing Kling 2.1 with audio via Thinksound in @replicate (1st 10 sec vid) to Veo3 in Flow Studio (2nd 8 sec vid). Very impressed with Thinksound. Landscape design and real project photo (for i2v) by VizX Design Studio. https://x.com/Clearstory3D/status/1944505549543833656
I AM DYING 😭🤣 made this with Veo3. also side note this just goes to show that any AI brookejlacey video will never, ever be as good as the original. I will be the ultimate reality dealer, mark my words. https://x.com/brookejlacey/status/1944477615827611691
I made a stop-motion animation for Chanel No. 5 using Veo3. Prompt share： A claymation Paris street at midnight — the Eiffel Tower sparkles in the background, and a tiny clay bottle of Chanel No.5 tiptoes across cobblestones. Make sure everything used stop motion animation. https://x.com/crystalsssup/status/1942162692938354804
I made this in 4 hours! Veo 3 image to video character consistency. Midjourney for character creation and style. Runway ML for coverage. Veo 3 image to video with json prompts #AIart #veo3 #midjourneyv7 #runwayml #rockwilerai #ai #PromptEngineering https://x.com/therockwiler/status/1942883991117336812
It’s kinda crazy how easy it is to make ads with AI now! This is my very first try using VEO3. Not bad, right? 😉 https://x.com/agentsrihan/status/1942987346921533463
Just saw the vampire rap generated with Veo3 by my friend @WuxiaRocks. I didn’t know Veo3 can do such cool rap lipsync. Now, I want to create cool raps for brands I love. First up, @Netlify. @biilmann, what do you think?😉 https://x.com/zeng_wt/status/1943684922214171125
people are using veo3 to bring history to life in the form of vlogs 🤣 via HistoryVisualizedbyAI on YouTube https://x.com/tanayj/status/1934373978098778145
Prompt + image segmentation with @GoogleDeepMind VEO3 + @ultralytics 🚀 Here’s the simple workflow: ✅ Generated a video using VEO3 (prompt shared in the comments) ✅ Processed the clip directly with YOLOE for prompt-based image segmentation. Prompt in the comments👇 #AI https://x.com/muhammdrizwanmr/status/1941015898082468277
Quick play with Veo3 + Astra @topazlabs https://x.com/AllarHaltsonen/status/1941202785363788125
sailing like You’ve never seen before 🚤🌊 hyperreal waves, golden sunsets, & raw human emotion brought to life with cinematic precision. generated using #Veo3, now available on @moofeedcom. this isn’t just AI. It’s storytelling in motion. https://x.com/iUllr/status/1943956162874867858
The actual budget friendly launch trailer. Every shot generated with @GoogleDeepMind Veo3. With the philosophical idea that “”Death isn’t the end, forgetting is””. We proudly introducing EzCall AI. Time to speak the words you never got to say https://x.com/zhaoyuWu8/status/1942285389651403102
This entire short film was made using AI. No actors. No cameras. Just prompts, imagination, and tech. Watch it now #AIshortfilm #AIFilmmaking #AhmedabadCrash #RadheWorks #GenAI #OpenAI #FutureOfCinema #CinematicAI #Veo3 https://x.com/punit19nov/status/1942220841657508003
This is AI … but with real actors! The Hollywood film is about to change. We made with Veo3, Runway Reference and Flux Kontext using Me and My friend’s @Jamesgulles_ performances. Will the future be shaped by AI creators or by filmmakers? https://x.com/herokominato/status/1941844050451243187
Two word VEO3 prompt experimentations: > Cat Kaleidoscope 🔊 I find this calming kind of like ASMR but with hypnotic video https://x.com/rBKeeper/status/1943202740945006659
Veo 3 Fast and Veo 3 image-to-video are now available in the API! 📹 Veo 3 Fast is $0.40 per second of video (with audio) and comes with production ready rate limits and has comparable quality in certain cases! https://x.com/OfficialLoganK/status/1950959720606396655
Veo3 (Fast) is actually much better with consistent character than Veo3 (Quality). Here’s a 4-scene video of a Japanese tight rope walker on top of a sky scraper. Prompts in the comment. https://x.com/juminoz/status/1942399268192674285
Veo3 fast { “”shot””: { “”composition””: “”High-angle tracking shot from a helicopter, 200mm telephoto lens on a stabilized gimbal system, shot on RED Helium 8K S35″”, “”camera_motion””: “”aerial tracking following the emus’ path, with gradual zoom-in””, “”frame_rate””: https://x.com/IamEmily2050/status/1941126453715948005
FantasyPortrait: Enhancing Multi-Character Portrait Animation with Expression-Augmented Diffusion Transformers https://fantasy-amap.github.io/fantasy-portrait/
4d gaussian splat processed from one of the largest volumetric capture stages ever made. And fun fact, it was Intel that originally made it! I remember seeing the results in 2017, and my mind was blown. You could capture something once, and reframe it infinitely in post. Alas, https://x.com/bilawalsidhu/status/1952000783492186424
Relightable Full-body Gaussian Codec Avatars https://neuralbodies.github.io/RFGCA/
more tests reskinning google earth photogrammetry renders with runway’s aleph video-to-video ai model https://x.com/bilawalsidhu/status/1950717547206037511
i think the whole interactive video vs. explicit 3d debate is about get supercharged this week. meanwhile, here’s me reskinning 3d gaussian splat renders with runway aleph. https://x.com/bilawalsidhu/status/1952489882024386819
This is game engine 2.0. Some day, all the complexity of UE5 will be absorbed by a data-driven blob of attention weights. Those weights take as input game controller commands and directly animate a spacetime chunk of pixels. Agrim and I were close friends and coauthors back at”” / X https://x.com/DrJimFan/status/1952747404379504855
Elon says Tesla already has world model for Optimus. https://x.com/TheHumanoidHub/status/1952771309383077906
HumanOLAT: A Large-Scale Dataset for Full-Body Human Relighting and Novel-View Synthesis https://vcai.mpi-inf.mpg.de/projects/HumanOLAT/
I really like 3d point clouds https://x.com/bilawalsidhu/status/1950937892894687619
Physically Controllable Relighting of Photographs”” TLDR: Explicit control of light sources akin to CG pipelines; different types of light sources; self-supervised training methodology using differentiable rendering to train their neural renderer with real-world photograph. https://x.com/Almorgand/status/1952757175144558966
RAP: Real-time Audio-driven Portrait Animation with Video Diffusion Transformer https://markson14.github.io/RAP/
We’re thrilled to release & open-source Hunyuan3D World Model 1.0! This model enables you to generate immersive, explorable, and interactive 3D worlds from just a sentence or an image. It’s the industry’s first open-source 3D world generation model, compatible with CG pipelines https://x.com/TencentHunyuan/status/1949288986192834718
Cameras as Relative Positional Encoding”” TLDR: comparison for conditioning transformers on cameras: token-level raymap, attention-level relative pose encodings, a (new) relative encoding Projective Positional Encoding -> camera frustums, (int|ext)insics for relative pos encoding https://x.com/Almorgand/status/1951331762463822212
Our new AI model AlphaEarth Foundations is mapping the planet in astonishing detail. 🌏🔍 Scientists will now be able to track the impact of deforestation, monitoring crop health, and more – significantly faster, thanks to our new datasets. 🧵 https://x.com/GoogleDeepMind/status/1950563700286398965
The mysterious Horizon-alpha model. Same prompt as I have given a dozen other models: “”create a visually interesting shader that can run in twigl app make it like the ocean in a storm”” By far the best, and extremely fast to create. https://x.com/emollick/status/1950973603631055068
🎮 Your new AI Playground is here 🤖 Find all the hottest tools like Kling, Veo3, Seedance, and PixVerse in one spot (with more coming). No more digging. Just explore, experiment, and create. Learn more here: https://x.com/veedstudio/status/1944749874911539526
Everyone is prompting Veo 3 with code & JSON But you can also prompt solely through allusions to poetry: “”What Wordsworth beheld, single in the field, interrupted by what Robert Frost knew from desire”” That is “”Yon solitary Highland Lass”” & “”I hold with those who favor fire”” https://x.com/emollick/status/1952553426316877887
