Most AI video tools are still focused on one workflow: enter a prompt, generate a video, then start over whenever something needs to change.
But @ViduAI_official is already pushing beyond basic generation with Vidu S2—toward a real-time interactive video ecosystem where the content can change while the video is being generated.
There are five technical upgrades making this possible:
1.Backbone–Refiner architecture produces structure and motion with parallel refinement for real-time 720P output.
2.Timestamp-aligned positional encoding enables seamless integration of reference images into motion and lighting.
3. An upgraded VLM Agent monitors video progress and plans subsequent actions.
4. Reinforcement learning applied to the causal model maintains training improvements during deployment.
5. Real-time editing of style, outfit, character and background avoids full regeneration.
This could unlock entirely new experiences across live streaming, gaming, entertainment, and e-commerce.
Imagine viewers using comments to change a streamer’s outfit, character, or environment in real time; a person moving continuously between anime, gaming and cinematic worlds; or fashion models trying on different branded outfits live while the campaign style changes around them.
That feels less like simply generating a video and more like interacting with—and directing—a living video world as it unfolds.
You can try Vidu S2 here: https://t.co/2p1xMnn3oU
And if you use the exclusive invite link, you can get 300 credits to explore the platform and its features: https://t.co/nW9KD3LYbO…
Right now, the entire Vidu S series is available for free trial, so don’t miss the chance to try it out. You can also get 300 credits when exploring other features. Take this opportunity to experience Vidu S2 and see what you can create.
#ViduAI #ViduS #ViduS2
This is the kind of feature that makes observability actually useful.
Jitsu makes it easy to connect with Datadog, Grafana, or New Relic through a single OTLP endpoint—without extra agents or unnecessary complexity.
Having dead letters, function logs, and failed batches visible in one place is a huge win for teams.
And keeping raw events private while supporting destination-specific alerts is a thoughtful touch.
Great work, @sohmehta 👏
Launching tokenized assets used to mean building the full stack first.
@awscloud provides the secure cloud foundation institutions already trust. @OpenAssetsInc brings the tokenization layer on top: issuance, compliance, and connectivity.
Together, infrastructure that once took years can now deploy in weeks.
The biggest on-chain ARCade launching this week with 4 minigames on ARC @arc! 💰💵
Sign up to waitlist here: https://t.co/nXesBJitQ9
Like, retweet, tag your friends, and comment your wallet for a surprise👀🎰🎰
The exact launch time of Shares will be announced tomorrow.
Tokenised $HOOD is acquired in weekly buyouts, using revenue generated from trading fees.
It is then distributed to all $SHR holders on chain.
Buy Shares, become a shareholder.