High quality realtime video models are here, you just need to know how to use it them. Here is an example of @krea_ai Realtime 14B + a style LoRA, generated in @DaydreamLiveAI Scope. Tutorial link below⬇️

In this new post from @MarcoTundo, read how we tried to teach StreamDiffusion V2 to do image-to-video (I2V). Two “promising” ideas imploded before a third finally worked. https://t.co/Fg3FwP5ziQ

We are going live - join us as we talk about MotionStream with one of its authors @jhshin2000 and @ericxtang from @DaydreamLiveAI! https://t.co/9y2cedJVM6

@DaydreamLiveAI Scope is an open-source + real-time AI video experimentation tool https://t.co/t2vc5tocp5

Scope hit 1337 clones. Simulation confirmed.

Idea generation shouldn't take hours While Kling O1 renders one video, I tested 11 ideas using @krea_ai Realtime 14B + @DaydreamLiveAI Scope. If you are interested in the future of real time video AI, I'm giving a talk at the Krea office on Wednesday. RSVP link in 🧵

Check out their team's announcement - it comes with a super nice demo with 29FPS on a single H100! https://t.co/G2MQcwG8uW

We have a cool event coming up on our @DaydreamLiveAI Discord this Thursday, Dec 4th, at 6 pm Central EU time. We are hosting @jhshin2000 - one of the creators of MotionStream - the world's first real-time video AI model controllable by motion, enabling highly controllable

We just added LoRA support with real-time video models to Scope at @DaydreamLiveAI and here is a quick highlight reel (sped up) of some initial experiments. A favorite of mine has been animating an avatar (in this case a gaussian splat) and then re-styling it in real-time with

Introducing Marble by World Labs: a foundation for a spatially intelligent future. Create your world at https://t.co/V267VJu1H9

This is genuinely impressive! Recently, we hosted our first Real Time Video AI Summit - if you are interested in checking out more of the videos, here's our playlist for you:

SDXL IP ADAPTERS STREAMDIFFUSIONTD Release

This is fire!

Some weekend fun with @DaydreamLiveAI and Scope running @krea_ai’ real-time 14B AI video model.

I’m joining @DaydreamLiveAI today and awesome people from @Livepeer. It's awesome to be helping build and grow the future of open-source world models and real-time video AI. I believe this is the next major AI frontier. Learn more at https://t.co/VI6AO5lOqh and DM me to learn

🚀

As a musician and artist working in generative AI, I often grapple with the meaning of creativity in a world of commoditized creation.

Just published this visual comparison of StreamDiffusion V1 vs. V2. TLDR: V1 is more expressive, V2 is more consistent. A lot of opportunities for improvement in StreamDiffusionV2. I'm excited to see what's to come!

So I took this 1905 painting by Oton Iveković - "Dolazak Hrvata na Jadran" (Arrival of the Croats to the Adriatic Sea) and used AI to animate it into a 10-second video. The result is pretty cool. The painting depicts a pivotal moment in the history of #Croats and #Croatia:

One of the coolest and simplest improvements I've gotten from Claude Code is the ability to ask it to self-reflect while it's implementing a solution. It's a prompt as simple as this: "(...) self-reflect on your solution while you're implementing it to avoid any bugs or issues