Infrastructure
Daydream
Open-source, local-first runtime for realtime generative-video pipelines — run a video-diffusion node graph and stream it over WebRTC.
livepeer.org · listing ≠ endorsement · re-checked 2026-09-03
- Operator
- Livepeer ecosystem · open source
- Format
- Node-graph pipeline → WebRTC
- Model
- Bundles StreamDiffusion V2 · Krea Realtime · LongLive
- Access
- Local-first · cloud via Livepeer GPUs
- Role
- Realtime-gen runtime/transport for live rooms
What is Daydream?
Daydream runs autoregressive realtime video models — it bundles StreamDiffusion V2, Krea Realtime, LongLive and more — as a node graph, and exposes a WebRTC API on localhost so generated frames stream to a browser or straight into OBS, TouchDesigner or Unity in real time. Cloud inference routes through Livepeer's decentralized GPU network. It is infrastructure: you bring the model and the show.
What Daydream does
- Runs autoregressive realtime video models as a node-graph pipeline
- Bundles StreamDiffusion V2, Krea Realtime, LongLive and more
- Exposes a WebRTC API so generated frames stream live to a browser or into OBS/TouchDesigner
- Cloud inference routes through Livepeer’s decentralized GPU network (real-time AI in beta)
- A tiny WebRTC client (@daydreamlive/browser) with WHIP/WHEP, zero dependencies
Who is Daydream for?
- Building a live generative-video stream that renders locally and broadcasts over WebRTC
- Scaling realtime pipelines to cloud GPUs through Livepeer without a centralized cloud contract
- Piping live AI frames into OBS, TouchDesigner or Unity for broadcast
Pros and cons of Daydream
Pros
- Open source and local-first — you own the pipeline
- Broad WebRTC/NDI/Syphon/Spout integrations make it production-friendly
- Decentralized cloud option (Livepeer) rather than a single vendor lock-in
Cons
- Infrastructure, not a product — you bring the model, the scene and the ops
- Cloud realtime inference is still beta
- Bundled-model quality varies and is model-, not platform-, dependent
What people are saying
Verbatim quotes from public coverage — collected 2026-09-03.
“Open-source, local-first platform for running real-time interactive generative AI video pipelines.”
Frequently asked questions
What is Daydream?
An open-source, local-first runtime for running real-time interactive generative AI video pipelines and streaming them over WebRTC.
Who is Daydream for?
Developers and artists building live generative video — it is infrastructure you compose with, not a hosted channel.
How does Daydream relate to Livepeer?
Daydream is part of the Livepeer ecosystem; cloud inference can run on Livepeer’s decentralized GPU network.
Alternatives in Infrastructure
3
fal.ai
Low-latency generative inference — the “livestream-from-a-model” layer behind much of this wave.fal.ai
LiveKit
Open-source WebRTC realtime A/V infrastructure for voice and video agents — the engine behind ChatGPT voice mode.livekit.io