The internet of cats beamed right into your eyeholes twitch.tv/vibediffuserai #stablediffusion #streamdiffusion

relax in the mountains twitch.tv/vibediffuserai #stablediffusion #streamdiffusion

#SDXL upgrade. text kinda works now :) plus prompt adherence way better. #streamdiffusion #stablediffusion.

Saturn clips.twitch.tv/HorribleFitLet…

SDXL/GPT4o/StreamDiffusion Music Vis x.com/i/broadcasts/1…

SDXL/GPT4o/StreamDiffusion AI Music Visualizer x.com/i/broadcasts/1…

AI Music Visualizer - SDXL/GPT4o/StreamDiffusion x.com/i/broadcasts/1…

AI Music Visualizer - SDXL/StreamDiffusion/GPT4o x.com/i/broadcasts/1…

AI Music Visualizer x.com/i/broadcasts/1…

TrippyTunesTV AI Music Visualizer - SDXL/GPT4o/StreamDiffusion x.com/i/broadcasts/1…

Experimenting around with streamv2v(github.com/Jeff-LiangF/st…) , definitely some improvements on temporal consistency, even with my crazy input videos. Managed to apply the sdxl support from github.com/hkn-g/StreamDi…, and it runs 15-20fps on a 4090.

StreamV2V AI music Visualizer - SD-Turbo x.com/i/broadcasts/1…

if your playing with diffusion models, definitely check out res-adapter.github.io and work it into your pipelines if your using non-512x512 resolutions. Easy to hack into stuff like #streamdiffusion if your generating 16x9 and knocks down the brain tumors on 1.5 derived models

Playing with OneDiff accelerated AnimateLCM pipelines. This was generated realtime.. need to try some last frame feedback with SparseCtrl...

It was only a matter of time. Real Time video generation with DiT could be on the horizon.

Wow, onediff accelerated diffusers AnimateLCM + rife_ncnn_vulkan_python makes a pretty decent stack. Any tips for autoregressive flows for AnimateDiff/AnimateLCM ideally with Diffusers, especially for alternate resolutions since SparseCtrl RGB really falls apart at non 512x512.

Remember StreamDiffusion? x.com/i/broadcasts/1…

Introducing MirageLSD: The First Live-Stream Diffusion (LSD) AI Model Input any video stream, from a camera or video chat to a computer screen or game, and transform it into any world you desire, in real-time (<40ms latency). Here’s how it works (w/ demo you can use!):

StreamDiffusionV2 - after a bit of tweaking of noise_scale and the adaptive noise scaler. It's pretty flexible. Can't wait to see what people build on this.

Playing with github.com/daydreamlive/s…