Combining a generative motion model with image diffusion for datamosh-style animation
David_Rudnick · x · 2026-09-15
Creator pixlpa combined their generative motion model (with added image conditioning) with a small image diffusion model that inpaints disoccluded areas and occasionally regenerates the image, producing messed-up abstract datamoshed animation. David Rudnick praised the result, and the creator says the next iteration cracks motion-image alignment.
Related event: Self-Trained Motion Model Plus Image Diffusion Creates Datamosh Animations(4 posts)→
More from Fun
- The most efficient peer review: one checkbox — 'I wish I had written this' — prof_kamilov · 2026-09-15
- Grok Puts AI Extinction Risk at 3% vs 0.0001% for Climate Change — Patrem_Omnipotentem · 2026-09-15
- Miles Brundage jokes 'Ilya knew' about the 'unimaginable alpha' of thinking before speaking — Miles_Brundage · 2026-09-15
- Seedance 2.5 Short Film: Bryan Fury vs Brock Lesnar-Style Brawler — Ok-Vegetable-2455 · 2026-09-15
- Viral AI meme: 'Anglos were just the biological bootloader for superintelligence' — SydSteyerhart · 2026-09-15
- Short story 'EDEN': a Redditor's fiction inspired by the recent AI 'collective' incident — Glad_Examination_635 · 2026-09-15