Latent Blend: embedding interpolation unlocks genuinely novel visuals from image models
graycrawford · x · 2026-11-12
graycrawford highlights Latent Blend, an image-model technique dedicated to embedding interpolation. By blending images in latent space, it produces genuinely novel visuals that prompts alone can't reach. It's live in his web app: users can ask the agent about Latent Blend, or right-click images directly to run a blend sweep/grid. A dedicated thread is promised soon.
More from Multimodal
- Open-source ComfyUI loader runs MiniMax H3 video LoRAs with H3 Image — IntellectuallDeviant · 2026-11-12
- Fable 5.2 builds a playable FPS in 30 minutes; Scoble says distribution is the real bottleneck — Scobleizer · 2026-11-12
- Audio-first video workflow: Seed Audio 1.0 + SeeDance 2.5 vs native audio, compared — TomLikesRobots · 2026-11-12
- Creator pushes Google's video model to render a working 3D pop-up book city — kimmonismus · 2026-11-12
- 'Harry Potter as a 1973 film' AI video shows retro restyling technique — techhalla · 2026-11-12
- AI video meltdown: glowing-eyed toilet man meets propeller-hat Wojak — jaykrown · 2026-11-12