From Illusion Diffusion to Complex Workflows: 3 Years of GenAI Progress
osanseviero · x · 2026-08-31
A reflection on the rapid progress of generative AI over the last three years, moving from simple fun with Illusion Diffusion to much more advanced capabilities today. The post shares a specific creative workflow example demonstrating how to morph a medieval town image into a corgi using a combination of Runway Gen-2, Illusion Diffusion, and a pixel art LoRA by nerijs.
More from Multimodal
- High-Res Noise Masks Enable Pixel-Perfect Latent Motion Transfer — stonyleinchen · 2026-08-31
- H3 LoRA Loader Node Isolates Audio and Video Modalities — Dantemss · 2026-08-31
- Claude vs. ChatGPT vs. Figma Make: UI design showdown using the same prompt — Tegadesigns · 2026-08-31
- LumiTokens: 3D Relighting via Token-Space Lighting Transformation — qixing_huang · 2026-08-31
- Open-source Topaz alternative: SeedVR2+TensorRT for faster video restoration — Cheap_Credit_3957 · 2026-08-31
- Creator shares motion graphics ad made with Google Gemini Flash 1.1 — aziz4ai · 2026-08-31