Diffusion images lack controllability: researchers argue for painter-style, tool-call-based generation
josephdviviano · x · 2026-11-17
A technical exchange on image-generation paradigms. @LinusMixson argues diffusion's core weakness is controllability: incorporating feedback is extremely hard, and OpenAI's Byzantine gpt-image-2 pipeline — likely compositing multiple diffusion outputs via tool calls — has remained SOTA for months. He sees more promise in code-based ("Claude paints in code") autoregressive approaches enabling painter-style in-stream composition. josephdviviano agrees, proposing diffusion be treated like a tool call doing many small transformations rather than a one-shot generator.
More from Models
- NVIDIA launches Nemotron for Science: RL post-training adds 17% on drug-metabolism prediction — NVIDIAAI · 2026-11-17
- NVIDIA launches Nemotron for Science: fully open models, datasets and agent recipes — NVIDIAAI · 2026-11-17
- User Has Spent $2,513 on ChatGPT Since 2023 but Eyes a Switch to Claude — MoreLunch1547 · 2026-11-17
- Claude Fable 5.2 writes 43% shorter replies and cuts em dash usage by 99% — ArtificialAnlys · 2026-11-17
- InferenceBench update: Gemini 4 Carbon takes #3, breaking Anthropic's leaderboard sweep — maksym_andr · 2026-11-17
- ChatGPT hallucinates an entire nonexistent client email while sorting the user's inbox — jdjohnson · 2026-11-17