Hard Fork AI OpenAI’s Image AI Improves Lighting and Texture Realism
8 snips
Dec 17, 2025 The podcast dives into the exciting advancements of OpenAI's new image model, Image 1.5. Discover how its launch is fueled by competition with tech giants. The discussion highlights stunning performance improvements, faster generation, and enhanced editing features for lighting and composition. Jaeden showcases successful complex prompts and shares insights on the model's limitations regarding likeness accuracy. Also covered are workarounds for uploading specific assets and the leap to higher quality outputs, including 4K support.
AI Snips
Chapters
Transcript
Episode notes
Image 1.5 Signals OpenAI Comeback
- OpenAI released a new image model called Image 1.5 that significantly improves realism and speed.
- Jaeden Schafer felt the update was overdue and shows OpenAI racing to regain market share from competitors.
Speed And Instruction-Following Are Key
- Image 1.5 is reportedly four times faster and better at following instructions and edits.
- Jaeden believes speed and precision are vital to stop users switching to faster models like Nano Banana.
Upgrade Images To Improve Video Output
- Expect Sora's video generation to improve after image-model upgrades because video models build on image models.
- Use higher-quality image models first if you want better video generation downstream.