OpenAI Upgrades Its Image Generation
Bi-weekly news commentary
Hello friend,
How are you? I hope you had a great week.
The gist:
OpenAI has just updated its GPT-40 model with more realistic images. This results from a full-year effort by an army of human trainers. GPT-40 replaces DALL-E3, which, if you recall, used to be a standalone image generator. This update is available to ChatGPT Free (limited prompts), Plus, Team, and Pro users. The model is trained on public data and licensing deals with companies like Shutterstock and News Corp. The stakes are high as this release comes with lower content guardrails which opened the floodgates to visual creations like the anime hot trend of images styled after Studio Ghibli, and bad actors creating misleading deepfakes. Artists can opt out of their work from being used in the LLM but copyright infringements are inevitable.
Find out the key use cases to leverage from this powerful visualization engine to fuel your visual storytelling strategy. You’ll also find an eye-opening side-by-side comparison of the same prompt I’ve been testing since 2022 - to witness the remarkable improvements the technology has gained.