OpenAI has announced the introduction of a new feature that will allow you to generate an image directly in ChatGPT. This feature will be available based on the omnimodal GPT-4o model. It will be available to everyone in the near future.
This technology will work on a model that better understands complex queries and improves the quality of visualization. This is especially important when creating images with multiple objects or text.
The new system, unlike DALL-E, which uses diffusion technology, generates images gradually: from left to right and from top to bottom. Thanks to this, artificial intelligence more accurately interprets complex queries, especially when you need to depict multiple objects or text. The image generation process will take longer, but will be more efficient.
OpenAI is also strengthening its security. It will block attempts to remove watermarks, filter requests to create NSFW content, and block the creation of content that exploits children.
Recall that ChatGPT recently learned to create tasks and remind users about them.
Only registered users can leave comments