OpenAI launches DALL·E 3: a new chapter in AI image generation
OpenAI recently released its latest image generation model, DALL·E 3. This model has attracted widespread attention in the field of AI because it brings a series of innovations and improvements.
Main features and new functions of DALL·E 3:
Greater detail and accuracy: DALL·E 3 is able to understand more details and subtleties than its predecessor, allowing users to easily transform their ideas into highly accurate images.
Research Preview: DALL·E 3 is currently in the research preview stage and will be available to ChatGPT Plus and enterprise customers in October, and will be launched in Labs later.
Precise text-to-image conversion: Modern text-to-image systems tend to ignore certain vocabulary or descriptions, forcing users to learn hint engineering. DALL·E 3 has made huge strides in producing images that are fully consistent with the text provided.
Comparison with DALL·E 2: Even with the same tips, DALL·E 3 has significant improvements over DALL·E 2.
Integration with ChatGPT: DALL·E 3 is built on ChatGPT, which means you can use ChatGPT as your brainstorming partner and tip perfector. Just ask ChatGPT what you want to see in a simple sentence to a detailed paragraph.
Safety: As with previous versions, we've taken steps to limit DALL·E 3's ability to generate violent, adult, or hateful content.
To prevent harmful generation: DALL·E 3 has taken steps to refuse requests for the names of public figures. In collaboration with red team experts, we have improved safety performance for public figure generation and harmful biases related to visual over/under representation.
Internal testing: We're also working on the best ways to help people identify whether an image was created by AI. We are experimenting with an Origin Classifier, a new internal tool that helps us identify whether an image was generated by DALL·E 3.
Creative Control: DALL·E 3 is designed to refuse requests to produce images in the style of existing artists. Creators can now also choose to exclude their images from training our future image generation models.
ChatGPT multi-modal is launched, adding new features:
Voice function: ChatGPT now supports voice input and output. Users can interact with it directly by voice without having to input through the keyboard.
Image recognition function: Users can upload pictures, and ChatGPT can recognize the content in the pictures and generate relevant text descriptions based on the pictures.
Reconnect the newbing function: This is a powerful search function that enables ChatGPT to search for relevant information on the Internet and provide users with richer content.
For example, if a user uploads a landscape picture, ChatGPT can identify elements such as mountains, water, and trees in the picture and generate a description: "This is a beautiful landscape painting with towering peaks, turquoise lakes, and lush woods."
Special tips:
For those who have not experienced ChatGPT and DALL·E 3, or want to experience them, there is good news! You can directly purchase an account or recharge on behalf of the world's leading artificial intelligence derivative service provider (official website: Neuronicx.com ), and log in directly to experience it. Because the above new features are only available to Plus members, if you want to experience these features, it is recommended to purchase a Plus member account.
In short, OpenAI's DALL·E 3 and ChatGPT have brought us unprecedented AI experience. Whether in text generation, image recognition or other aspects, they have demonstrated powerful capabilities. For AI enthusiasts and professionals, this is undoubtedly an era worth looking forward to.