Openai has recely added a new feature called “Images in ChatGPT” to its artificial ielligence chats, which enables images to create images using the Gpt-4O model. This new tool enhances the quality of the color in the images and improving the text inside them dramatically.
According to Openai spokesman Tara Christinon, this feature has been available to all users today. He explained that people who use the usual version of ChatGPT, despite some limitations, could request image production without a specific ceiling. However, it is likely that these restrictions will be changed in the future by increasing users or changing needs.
Gabriel Go, director of the Openai Research Group, said the “Gpt-4o omnimodal” model has been used to develop this tool; A model that is capable of processing differe data. Go also poied to the significa progress of the “Binding” feature, adding, “In previous models, maiaining the coordination between differe parts of the image was challenging, especially when the number of image elemes increased.” For example, artificial ielligence sometimes produced a red star and a red triangle, a red star and a blue triangle.
However, the new model has eliminated this inconsistency and is now capable of accurately in the image between 15 and 20 elemes. However, older models could only manage 5 to 8 elemes without error. Gabriel Go further explained that this feature uses the “Autoregressive” method; That is, ChatGPT creates images from the top to the bottom and from left to right. This approach is differe from previous models such as DALL · E, which was processed by the “Diffusion Model” based on the whole image, and therefore the images are more harmonious.
According to Openai’s research team manager, this feature is the result of coinuous efforts and numerous tests of his team in rece mohs. He noted that although the tool still faces challenges in the display of small and small texts within the images, it has been able to maiain the quality of text and image simultaneously. This progress is an importa step in enhancing the experience of ChatGPT users and reflects Openai’s commitme to coinuous improveme of its products.
This new feature not only enhances the accuracy and quality of images, but has paved the way for the more varied applications of these chats in the field of technology. It is expected that by receiving users’ feedback, the tool will be even more efficie in the future and its curre restrictions will gradually decrease.
Source: The Verge
