During a lifestyle on Tuesday, CEO of Openai Sam Altman announced the first major update for chatgpt image generation skills in more than a year.
Chatgpt can now use the company’s GPT-4o model to naturally create and modify images and photos. The GPT-4o has long supported the Chatbot platform with him, but so far, the model has been able to generate and modify the text-not images.
Altman said that the generation of the local GPT-4o image is directly in Chatgpt and Sora, OpenAi video generation product, for subscribers in the company’s 200-month-old plan. Openai says the feature is coming soon to plus and free chatgt users, as well as developers using the company API service.
GPT-4o with the output of the figure “thinks” a little longer than the image generating model that effectively replaces, distinct 3, to do what Openai describes as more accurate and detailed images. The GPT-4o can edit existing images, including images with people in them-turning them or “inpainting” details both in the foreground and background objects.
To strengthen the new image feature, Openai told The Wall Street Journal that he trained the GPT-4o in “Available Public Data” as well as property data from its partnerships with companies such as Shuttertstock.
Many generators of generators see training data as a competitive advantage, so they keep them and any information about it near the chest. But the details of training data are also a possible source of IP -related lawsuits, another breakdown for companies to reveal a lot.
“We are respecting the rights of artists in terms of how we do the result, and we have policies instead of preventing us from generating images that directly imitate the work of living artists,” said Brad Lightcap, Openai’s Operational Chief in a diary statement.
Openai offers a form of choice that allows creators to request that their works be removed from his training data. The company also says it respects requests to prevent its online replacement bots from training data collection, including images, from websites.
The improved feature of chatgpt image generation continues on the heels of the experimental exit of Google’s local image for 2.0 flash gemini, one of the company’s flag models. The powerful characteristic went viral on social media – but not necessarily for the best reasons. The 2.0 flash image component turned out to have little guards, allowing people to remove waters and create images that describe copyright protected characters.
This article was updated at 12:00 pt to include the OpenAi statement in the Wall Street Journal about GPT-4o training data.
