Artificial Intelligence (AI) continues to amaze us with its progress in image generation. YouTuber MattVidPro AI shared a video offering a glimpse into OpenAI’s latest model, DALL-E 3, which takes image generation to a whole new level. Matt expresses his excitement, claiming it surpasses other models like Midjourney and Stable Diffusion, including its predecessor, DALL-E 2.
Although DALL-E 2 was groundbreaking upon its release last year, subsequent versions like Midjourney and Stable Diffusion improved its image generation capabilities. Nevertheless, OpenAI didn’t stop there. The video, obtained from a user on Matt’s Discord server, teases the potential of DALL-E 3.
One remarkable feature of DALL-E 3 is its coherent and accurate text generation within images, an area where AI models have historically struggled. The video showcases examples of impeccably spelled and well-rendered text, including logos, product names, and even complete sentences.
However, it’s important to note that the version of DALL-E 3 shown in the video is still a work in progress. OpenAI has removed safety features, allowing the model to generate violent and potentially disturbing images.
Although the results are astonishingly realistic and detailed, concerns arise regarding potential misuse or the creation of inappropriate content.
The video assures viewers that OpenAI will fine-tune the model and implement safety measures before any public release. This is to ensure adherence to ethical guidelines and prevent the generation of harmful or offensive content.
As for the release date, it remains unknown. MattVidPro AI speculates that it could possibly happen before the year’s end.