OpenAI announces AI 'DALL-E 3' that can fully integrate with ChatGPT and generate images from conversations, more faithful to text and more secure than the previous generation
AI company OpenAI announced the image generation AI ' DALL-E 3 ' on September 20, 2023. DALL-E 3 features a large-scale language model, GPT-based chatbot AI, and integration with ChatGPT, and is scheduled to be available to ChatGPT Plus and Enterprise customers in October 2023.
DALL·E 3
OpenAI's new AI image generator pushes the limits in detail and prompt fidelity | Ars Technica
https://arstechnica.com/information-technology/2023/09/openai-announces-dall-e-3-a-next-gen-ai-image-generator-based-on-chatgpt/
'DALL-E' is an image generation AI announced by OpenAI in 2021 that can output images with content in line with text input called 'prompts.' DALL-E's language processing uses parameters from GPT-3, OpenAI's large-scale language model.
AI ``DALL E'' that automatically generates illustrations and photos from words such as ``Pikachu riding a bike'' and ``avocado chair'' - GIGAZINE
In 2022, 'DALL-E 2', an improved version of DALL-E, was released. This DALL-E 2 can generate higher resolution images with lower delay than DALL-E, and has added an 'Inpainting' function that allows you to specify and edit a certain area of a photo.
Introducing the high-resolution, low-latency version 'DALL E 2' of 'DALL E' that creates images from input text information - GIGAZINE
The DALL-E 3 announced this time is also an AI that generates images from text, but it is characterized by being built on ChatGPT, a chatbot AI developed based on GPT. DALL-E is provided as an integrated feature of the paid plan ChatGPT Plus , and by speaking to ChatGPT and entering an image prompt, ChatGPT will generate an image for you.
OpenAI says, ``Current image generation AI tended to ignore some words and descriptions when generating images from text, requiring users to
OpenAI also states that DALL-E3 has limits on its ability to generate violent, adult, or hateful content. In addition, it is designed to reject requests for the style of living artists, and creators can now opt out of their images from training the image generation model through a dedicated form .
OpenAI publishes example images generated with DALL-E 3 along with input text.
'An illustration of an avocado sitting on a therapist's chair with a gaping hole in the center telling the therapist, ``I feel empty inside.'' The therapist is writing notes with a spoon.'
'Photo of a spherical chair inspired by a lychee. The outside is bumpy and white, the inside is luxurious, and the background is tropical wallpaper.'
'Paper craft art depicting a girl gently hugging a cat. The girl and the cat are sitting in a potted plant, the cat purrs contentedly, and the girl is smiling. Made of handmade paper flowers and leaves. It’s colorful.”
“A bright yellow banana-shaped sofa sits in the cozy living room, its curves supporting a plethora of colorful cushions. A patterned rug on the wooden floor adds eclectic charm. There is a potted plant in the corner that stretches towards the sunlight coming in through the window.
'Little potato kings with majestic crowns sit on their thrones, ruling vast potato kingdoms filled with potato subjects and potato castles.'
“A middle-aged Asian woman with silver streaks in her black hair seems to be broken and shattered, intricately embedded in a sea of broken porcelain. The porcelain is made of glossy and matte blues, greens, Glowing with a harmonious splatter paint pattern of orange and red, she captures her dance in a surreal juxtaposition of movement and stillness. Her skin tone is the same pale hue as porcelain, giving her figure an air of mystery. is attached.”
Below are the images generated by DALL-E 2 (right) and DALL-E 3 (left) with the text 'Oil painting depicting a basketball player's dunk drawn like a nebula explosion.'
DALL-E 3 is in research preview at the time of writing, and is scheduled to be made available to ChatGPT Plus and Enterprise customers in early October 2023, and will also be made available to Labs Open AI in the second half of fall 2023. . Also, like DALL-E 2, OpenAI stipulates that images generated with DALL-E 3 are for use by customers, and OpenAI's permission is not required for reprinting, selling, or commercializing them.
Related Posts:
in Software, Web Service, Web Application, Posted by log1i_yk