Microsoft announced the new Image Creator in its AI-powered Bing search engine in March. This tool can turn words into pictures according to the user’s imagination. This gives users access to infinite image possibilities right from within Bing. For the specific image users need, they don’t have to scourge the internet for the same, with this tool they can simply create it. The tool can come in handy in multiple situations. It can be a vivid dream, a birthday invitation to fit a unique theme or a new concept they’re trying to land at work. This tool can create images that are not only personal and specific but also unique to express the user’s idea. The new Bing Chat also allow users to ask for these pictures in the flow of their conversation.
Behind the scenes of the Image Creator
The tech giant has shared the thinking and the tech behind the Image Creator tool. Moreover, the company also has added some tips for how to best use the feature in the product.
Image Creator uses OpenAI’s latest DALL·E model. This is a powerful language model that can generate coherent and diverse texts on almost any topic. Except, instead of taking the text as input and producing text as output, DALL·E takes text as input and produces an image as output. It uses a technique called self-attention to learn the relationships between text and images.
This helps the tool to produce realistic images that can combine concepts, attributes, and styles. For example, users can ask Bing Image Creator to draw a “cat wearing a hat and sunglasses” or a “surreal painting of a pineapple”. Such prompts will generate images that match their request. Users can also include a significant number of details in their query to get more specific results. For the best results, users need to include a subject, details and the scene. Instead of directly using the Image Creator vertical, users can create images iteratively within a Bing chat session.
Why Microsoft made the tool available in phases
Microsoft made the Image Creator tool available in a few markets in October 2022 but did a wider rollout of the same at a later date in March. The company has also explained the reason for delaying the rollout in some markets.
The team at Microsoft wanted to test and learn the implications of responsible AI. It’s important to note that in addition to generating AI images with the latest DALL·E models from OpenAI, the company prohibits the use of Image Creator to produce content that can inflict harm on individuals or society.
To achieve this, the company blocks certain prompts and images and has used the last few months to help inform this approach. If users think that Microsoft is “over blocking” images they can use the appeal process which will also help the company to improve its classifiers.
The company is also expecting an improved model from OpenAI. Microsoft claims that the latest model has increased the quality of the images with more detail and understands more visual concepts, which leads to more relevant results. Image Creator is now available to try for free in English everywhere that the new Bing is available.
Post a Comment