What you need to know
- Bing Image Creator just received a major update in the form of DALL-E 3.
- DALL-E 3 is a model made by OpenAI that can create more accurate and higher quality images than previous models available.
- Anyone can try the improved Bing Image Creator to make anything from dogs jumping through leaves to kaiju Clippy destroying a city.
Bing Image Creator just took a massive step forward by adding support for DALL-E 3. Images generated by the tool should now be more accurate and higher quality than those made with previous models. Microsoft announced DALL-E 3 integration with Bing Image Creator a couple of weeks ago, but the functionality is now available for everyone to try.
If you don’t care about the ins and outs of how DALL-E 3 improves image generation and just want to play with the new feature, jump on over to Bing.com/create. Just beware that it’s incredibly fun to be able to make images in a few seconds using natural language. You may want to set aside some time during your lunch break instead of getting distracted at work.
I asked Bing to generate a kaiju Clippy and the results are frankly horrifying. pic.twitter.com/kF0JAG6UxzOctober 4, 2023
Over 1 billion photos have been made by Bing Image Creator, according to Microsoft. That count is likely set to spike with the rollout of DALL-E 3. Images made by the improved Bing promise the following:
- Relevance and prompt following: DALL-E 3 follows the user’s prompt with even more precision and reliability than any previous models. For the best results, we recommend providing greater level of detail in the prompt—the more information, the more refined the final image will be.
- Coherence: DALL-E 3 generates images that are even more photorealistic than other models for a varied set of prompts. The images are not only visually appealing, but also logically consistent with the prompt.
- Aesthetics: DALL-E 3 generates images that are not only realistic, but also creative and artistic. The images can be uniquely styled with flair that meets your creativity.
When entering a prompt into Bing Image Creator, you can specify art style, details, and other elements of the image. For example, you can ask for a photo to appear realistic, be made in the style of anime, or have the tool create a pixel art image.
While being able to create realistic images is entertaining and impressive, there are ethical concerns surrounding the technology. Microsoft highlighted that all AI-generated images made by Bing Image Creator feature an invisible digital watermark that meets the C2PA specification. That spec aims to prevent misleading information from being presented as genuine.
There’s also a content moderation system in place to prevent harmful or inappropriate images from being generated, though we’ll have to see how that holds up in practice. Bing Chat was recently tricked into solving a CAPTCHA, so people may be able to get around guidelines for Bing Image Creator.
Bing Image Creator slowed down over the weekend due to a spike in usage, but Microsoft appears to have fixed any issues.
