Initially a pipe dream, AI image generation has come a long way since it arrived a few years ago. Tools like Midjourney, DALL-E, and Jasper Art are all incredibly powerful and allow users to generate just about anything that comes to mind. But OpenAI is stepping things up with the introduction of DALL-E 3, which is getting a public release later this year. But what exactly will DALL-E 3 bring to the table? This is our DALL-E 2 vs DALL-E comparison.

DALL-E 2 vs DALL-E 3: Prompt interpretation

DALL-E 2 is great at interpreting text prompts, though it’s not without issue. The AI image generator doesn’t correctly interpret all text prompts and is terrible at generating particular objects.

One of the most significant improvements made in DALL-E 3 is that the new version better understands text prompts, specifically longer ones. DALL-E 3 has also improved in areas that had previously posed problems for image-generation tools, including human details like hands and reflections.

One brand new feature that DALL-E 3 adds is the integration with ChatGPT. The AI chatbot has been built directly on top of DALL-E 3. It can now be utilized as a “brainstorming partner” to assist the user with creating image ideas via conversational exchanges. Users new to AI image generation can utilize ChatGPT to iterate on their text prompts, with the AI assistant offering helpful ideas for improving image generation.

DALL-E 2 vs DALL-E 3: Image quality

When directly comparing the outputs of DALL-E 2 with DALL-E 3 for the same prompt, it’s clear that the new version produces a significantly better image. DALL-E 3 generates images that feature greater detail with sharper lighting, textures, and more detailed backgrounds. In nearly every way, the image quality is much better than DALL-E 2’s results.

Another facet of DALL-E 3’s image quality is its ability to generate text successfully within the image. This has long been a problem area for even the most powerful AI image generation software, but it seems it is finally getting fixed with this new version.

While DALL-E 2 was only ever available via a stand-alone tool on OpenAI’s website, DALL-E 3 is now available directly through Microsoft’s search engine, Bing. The feature is available through Bing Chat, allowing users to ask for prompts from the AI image generator via Bing. DALL-E 2 never featured any integrations such as this.

This feature is only available for ChatGPT Plus users, who must pay a monthly subscription.

DALL-E 2 vs DALL-E 3: Safety features

Another major change that DALL-E 3 has made over DALL-E 2 is focusing more on safety protocols. This includes preventing the generation of images that feature adult, violent, or hateful imagery. This means that if a user inputs a text prompt requesting the generation of an image deemed inappropriate or explicit, DALL-E 3’s safety protocols will flag the image generation and deny the user’s request.

The safety feature will also stop the generation of an image that may fall under copyright infringement. This means that DALL-E 3 will decline to generate any images that resemble living public figures or directly imitate the style of living artists.

These changes have been implemented to stay within copyright laws and to avoid the creation of potentially offensive or hateful imagery.

When is DALL-E 3 coming out?

DALL-E 3 was officially released in August 2023, though it’s only available for wide use through Microsoft’s Bing Chat. Outside of Bing, Version 3.0 is currently only available to researchers, and it is not available to be used by the general public in other forms.