Plategirl by OpenAI

DALL-E 3 might just have Killed Prompt Engineering for Image Creation

DALL-E 3 can accurately represent a scene with specific objects and the relationships between them.

This says it all. You can now describe a scene and objects and DALL-E 3 creates the corresponding image. If you have ever tried to get a specific scene from a tool like Midjourney or DALL-E 2, you know how difficult this can be.  It’s one thing to get something cool looking more or less by trial and error, but it’s extremely difficult to get specific things with prompt engineering for image generation. Scenes where different objects have to follow specific relationships described in the prompt are something that just didn’t work and it’s still quite the art to translate your mental image into a working prompt.

dalle-image-map by OpenAI
dalle-image-map by OpenAI

To that end, many people have already experimented with generating prompts with ChatGPT. Now, OpenAI has gone a step further and built DALL-E 3 on top of ChatGPT so that the latter can act as a creative partner. With that you can ask ChatGPT what you want to see in picture describing it with a simple sentence or a detailed paragraph. For instance, the beautiful hero image of this blog post is represented by this prompt:

A middle-aged woman of Asian descent, her dark hair streaked with silver, appears fractured and splintered, intricately embedded within a sea of broken porcelain. The porcelain glistens with splatter paint patterns in a harmonious blend of glossy and matte blues, greens, oranges, and reds, capturing her dance in a surreal juxtaposition of movement and stillness. Her skin tone, a light hue like the porcelain, adds an almost mystical quality to her form.

Therefore, when given a concept, ChatGPT can seamlessly craft customized, intricate prompts for DALL-E 3, transforming your idea into vivid imagery. If an image catches your eye but needs a touch-up, simply instruct ChatGPT with a brief directive.

In terms of safety, DALL-E 3 has measures to limit its ability to generate violent, adult, or hateful content. It also declines requests that ask for a public figure by name. OpenAI is researching ways to help people identify when an image was created with AI, but hasn’t shared anything yet.

OpenAI has made a significant move to safeguard the interests of living artists—a crucial measure to foster understanding and potentially sidestep future legal disputes. DALL·E 3 will refrain from replicating their unique styles, addressing a major concern from the artistic community. Moreover, artists will have the option to exclude their images from the training datasets of upcoming image generation models.

DALL-E 3 is set to be accessible to ChatGPT Plus and Enterprise users come early October. Much like its predecessor, DALL-E 2, the visuals you produce with DALL-E 3 are entirely yours. You have the freedom to republish, commercialize, or merchandise them without seeking OpenAIs approval.

Scroll to top